research article machine vision based automatic...
TRANSCRIPT
Research ArticleMachine Vision Based Automatic Detection Method ofIndicating Values of a Pointer Gauge
Jiannan Chi12 Lei Liu1 Jiwei Liu3 Zhaoxuan Jiang1 and Guosheng Zhang2
1School of Automation and Electrical Engineering University of Science and Technology Beijing Beijing 100083 China2Key Laboratory of Operation Safety Technology on Transport Vehicles Ministry of Transport Beijing 100088 China3School of Automation Shenyang Aerospace University Shenyang 110136 China
Correspondence should be addressed to Jiannan Chi ustbjncustbeducn
Received 4 November 2014 Revised 27 December 2014 Accepted 4 February 2015
Academic Editor Chih-Cheng Hung
Copyright copy 2015 Jiannan Chi et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
This study proposes an automatic reading approach for a pointer gauge based on computer vision Moreover the study aims tohighlight the defects of the current automatic-recognitionmethod of the pointer gauge and introduces a method that uses a coarse-to-fine scheme and has superior performance in the accuracy and stability of its reading identification First it uses the regiongrowing method to locate the dial region and its center Second it uses an improved central projection method to determine thecircular scale region under the polar coordinate system and detect the scale marks Then the border detection is implementedin the dial image and the Hough transform method is used to obtain the pointer direction by means of pointer contour fittingFinally the reading of the gauge is obtained by comparing the location of the pointer with the scale marksThe experimental resultsdemonstrate the effectiveness of the proposed approachThis approach is applicable for reading gauges whose scalemarks are eitherevenly or unevenly distributed
1 Introduction
With the rapid development of information technology thedigital meter has been widely applied But pointer gaugeis still very popular in various fields due to its simplestructure high reliability low price and easy operationHowever owing to the nondigital signal output of the pointergauge a computer cannot conduct the processing and remotetransmission of the data collected by the pointer gauge Thislimits its application Therefore the method of providing thepointer gauge with digital features such as automatic readingand transforming the collected value into the digital signalshould be urgently solved for wider application In particularin the process of calibration or metrological verification [1]the checkers need to read the output values of the standardand the indicating value of the calibrated gauge respectivelyand then make a comparison Random errors may occur inthis case due to the limited acuity of human observationFurthermore when an operator is far away from the gaugethe operator has to repeat the reading and manually recordthe indicating value This not only increases the operatorrsquos
burden but also lowers the efficiency of the gauge calibrationTo overcome the above limitations and to make the pointergauge more user-friendly new technologies are expected todigitalize its reading At present machine vision is widelyutilized to detect the gauge dial and pointer and then conducta reading
This study aims at the application of pressure gaugeverification and proposes a computer vision based automaticreading approach for a pointer gauge that adopts a coarse-to-fine scheme In this approach the dial region and itscenter of the pointer gauge are located by first using theregion growing method Then the circular scale region isdetermined by the adaptive threshold method under thepolar coordinate system In the circular region the scalemarks distribution diagram is produced using improvedcentral projection According to the scale marks distributiondiagram the main scale marks are located in the circularscale region of the dial In the following steps the Houghtransformation is used to detect the pointer in the dialregion and obtain its direction Finally the distance method
Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2015 Article ID 283629 19 pageshttpdxdoiorg1011552015283629
2 Mathematical Problems in Engineering
PC Image
Drive card Stepping
Other signal
CCD or
Stepping
High
standard
Tested
Worktable
Benchmarksignal
generator
collectors
capture card
motor driver
CMOS camera
motor
precision
gaugegauge
Figure 1 Hardware structure of the automatic recognition of the indicating value of the pointer gauge
Image Image
acquiring
Preprocessing gauge image
Pointerand scale
Automatic
reading
Gaugereadingoutput
Reading identification
preprocessing marks detection identification
Detection
of pointer
and scale
Figure 2 The image processing procedure of the automatic recognition of the indicating value of the pointer gauge
is adopted to obtain the indicating value of the gauge bycomparing the pointer direction with the position of the scalemarks
Since the indicating value is determined by comparing thepointer direction and the locations of the scale marks thisapproach is also suitable for gauges wherein the scale marksare unevenly distributed Therefore the proposed approachis in fact a general approach for gauge calibration
2 Related Work
21 Configuration of Hardware System Used in Gauge Ver-ification The hardware system used in gauge verificationgenerally comprises the following configuration [2ndash5] whichis shown in Figure 1 During the working process the highprecise gauge and the calibrated gauge are placed on theworktable A computer sends impulse signals to the motordriver Therefore the stepping motor is controlled to maketheworktablemove to allow for the adjustment of the locationof the gauge so that the gauges can be appropriately placedin the camerarsquos view while the standard signal source sendsreference signals to the calibrated pointer gauge and the highprecise gauge The indicating value of the high precise gaugeis treated as standard data Next computer vision is applied toautomatically recognize the values indicated by the calibratedand standard gauges Finally a comparison is made betweenthe two values and the verification result is reached
For some other pointer gauges the standard data pro-duced by the reference signal generator is directly inputted
into the computer by the other signal-acquisition equipmentThe verification result is achieved by comparing the standarddata with the indicating value of the calibrated gauge whichis extracted using computer vision
22 The Process of Automatic Reading of the Pointer GaugeBased on Computer Vision Figure 2 shows the procedure ofthe image processing used to recognize the pointer gaugevalue First to improve the image quality the gauge imagehas to be preprocessed by means of noise suppression andimage enhancement Next image processing and analyzingmethods are used to detect the pointer and scale marksAccording to the pointer direction and the location of scalemarks the value indicated by the pointer gauge can bedetected Finally the indicating value results are outputtedTherefore the automatic recognition methods of the gaugeindicating value mainly include three parts
(1) image preprocessing aimed to improve image qualityand suppress noise
(2) gauge pointer and scale marks detection based onimage processing and analyzing
(3) automatic recognition of the pointer gauge indicatingvalue
23 Related Work of the Pointer Gauge Indicating ValueDetection In the existing literature the pointer gaugeimage should be preprocessed to remove noise or enhance
Mathematical Problems in Engineering 3
the image especially for the images extracted from industrialfields Image preprocessing improves the image resolutionThus it is easier to detect the features relating to pointer andscale marks by means of image analyzing
Image preprocessing of the pointer gauge mainly com-prises two parts First image preprocessing is used to filterout interference and noise that result from the change ofenvironment For example in [6] the homomorphic filteringmethod was adopted to process the gauge image that hasa large area of glare and reflection Second image prepro-cessing is used to correct an image distortion caused bycamera imaging or different shooting angles For instance[7] discussed how to reduce the reading error when thegauge dial was uneven with the camera lens surface Thepapers [8 9] proposed an intelligent method of gauge valuereading to avoid the influence of imaging distortion thismethod was able to correct the angle between the pointer andthe zero-scale marks and the angle between the zero-scalemarks and the other longer scale marks It also adopted theNewton interpolation method to approximate the functionalrelationships between the pointer angle and its indicatingvalue
After image preprocessing image segmentation shouldbe implemented to obtain a binarization image wherein thethreshold-segmentation method is always applied [10ndash13] Inthe following step image thinning needs to be conducted inthe binarization image to achieve a one-pixel-wide pointerimageThen straight-line fitting is used to detect the pointerSo the key point of the automatic recognition of the gaugeindicating value is the detection of the pointer and scalemarks In the existing literature several methods for pointerdetection such as subtractionmethod the templatematchingmethod the Hough transform method and the central pro-jectionmethodwere proposed In the subtractionmethod [414 15] two gauge images with roughly the same backgroundswere taken by camera while the pointer pointed toward adifferent scaleWhen one image was subtracted from anotherone there were only two pointers left in the difference imageNext the difference image was segmented to detect thepointer Finally the gaugersquos indicating value could be obtainedby calculating the position of the pointer and scale marksThe subtraction method was easily implemented by meansof image subtraction However the different image not onlyreserves the pointer target but also leaves out a lot of noise dueto the influence of illuminationTherefore in many cases thesubtraction method does not produce reliable results
In the template feature method [16] the features thatrepresent the pointer and its direction such as shape areaand gray distribution formed amultiple parameters templateused to match the pointer in the gauge image for pointerdetection Since the template comprises multiple cues of thepointer and its direction the template feature method couldavoid the influence of an uneven illumination and otherenvironmental changing factors However the process of atemplate match is too complicated to accurately detect thepointer
The Hough transform method has often been used inpointer detection In [17 18] after edge detection and imagethinning in the gauge image the Hough transform method
or its improved version was applied to detect the pointer byfitting the pointer edge The Hough transform method couldachieve good results for the pointer gauge that has manywords and symbols on the dial In [19 20] a novel pointerdetection method called central projection was describedThemain idea of the central projection method was to searchfor the pointer location bymeans of calculating the projectionfrom the feature points to the known central point Everyprojection value corresponded to the angle of the straightline that connected the projection and central points Theremust be an angle that corresponds to the largest numberof projection points and represents the pointer Hence thecentral projection method was always used to determine thelocation of the gauge pointer
The least squares method [21] was also a straight-linefitting method just like the Hough transformmethod whichwas used to detect the pointer Compared with the Houghtransform method the least squares method required fewercalculations
After the pointer was detected angle and distance meth-ods [22]were always used to determine the indicating value ofthe point gauge The angle method recognized the indicatingvalue by calculating the deflection angle between the detectedpointer and the zero-scale mark according to the formula119872 times 120573120572 Here 119872 was the range 120572 was the premeasuredangle between the maximum- and zero-scale mark and 120573
was the angle between the current pointer and zero-scalemarkThe distancemethod [23] obtained the indicating valueby calculating the distance between the pointer and nearestscale mark In this method each scale reticule was locatedand labeled with the corresponding scale number in a two-dimensional coordinate system If the distance between thepointer and each reticule is as follows
119889119894=
1003816100381610038161003816119860119909119894 + 119861119910119894 + 1198621003816100381610038161003816
radic1198602 + 1198612 (1)
then the indicating value of the pointer is
119878119899= 119878119871+
1198891
1198891+ 1198890
times (119878119867minus 119878119871) (2)
Here (119909119894 119910119894) is the coordinate of the points in the scale marks
119889119894is the distance between the point and the line 119860119909 + 119861119910 +
119862 = 0 119878119899represents the indicating value of the pointer 119878
119867
and 119878119871are the values of the scale marks which are located
on both sides of the pointer and 1198891and 119889
0are the distance
between the pointer and the nearest scale marks on bothsides respectively The angle method is simpler than thedistance method but it is unsuitable for the pointer gauge inwhich the scale is uneven
3 Overview of Our Work
As mentioned above the current methods for gauge readingrecognition are limited in terms of their practical applicationmainly due to two deficiencies (1)Theexistingmethods wereall implemented in the global gauge images Therefore thepointer and scale marks segmentation cannot be accurately
4 Mathematical Problems in Engineering
conducted when there is some interference or where thereis some complex writing on the gauge dial (2) In manystudies the scale marks are not precisely detected Hence theindicating value of the pointer gauge with an uneven scalecannot be recognized
This study aims to overcome the shortcomings of theexisting methods As its contextual application it also aimsto improve the calibration of the pressure gauge and developan automatic recognition method of pointer gauge readingsbased on machine vision This method implements a schemethat takes detection from coarse to fine in terms of thepointer and scale marks detection First the pointer gaugersquosdial region is located in a global image Then the dialrsquosannular scale region is located In the following steps thescale marks are detected in the annular scale region and thepointer is detected in the dial region Finally based on theresults of the detection of the pointer and scale marks theimproved distance method is used to recognize the gaugersquosvalue Since the improved distance method is implementedfrom the coarse region to the fine target it has a strongability for suppressing interference Since the indicating valueis obtained from the pointer and scale marks detection themethod is also suitable for reading the gauge whose scalemarks are unevenly distributed
31 Machine Vision Based Pressure Gauge Verification SystemMetrological departments usually calibrate pressure gaugeswith piston manometers In this case the standard pressureoutput of piston manometers given by manually addedweights is passed on to the tested pressure gauge Then thequalification of the pressure gauge is determined throughreading analysis and comparison of the indicating value Ingeneral verification personnel manually handle this processthrough observation by recording and analyzing the indicat-ing value of the gauge panel Many problems still exist withthis verification method They include (1) low accuracy andrepeatability in human visual observations and (2) poor workefficiency due to the large workload of verification personnelHence development of an automatic verification methodfor the pressure pointer gauge will boost the advancementof the technology used in their production and verificationcapabilities The system composition for pointer pressuregauge verification is shown in Figure 3 We can see thatimages of pointer gauges are obtained through a camera andprocessed and analyzed by computer wherein the indicatingvalues of the standard pressure gauge and the test gaugeare detected and recognized The final results of the gaugereading will be displayed on amonitor to operators and reacha verification conclusion based on the analyses of the abovereadingsThe detection software also has suchmultifunctionsas the storage of historical data graphical display and imageswitching to realize the automatic verification of pointergauges
32 Introduction of OurWork In this study a new automaticrecognition method of the pointer gauge that indicates valueis proposed Figure 4 shows a block diagram of this method
In general the computer vision based automatic recog-nition method of indicating value comprises two parts One
Piston gaugeCalibrated gauge
Standard gauge
Display PC
Camera
Image capture card
Figure 3 Schematic of a machine vision based pressure gaugeverification system
part is the detection of the indicating value of the pointergauge This mainly includes the scale detection and thepointer detection which refer to the location of the dialregion and the dial center the determination of the annularscale region in the dial the scale detection and the pointerdetection The other is the automatic recognition of thereading Namely it makes the final judgment of the gaugereading according to the detected direction of the pointer andthe distribution of the scale marks As seen in Figure 4 theautomatic recognition of the indicating value proposed inthis paper falls into five steps
(1) The dial region and its center location first in theentire image the dial region can be determined bysemi-interaction (using the mouse to drop downthe box in the image) The pixel point with thebiggest gray value is searched out in the dial regionand taken as the seed point of the region growingmethod used to produce the circular region in thedial In the process of the region growing methodthe similarity threshold value between the pixel pointsis automatically determined Finally according to theresults of the region growing method the radius andcenter of the circular dial are obtained
(2) Determination of the scale region in the dial theimage coordinate is first converted into the polarcoordinate with the dial center as its origin whichis used to express the coordinate of each pixel in theimage Then by comparing the pixel sum of the lowgray value with that of the high gray value in theannular region the annular scale region in the dial isdetermined And finally it is copied into a new imageas the annular scale region
(3) The scale marks are detected using the central pro-jection method In the newly generated annular scaleregion image first the scale mark segmentation isconducted by theOSTUmethod to form the binariza-tion scale imageThen the angle of each line that linksthe black pixel with the circle center is calculatedand the number of each angle is obtained The angleswith large numbers (connecting line) in a narrowangle range correspond to the thick scalemarks (mainscales) on the dial Thus the angles of the scale marksin the scope of 0∘ndash360∘ are obtained
Mathematical Problems in Engineering 5
Scale detection by central projection method
Segment the image in annular scaleregion by OSTU to obtain the
binarization scale image
Traverse the black pixel points in theimage link each black point to the
circle center and then calculate the angleof the connecting line to get that of
each coarse scale mark
Detection of gauge pointer by Houghtransform
Utilize the Canny operator to performborder detection on the dial image
Use the adaptive threshold value toperform the Hough transform on dial
border image and fit the gauge pointercontour
Calculate the pointer direction within
The indicating value reading
Calculate the indicating valueaccording to the pointer direction and
the dial scale distribution
Dial region and center location
Search the pixel point with maximumgray in the dial region as the seed
Conduct region growing from the seedpoint and adaptively determine thesimilarity principle between the pixels
Obtain the center and radius ofcircular dial region by region growing
Determination of annular scale region in the dial
Convert the image into the polarcoordinates with the dial center as its
origin center
Determine the annular scale region bycomparison between the pixel sum oflow gray value and that of high gray
value
Copy the circular scale regionestablished into a new image
the scope of 0ndash Obtain the scale in the scope of 0ndash360degrees
360 degrees
Figure 4 The flow chart of the pointer gauge indicating value recognition
(4) Pointer detection first the Canny operator is usedto implement the border detection Then the Houghtransform method is used to fit the detected edgeand obtain the pointer contour Finally the pointerdirection is obtained according to the center positionand the pointer contour
(5) The indicating value is calculated according to thepointer direction and the dial scale distribution
4 Indicating Value Recognition Approach ofPointer Gauge
41 Pointer Gauge Indicating Value Detection As mentionedabove the detection of the indicating value is achievedthrough the scale and pointer detection As for the scaledetection first the gauge center and dial region should belocated followed by the extraction of the annular scale region
6 Mathematical Problems in Engineering
(a) Typical pointer gauge (b) The results of the dial region growing method
Figure 5 The diagram of the dial region growing results
and the scale mark detection Finally the detection of thepointer direction is performed
411 Dial Region and Its Center Location As for the detectionof the indicating value first the gauge pointer and its directionshould be detected to determine its rotation center In generalthe gaugersquos dial center is the pointerrsquos rotation center Hencefirst the dial center should be detected In this study thecircular dial region is detected for determining the dial centerAs is shown in Figure 5(a) typically there are the labelswords and scale marks in the circular dial Therefore it isdifficult to confirm the circular region of the dial by suchsegment methods as the border and threshold values Inview of the above this paper introduces the adaptive-region-growing method to locate the gaugersquos circular region andfurther to obtain its dial center The specific steps are asfollows
(1) The placement of the seed [24] since the dial isusually white the region with a high gray level orthe pixel point with the highest gray level in thegauge is certainly located at the part where the dialis Therefore the latter is selected as the seed point ofthe region growing
(2) Subject to the similarity of the pixel gray level theexpansion of the dial region is conducted from theseed point However it is hard to fix an appropriatesimilarity threshold value of the gray level betweenpixels As is shown in Figure 5(b) the circular dialregion cannot be completely produced with a smallthreshold value while an oversized dial region will begenerated with a large threshold value
This study proposes an adaptive method to determinethe similarity threshold value of the gray level the concreteprocess is as follows
(A) A small similarity threshold value 119879 of the graylevel is empirically chosen and then the dial regiongrowing method is conducted from the seed point
(B) The ratio of the circumscribed rectanglersquos length andwidth of the newly generated dial region is calculatedto detect whether it is in the 095ndash105 range
(C) If the ratio of the length and width fails to meetthe conditions in (B) the similarity threshold valuewould be increased with a certain step length (119878)and the first step (A) would continue to conductthe region growing from the seed point Howeverif it meets the conditions in (B) the growing regionwill be the circular dial region the center of thecircumscribed rectangle will be the circle center andthe smaller value between the length and width willbe the diameter of the circular region (Figure 6)
412 Polar Coordinate Based Scale Region Location
(1) Image in Polar Coordinates In general in light of thecircular permutation the pointer gauge scale marks aredistributed around the center Therefore to correctly andexpediently show where the scale marks in the image arethe polar coordinate system is utilized to position the pixelsin the image In this study a conversion from the imagecoordinates into the polar coordinates is performed Thismeans that the pixel coordinates expressed by the rectangularimage coordinate system are converted into those expressedby polar coordinates The polar coordinates defined in thisstudy are shown in Figure 7 the circle center being its origin
(2) Detection of the Scale Region in the Dial The procedurefor scale region detection based on the adaptive thresholdmethod is as follows
(A) The circular dial radius detected in Section 411 istaken as the initial radius 119877 its decreasing step lengthbeing 119887 and the initial 119899 = 1 In the gauge dial thegray value difference between the scale marks anddial background is generally rather large Thereforethe adaptive threshold value method can be adoptedto segment the scale In the polar coordinate theannular binarization image is gained by employing
Mathematical Problems in Engineering 7
(a) (b)
(c) (d)
Figure 6 The region growing results under different pixel similarity threshold values
90∘
60∘
210∘
150∘
120∘
180∘
240∘
270∘300∘
330∘
30∘
0∘
Figure 7 Schematic of the polar coordinate
the OSTU method [25] to perform the binarizationsegmentation in the annular region
(B) The ratio between the number of black andwhite pixelpoints in the annular binarization region between theradius 119877 and 119877minus119899lowast119887 is calculated and denoted by 119903
1
After reducing the inner radius according to the steplength 119887 the ratio between the number of black and
white pixel points in the region between 119877minus(119899+1)lowast119887and 119877 is computed and denoted by 119903
2
(C) If 1199031is obviously different from 119903
2 the inner radius
and the outer radius of the scale region would be 119877 minus(119899 + 1) lowast 119887 and 119877 respectively
(D) If 1199031is approximated to 119903
2 the annulus between 119877 minus
(119899 + 1) lowast 119887 and 119877 would still be in the circular scaleregion At thismoment let 119899 = 119899+1 the ratio betweenthe number of black and white pixel points in theannulus between 119877 minus (119899 + 1) lowast 119887 and 119877 is computedand is still denoted by 119903
2 Then go to (B) (Figure 8)
After obtaining the annular scale region in the abovepolar coordinate the scale region is selected on the basis ofangle and radius and copied into a new image called the scalemark region image119872 (Figure 9)
413 Scale Mark Detection Based on the Improved CentralProjection Method In the newly generated image119872 whichis the annular scale region the OSTU method is adopted tosegment the scale marks to form the binarized scale imageIn this image the pixel value of the scale mark is 0 (black)and those in other places are 255 (white) Each black pixelpoint and the circle center are linked and the slope 119870 ofthe connecting line is calculated according to the formula119896 = tan120572 wherein 120572 is the angle between the 119909-axis and
8 Mathematical Problems in Engineering
(a) (b)
(c)
Figure 8 Schematic of scale mark region detection
(a) (b)
Figure 9 The scale mark region diagram and its binarized image
Mathematical Problems in Engineering 9
x
y
0
(a) Schematic of the image coordinate system
x
y
1
1
2
2
3
3
4
4
minus1
minus1
minus2
minus2
0minus3
minus3
minus4
minus4
(b) The rectangular coordinate system
Figure 10 Schematic of the image coordinate system and the rectangular coordinate system
the connecting line between the black pixel point and thecircular center Due to the origin of the image coordinatebeing in the top left corner and the119910-axis running downwardit is inconvenient to denote the central angle within thescope of 0∘ndash360∘ Therefore in this study the rectangularcoordinate system with the image center as its origin isintroduced to express the angle between the black pixelpoint on the scale mark and the circle center instead of theimage coordinate system As is shown in Figure 10 the anglesbetween the black pixel point and the circle center are in the0∘ndash360∘ range
After calculating the number for angle 120572 the angle witha large number (connecting lines) in a small angle range isthat of the thick scale mark (main scale) on the dialThe stepsto detect the scale mark by central projection are detailed asfollows
(1) The black pixel point and the image central pointthat is the circle center (119909
0 1199100) in the new image are linked
to determine the slope 119896 which is converted to an angleaccording to the following formula
120572 = arctan (119896) lowast 180
120587 (3)
(2) As shown in Figure 10 the image coordinate systemis transformed to a rectangular one The angle 120572 followingformula (4) is converted into an angle in the rectangularcoordinate
120579 = minus120572 119894 lt 119903 119895 gt 119903 the first quadrant
120579 = 180 minus 120572 119894 lt 119903 119895 lt 119903 the second quadrant
120579 = 180 minus 120572 119894 gt 119903 119895 lt 119903 the third quadrant
120579 = 360 minus 120572 119894 gt 119903 119895 gt 119903 the forth quadrant
(4)
Occ
urre
nce t
imes
Angle (deg)1205791 1205792 1205793 12057940
Figure 11 The angle histogram
(3) An angle histogram is generated by recording theoccurrence frequency of each angle Figure 11 shows the anglehistogram
(4) The angle with the most occurrence times in thehistogram is selected and denoted by max all the angleswhose occurrence times are more thanmax 2 are foundTheoccurrence times of the angles are denoted by 119891(120579) Here 120579represents the angles that occur more than max 2
(5) For every angle 120579 we calculate the weighted averagevalue anglemean(119898) according the following to formula
anglemean (119898)
=119891 (120579 + 1) lowast (120579 + 1) + 119891 (120579) lowast (120579) + 119891 (120579 minus 1) lowast (120579 minus 1)
119891 (120579 + 1) + 119891 (120579) + 119891 (120579 minus 1)
(5)
Here119898 is the sign of the angle
10 Mathematical Problems in Engineering
12057221205721
1205720
1205723
Thick scale marksy
x
Figure 12 The location of the thick scale marks
(6) Calculate specific value 119875 according to the followingformula
119875 =anglemean (119898)
anglemean (119898 + 1) (6)
If 119875 is within the 09ndash11 range anglemean(119898) is the angleof the thick scale mark All the angles sorted out successivelyare taken as 120572
0 1205721 1205722 which correspond to the thick
scale marks Therefore the thick scale marks are located onFigure 12
414 Improved Hough Transform-Based Pointer DetectionThe edge detection is implemented in the pointer gaugeimage using the Canny operator to obtain the edge imageof the gauge Then the Hough transform method is adoptedto fit the straight line in the image to obtain all the straightlines among which the pointer border is included Finallyonly the pointer border can be determined by excluding theother straight edges
(1) The Hough transform method is adopted to fit thestraight line and find all the straight lines in the image Inaddition each straight line described in the image coordinatesystem by 119910 = 119896119909 + 119887 is calculated
(2)The distance 119889 from the circle center (1199090 1199100) to the
straight line is calculated according to the following formula
119889 =
10038161003816100381610038161199100 minus 119896 lowast 1199090 minus 1198871003816100381610038161003816
radic1 + 1198962 (7)
Based on prior knowledge we know that the distancefrom the circle center to the straight line of the pointer ismuch less than the diameter of the dial all the straight lines
Figure 13 Schematic of pointer fitting and its direction
conforming to 119889 lt 211990310 are picked out Here 119903 is thediameter of the dial
(3) The angle of each straight line in (1) is calculatedaccording to formula (8) and stored into an array angle[]
120593 =arctan (119896) lowast 180
120587 (8)
(4) As the angles of the pointer direction are acute onesgenerally in the range of 2∘ sim 10
∘ the angle Δ120593 is calculatedsubject to the following formula
Δ120593 =1003816100381610038161003816angle [119899] minus angle [119899 + 1]1003816100381610038161003816 (9)
If it is in the situation where 2 lt Δ120593 lt 10 the straight lineshould be selected as one side of the pointer Therefore bothlines where the two sides of the pointer lie can be found
(5) According to the straight lines of the two sides thepointer direction can be determined
42 The Pointer Gauge Indicating Value Recognition As isshown in Figure 11 the intersection point 119866 of the twostraight lines of the pointer border is calculated and thenlinked to the circle center 119874(119909
0 1199100) The angle of the
line 119874119866 is calculated In accordance with the position ofthe intersection point 119866 the angle 120573 in the rectangularcoordinate system is computed according to formula (1)(Figure 13)
According to the angle 120573 the readings 1198630 1198631 119863
119899
corresponding to each thick scale mark are obtained Basedon the angle that corresponds to the thick scale markand its number 119899 which is detected in Section 414 thepointer reading can be determined according to formula (10)
Mathematical Problems in Engineering 11
Table 1 Experimental samples
Sample name Image resolution Sample character The purpose of sample selectionThe pressure gauge with manylabels and a lot of writing on thedial
1000750A lot of writing on the dial which mayinfluence the location of the dial centerand the pointer segmentation
Verify the robustness of thealgorithm and the capability ofresisting interference
Flow gauge with the unevenscale 500500 Uneven distribution of the scale marks Verify the universality of the
algorithm
The scale within the adjacent thick scale marks is divided bylinear interpolation
119878indicating value
=
10038161003816100381610038161003816100381610038161003816
1205720minus 120573
1205720minus 1205721
10038161003816100381610038161003816100381610038161003816
(1198631minus 1198630) 120572
0lt 120573 lt 120572
1
10038161003816100381610038161003816100381610038161003816
1205721minus 120573
1205721minus 1205722
10038161003816100381610038161003816100381610038161003816
(1198632minus 1198631) + 1198631 120572
1lt 120573 le 120572
2
10038161003816100381610038161003816100381610038161003816
120572119899minus1
minus 120573
120572119899minus1
minus 120572119899
10038161003816100381610038161003816100381610038161003816
(119863119899minus 119863119899minus1
) + 119863119899minus1
120572119899minus1
lt 120573 le 120572119899
(10)
Here 1205720 1205721 1205722 120572
119899are the angles that correspond to the
thick scale marks
5 Experimental Results and Analysis
To verify the effectiveness of the proposed approach a seriesof experiments are conducted to test its performance
51 Experiment Design The experiments include two partsFirst the images of the different kinds of pointer gauges areselected to conduct the simulation experiments and verifythe effectiveness and applicability of the approach Thenthe pressure gauge calibration system which is composedof a float gauge high precise pressure gauge test pressuregauge camera and computer is established to verify theactual performance of this approach in the pointer gaugecalibration
52 Part One of the Experiments Reading Identification ofthe Various Pointer Gauges
521 Samples Selection As can be seen in Table 1 two groupsof representative samples are selected for the experimentThey include the pressure gauge with a lot of writing on thedial and the flow gauge with an uneven scale
522 Experimental Results and Analysis In this part ofthe experiment we adopt the proposed approach to theautomatic recognition of the indicating value of the abovementioned two types of gauges To verify the effectivenessof the proposed approach Figures 14 and 15 illustrate theexperimental results of each step
Figure 14(a) is the original pointer gauge image FromFigure 14(a) we can see that the pointer and scale marks
cannot be directly detected by segmentation methods sincethere is a lot of writing on the dial Therefore the proposedapproach applies an idea that conducts a coarse-to-finesearch The dial and scale regions are located first since theyare more easily detected than the pointer and scale marks Asshown in Figures 14(c) and 14(d) the dial region is locatedusing the adaptive-region-growing method and the scaleregion is detected under the polar coordinate system Thenin both regions the pointer and scale marks are accuratelydetected As shown in Figures 14(e) and 14(f) the improvedHough transform method is adopted to detect the pointerand the improved central projectionmethod is used to detectthe scale marks The location of the thick scale marks isdetermined according to the angle histogram generated fromthe central projection of the scale marks Their indicatingvalues are given according to the location of the thick scalemarks The scale values within any two adjacent thick marksare determined by linear division According to the directionof the pointer and the location of the scale marks the anglemethod which is adopted in the adjacent thick scale marksis used to indicate the pointer gauge value
The proposed approach detects the pointer direction andthe scale marks location respectively In Figures 15(e) and15(f) it can be seen that the location of the scale marks isprecisely obtained by the angle histogram and the pointer isdetected in the dial region Based on the location of the scalemarks and the pointer direction the distance between thepointer and the adjacent scale marks can be calculated usingthe distance method then the pointer indicating value canbe estimatedTherefore whether or not the distribution of thescale is even the proposed approach is able to precisely detectthe gaugersquos indicating value while the most current existingmethods cannot
53 Part Two of the Experiments The Recognition of the FloatPointer Gauge Indicating Value
531 The System and Procedure of Experiments To verifythe performance of the practical application of the proposedapproach in practical application the float gauge is used as thepressure gauge verification system The type of float gauge isY-047 The main technical parameters of the float gauge aredescribed as follows The range of output pressure is 001ndash025MPa the accuracy class is plusmn005 the rated pressureis 05MPa and the standard weights are from 001MPato 25MPa A high-precision gauge is used as the standardpressure gauge whose accuracy class is 15 The test gaugeis a gauge with general accuracy A visual detection systemis established in front of the float gauge An optical lens
12 Mathematical Problems in Engineering
(a) The original gauge image (b) The dial region of the manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 1205723 12057241205720 1205725 1205726 1205727 1205728 1205729 12057210
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 14 The detection results of the pointer gauge with a lot of writing on the dial
whose focal length can be adjustable from 10 to 20mm anda 500-megapixel HD CMOS camera with a 60 fps framerate are applied as image sensors The system is equippedwith a coaxial light source which is placed on the cameralens The hardware configuration of the computer used inthis experiment is Inter Pentium(R) Dual CPU E2200 and
22GHz The programming environment is VS2010 Theverification system of the pointer gauge that is based oncomputer vision is shown in Figure 16 The camera used inthe experiments takes images of the high-precision standardgauge and the test gauge The proposed approach is adoptedto detect the indicating value of both standard and test gauges
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
2 Mathematical Problems in Engineering
PC Image
Drive card Stepping
Other signal
CCD or
Stepping
High
standard
Tested
Worktable
Benchmarksignal
generator
collectors
capture card
motor driver
CMOS camera
motor
precision
gaugegauge
Figure 1 Hardware structure of the automatic recognition of the indicating value of the pointer gauge
Image Image
acquiring
Preprocessing gauge image
Pointerand scale
Automatic
reading
Gaugereadingoutput
Reading identification
preprocessing marks detection identification
Detection
of pointer
and scale
Figure 2 The image processing procedure of the automatic recognition of the indicating value of the pointer gauge
is adopted to obtain the indicating value of the gauge bycomparing the pointer direction with the position of the scalemarks
Since the indicating value is determined by comparing thepointer direction and the locations of the scale marks thisapproach is also suitable for gauges wherein the scale marksare unevenly distributed Therefore the proposed approachis in fact a general approach for gauge calibration
2 Related Work
21 Configuration of Hardware System Used in Gauge Ver-ification The hardware system used in gauge verificationgenerally comprises the following configuration [2ndash5] whichis shown in Figure 1 During the working process the highprecise gauge and the calibrated gauge are placed on theworktable A computer sends impulse signals to the motordriver Therefore the stepping motor is controlled to maketheworktablemove to allow for the adjustment of the locationof the gauge so that the gauges can be appropriately placedin the camerarsquos view while the standard signal source sendsreference signals to the calibrated pointer gauge and the highprecise gauge The indicating value of the high precise gaugeis treated as standard data Next computer vision is applied toautomatically recognize the values indicated by the calibratedand standard gauges Finally a comparison is made betweenthe two values and the verification result is reached
For some other pointer gauges the standard data pro-duced by the reference signal generator is directly inputted
into the computer by the other signal-acquisition equipmentThe verification result is achieved by comparing the standarddata with the indicating value of the calibrated gauge whichis extracted using computer vision
22 The Process of Automatic Reading of the Pointer GaugeBased on Computer Vision Figure 2 shows the procedure ofthe image processing used to recognize the pointer gaugevalue First to improve the image quality the gauge imagehas to be preprocessed by means of noise suppression andimage enhancement Next image processing and analyzingmethods are used to detect the pointer and scale marksAccording to the pointer direction and the location of scalemarks the value indicated by the pointer gauge can bedetected Finally the indicating value results are outputtedTherefore the automatic recognition methods of the gaugeindicating value mainly include three parts
(1) image preprocessing aimed to improve image qualityand suppress noise
(2) gauge pointer and scale marks detection based onimage processing and analyzing
(3) automatic recognition of the pointer gauge indicatingvalue
23 Related Work of the Pointer Gauge Indicating ValueDetection In the existing literature the pointer gaugeimage should be preprocessed to remove noise or enhance
Mathematical Problems in Engineering 3
the image especially for the images extracted from industrialfields Image preprocessing improves the image resolutionThus it is easier to detect the features relating to pointer andscale marks by means of image analyzing
Image preprocessing of the pointer gauge mainly com-prises two parts First image preprocessing is used to filterout interference and noise that result from the change ofenvironment For example in [6] the homomorphic filteringmethod was adopted to process the gauge image that hasa large area of glare and reflection Second image prepro-cessing is used to correct an image distortion caused bycamera imaging or different shooting angles For instance[7] discussed how to reduce the reading error when thegauge dial was uneven with the camera lens surface Thepapers [8 9] proposed an intelligent method of gauge valuereading to avoid the influence of imaging distortion thismethod was able to correct the angle between the pointer andthe zero-scale marks and the angle between the zero-scalemarks and the other longer scale marks It also adopted theNewton interpolation method to approximate the functionalrelationships between the pointer angle and its indicatingvalue
After image preprocessing image segmentation shouldbe implemented to obtain a binarization image wherein thethreshold-segmentation method is always applied [10ndash13] Inthe following step image thinning needs to be conducted inthe binarization image to achieve a one-pixel-wide pointerimageThen straight-line fitting is used to detect the pointerSo the key point of the automatic recognition of the gaugeindicating value is the detection of the pointer and scalemarks In the existing literature several methods for pointerdetection such as subtractionmethod the templatematchingmethod the Hough transform method and the central pro-jectionmethodwere proposed In the subtractionmethod [414 15] two gauge images with roughly the same backgroundswere taken by camera while the pointer pointed toward adifferent scaleWhen one image was subtracted from anotherone there were only two pointers left in the difference imageNext the difference image was segmented to detect thepointer Finally the gaugersquos indicating value could be obtainedby calculating the position of the pointer and scale marksThe subtraction method was easily implemented by meansof image subtraction However the different image not onlyreserves the pointer target but also leaves out a lot of noise dueto the influence of illuminationTherefore in many cases thesubtraction method does not produce reliable results
In the template feature method [16] the features thatrepresent the pointer and its direction such as shape areaand gray distribution formed amultiple parameters templateused to match the pointer in the gauge image for pointerdetection Since the template comprises multiple cues of thepointer and its direction the template feature method couldavoid the influence of an uneven illumination and otherenvironmental changing factors However the process of atemplate match is too complicated to accurately detect thepointer
The Hough transform method has often been used inpointer detection In [17 18] after edge detection and imagethinning in the gauge image the Hough transform method
or its improved version was applied to detect the pointer byfitting the pointer edge The Hough transform method couldachieve good results for the pointer gauge that has manywords and symbols on the dial In [19 20] a novel pointerdetection method called central projection was describedThemain idea of the central projection method was to searchfor the pointer location bymeans of calculating the projectionfrom the feature points to the known central point Everyprojection value corresponded to the angle of the straightline that connected the projection and central points Theremust be an angle that corresponds to the largest numberof projection points and represents the pointer Hence thecentral projection method was always used to determine thelocation of the gauge pointer
The least squares method [21] was also a straight-linefitting method just like the Hough transformmethod whichwas used to detect the pointer Compared with the Houghtransform method the least squares method required fewercalculations
After the pointer was detected angle and distance meth-ods [22]were always used to determine the indicating value ofthe point gauge The angle method recognized the indicatingvalue by calculating the deflection angle between the detectedpointer and the zero-scale mark according to the formula119872 times 120573120572 Here 119872 was the range 120572 was the premeasuredangle between the maximum- and zero-scale mark and 120573
was the angle between the current pointer and zero-scalemarkThe distancemethod [23] obtained the indicating valueby calculating the distance between the pointer and nearestscale mark In this method each scale reticule was locatedand labeled with the corresponding scale number in a two-dimensional coordinate system If the distance between thepointer and each reticule is as follows
119889119894=
1003816100381610038161003816119860119909119894 + 119861119910119894 + 1198621003816100381610038161003816
radic1198602 + 1198612 (1)
then the indicating value of the pointer is
119878119899= 119878119871+
1198891
1198891+ 1198890
times (119878119867minus 119878119871) (2)
Here (119909119894 119910119894) is the coordinate of the points in the scale marks
119889119894is the distance between the point and the line 119860119909 + 119861119910 +
119862 = 0 119878119899represents the indicating value of the pointer 119878
119867
and 119878119871are the values of the scale marks which are located
on both sides of the pointer and 1198891and 119889
0are the distance
between the pointer and the nearest scale marks on bothsides respectively The angle method is simpler than thedistance method but it is unsuitable for the pointer gauge inwhich the scale is uneven
3 Overview of Our Work
As mentioned above the current methods for gauge readingrecognition are limited in terms of their practical applicationmainly due to two deficiencies (1)Theexistingmethods wereall implemented in the global gauge images Therefore thepointer and scale marks segmentation cannot be accurately
4 Mathematical Problems in Engineering
conducted when there is some interference or where thereis some complex writing on the gauge dial (2) In manystudies the scale marks are not precisely detected Hence theindicating value of the pointer gauge with an uneven scalecannot be recognized
This study aims to overcome the shortcomings of theexisting methods As its contextual application it also aimsto improve the calibration of the pressure gauge and developan automatic recognition method of pointer gauge readingsbased on machine vision This method implements a schemethat takes detection from coarse to fine in terms of thepointer and scale marks detection First the pointer gaugersquosdial region is located in a global image Then the dialrsquosannular scale region is located In the following steps thescale marks are detected in the annular scale region and thepointer is detected in the dial region Finally based on theresults of the detection of the pointer and scale marks theimproved distance method is used to recognize the gaugersquosvalue Since the improved distance method is implementedfrom the coarse region to the fine target it has a strongability for suppressing interference Since the indicating valueis obtained from the pointer and scale marks detection themethod is also suitable for reading the gauge whose scalemarks are unevenly distributed
31 Machine Vision Based Pressure Gauge Verification SystemMetrological departments usually calibrate pressure gaugeswith piston manometers In this case the standard pressureoutput of piston manometers given by manually addedweights is passed on to the tested pressure gauge Then thequalification of the pressure gauge is determined throughreading analysis and comparison of the indicating value Ingeneral verification personnel manually handle this processthrough observation by recording and analyzing the indicat-ing value of the gauge panel Many problems still exist withthis verification method They include (1) low accuracy andrepeatability in human visual observations and (2) poor workefficiency due to the large workload of verification personnelHence development of an automatic verification methodfor the pressure pointer gauge will boost the advancementof the technology used in their production and verificationcapabilities The system composition for pointer pressuregauge verification is shown in Figure 3 We can see thatimages of pointer gauges are obtained through a camera andprocessed and analyzed by computer wherein the indicatingvalues of the standard pressure gauge and the test gaugeare detected and recognized The final results of the gaugereading will be displayed on amonitor to operators and reacha verification conclusion based on the analyses of the abovereadingsThe detection software also has suchmultifunctionsas the storage of historical data graphical display and imageswitching to realize the automatic verification of pointergauges
32 Introduction of OurWork In this study a new automaticrecognition method of the pointer gauge that indicates valueis proposed Figure 4 shows a block diagram of this method
In general the computer vision based automatic recog-nition method of indicating value comprises two parts One
Piston gaugeCalibrated gauge
Standard gauge
Display PC
Camera
Image capture card
Figure 3 Schematic of a machine vision based pressure gaugeverification system
part is the detection of the indicating value of the pointergauge This mainly includes the scale detection and thepointer detection which refer to the location of the dialregion and the dial center the determination of the annularscale region in the dial the scale detection and the pointerdetection The other is the automatic recognition of thereading Namely it makes the final judgment of the gaugereading according to the detected direction of the pointer andthe distribution of the scale marks As seen in Figure 4 theautomatic recognition of the indicating value proposed inthis paper falls into five steps
(1) The dial region and its center location first in theentire image the dial region can be determined bysemi-interaction (using the mouse to drop downthe box in the image) The pixel point with thebiggest gray value is searched out in the dial regionand taken as the seed point of the region growingmethod used to produce the circular region in thedial In the process of the region growing methodthe similarity threshold value between the pixel pointsis automatically determined Finally according to theresults of the region growing method the radius andcenter of the circular dial are obtained
(2) Determination of the scale region in the dial theimage coordinate is first converted into the polarcoordinate with the dial center as its origin whichis used to express the coordinate of each pixel in theimage Then by comparing the pixel sum of the lowgray value with that of the high gray value in theannular region the annular scale region in the dial isdetermined And finally it is copied into a new imageas the annular scale region
(3) The scale marks are detected using the central pro-jection method In the newly generated annular scaleregion image first the scale mark segmentation isconducted by theOSTUmethod to form the binariza-tion scale imageThen the angle of each line that linksthe black pixel with the circle center is calculatedand the number of each angle is obtained The angleswith large numbers (connecting line) in a narrowangle range correspond to the thick scalemarks (mainscales) on the dial Thus the angles of the scale marksin the scope of 0∘ndash360∘ are obtained
Mathematical Problems in Engineering 5
Scale detection by central projection method
Segment the image in annular scaleregion by OSTU to obtain the
binarization scale image
Traverse the black pixel points in theimage link each black point to the
circle center and then calculate the angleof the connecting line to get that of
each coarse scale mark
Detection of gauge pointer by Houghtransform
Utilize the Canny operator to performborder detection on the dial image
Use the adaptive threshold value toperform the Hough transform on dial
border image and fit the gauge pointercontour
Calculate the pointer direction within
The indicating value reading
Calculate the indicating valueaccording to the pointer direction and
the dial scale distribution
Dial region and center location
Search the pixel point with maximumgray in the dial region as the seed
Conduct region growing from the seedpoint and adaptively determine thesimilarity principle between the pixels
Obtain the center and radius ofcircular dial region by region growing
Determination of annular scale region in the dial
Convert the image into the polarcoordinates with the dial center as its
origin center
Determine the annular scale region bycomparison between the pixel sum oflow gray value and that of high gray
value
Copy the circular scale regionestablished into a new image
the scope of 0ndash Obtain the scale in the scope of 0ndash360degrees
360 degrees
Figure 4 The flow chart of the pointer gauge indicating value recognition
(4) Pointer detection first the Canny operator is usedto implement the border detection Then the Houghtransform method is used to fit the detected edgeand obtain the pointer contour Finally the pointerdirection is obtained according to the center positionand the pointer contour
(5) The indicating value is calculated according to thepointer direction and the dial scale distribution
4 Indicating Value Recognition Approach ofPointer Gauge
41 Pointer Gauge Indicating Value Detection As mentionedabove the detection of the indicating value is achievedthrough the scale and pointer detection As for the scaledetection first the gauge center and dial region should belocated followed by the extraction of the annular scale region
6 Mathematical Problems in Engineering
(a) Typical pointer gauge (b) The results of the dial region growing method
Figure 5 The diagram of the dial region growing results
and the scale mark detection Finally the detection of thepointer direction is performed
411 Dial Region and Its Center Location As for the detectionof the indicating value first the gauge pointer and its directionshould be detected to determine its rotation center In generalthe gaugersquos dial center is the pointerrsquos rotation center Hencefirst the dial center should be detected In this study thecircular dial region is detected for determining the dial centerAs is shown in Figure 5(a) typically there are the labelswords and scale marks in the circular dial Therefore it isdifficult to confirm the circular region of the dial by suchsegment methods as the border and threshold values Inview of the above this paper introduces the adaptive-region-growing method to locate the gaugersquos circular region andfurther to obtain its dial center The specific steps are asfollows
(1) The placement of the seed [24] since the dial isusually white the region with a high gray level orthe pixel point with the highest gray level in thegauge is certainly located at the part where the dialis Therefore the latter is selected as the seed point ofthe region growing
(2) Subject to the similarity of the pixel gray level theexpansion of the dial region is conducted from theseed point However it is hard to fix an appropriatesimilarity threshold value of the gray level betweenpixels As is shown in Figure 5(b) the circular dialregion cannot be completely produced with a smallthreshold value while an oversized dial region will begenerated with a large threshold value
This study proposes an adaptive method to determinethe similarity threshold value of the gray level the concreteprocess is as follows
(A) A small similarity threshold value 119879 of the graylevel is empirically chosen and then the dial regiongrowing method is conducted from the seed point
(B) The ratio of the circumscribed rectanglersquos length andwidth of the newly generated dial region is calculatedto detect whether it is in the 095ndash105 range
(C) If the ratio of the length and width fails to meetthe conditions in (B) the similarity threshold valuewould be increased with a certain step length (119878)and the first step (A) would continue to conductthe region growing from the seed point Howeverif it meets the conditions in (B) the growing regionwill be the circular dial region the center of thecircumscribed rectangle will be the circle center andthe smaller value between the length and width willbe the diameter of the circular region (Figure 6)
412 Polar Coordinate Based Scale Region Location
(1) Image in Polar Coordinates In general in light of thecircular permutation the pointer gauge scale marks aredistributed around the center Therefore to correctly andexpediently show where the scale marks in the image arethe polar coordinate system is utilized to position the pixelsin the image In this study a conversion from the imagecoordinates into the polar coordinates is performed Thismeans that the pixel coordinates expressed by the rectangularimage coordinate system are converted into those expressedby polar coordinates The polar coordinates defined in thisstudy are shown in Figure 7 the circle center being its origin
(2) Detection of the Scale Region in the Dial The procedurefor scale region detection based on the adaptive thresholdmethod is as follows
(A) The circular dial radius detected in Section 411 istaken as the initial radius 119877 its decreasing step lengthbeing 119887 and the initial 119899 = 1 In the gauge dial thegray value difference between the scale marks anddial background is generally rather large Thereforethe adaptive threshold value method can be adoptedto segment the scale In the polar coordinate theannular binarization image is gained by employing
Mathematical Problems in Engineering 7
(a) (b)
(c) (d)
Figure 6 The region growing results under different pixel similarity threshold values
90∘
60∘
210∘
150∘
120∘
180∘
240∘
270∘300∘
330∘
30∘
0∘
Figure 7 Schematic of the polar coordinate
the OSTU method [25] to perform the binarizationsegmentation in the annular region
(B) The ratio between the number of black andwhite pixelpoints in the annular binarization region between theradius 119877 and 119877minus119899lowast119887 is calculated and denoted by 119903
1
After reducing the inner radius according to the steplength 119887 the ratio between the number of black and
white pixel points in the region between 119877minus(119899+1)lowast119887and 119877 is computed and denoted by 119903
2
(C) If 1199031is obviously different from 119903
2 the inner radius
and the outer radius of the scale region would be 119877 minus(119899 + 1) lowast 119887 and 119877 respectively
(D) If 1199031is approximated to 119903
2 the annulus between 119877 minus
(119899 + 1) lowast 119887 and 119877 would still be in the circular scaleregion At thismoment let 119899 = 119899+1 the ratio betweenthe number of black and white pixel points in theannulus between 119877 minus (119899 + 1) lowast 119887 and 119877 is computedand is still denoted by 119903
2 Then go to (B) (Figure 8)
After obtaining the annular scale region in the abovepolar coordinate the scale region is selected on the basis ofangle and radius and copied into a new image called the scalemark region image119872 (Figure 9)
413 Scale Mark Detection Based on the Improved CentralProjection Method In the newly generated image119872 whichis the annular scale region the OSTU method is adopted tosegment the scale marks to form the binarized scale imageIn this image the pixel value of the scale mark is 0 (black)and those in other places are 255 (white) Each black pixelpoint and the circle center are linked and the slope 119870 ofthe connecting line is calculated according to the formula119896 = tan120572 wherein 120572 is the angle between the 119909-axis and
8 Mathematical Problems in Engineering
(a) (b)
(c)
Figure 8 Schematic of scale mark region detection
(a) (b)
Figure 9 The scale mark region diagram and its binarized image
Mathematical Problems in Engineering 9
x
y
0
(a) Schematic of the image coordinate system
x
y
1
1
2
2
3
3
4
4
minus1
minus1
minus2
minus2
0minus3
minus3
minus4
minus4
(b) The rectangular coordinate system
Figure 10 Schematic of the image coordinate system and the rectangular coordinate system
the connecting line between the black pixel point and thecircular center Due to the origin of the image coordinatebeing in the top left corner and the119910-axis running downwardit is inconvenient to denote the central angle within thescope of 0∘ndash360∘ Therefore in this study the rectangularcoordinate system with the image center as its origin isintroduced to express the angle between the black pixelpoint on the scale mark and the circle center instead of theimage coordinate system As is shown in Figure 10 the anglesbetween the black pixel point and the circle center are in the0∘ndash360∘ range
After calculating the number for angle 120572 the angle witha large number (connecting lines) in a small angle range isthat of the thick scale mark (main scale) on the dialThe stepsto detect the scale mark by central projection are detailed asfollows
(1) The black pixel point and the image central pointthat is the circle center (119909
0 1199100) in the new image are linked
to determine the slope 119896 which is converted to an angleaccording to the following formula
120572 = arctan (119896) lowast 180
120587 (3)
(2) As shown in Figure 10 the image coordinate systemis transformed to a rectangular one The angle 120572 followingformula (4) is converted into an angle in the rectangularcoordinate
120579 = minus120572 119894 lt 119903 119895 gt 119903 the first quadrant
120579 = 180 minus 120572 119894 lt 119903 119895 lt 119903 the second quadrant
120579 = 180 minus 120572 119894 gt 119903 119895 lt 119903 the third quadrant
120579 = 360 minus 120572 119894 gt 119903 119895 gt 119903 the forth quadrant
(4)
Occ
urre
nce t
imes
Angle (deg)1205791 1205792 1205793 12057940
Figure 11 The angle histogram
(3) An angle histogram is generated by recording theoccurrence frequency of each angle Figure 11 shows the anglehistogram
(4) The angle with the most occurrence times in thehistogram is selected and denoted by max all the angleswhose occurrence times are more thanmax 2 are foundTheoccurrence times of the angles are denoted by 119891(120579) Here 120579represents the angles that occur more than max 2
(5) For every angle 120579 we calculate the weighted averagevalue anglemean(119898) according the following to formula
anglemean (119898)
=119891 (120579 + 1) lowast (120579 + 1) + 119891 (120579) lowast (120579) + 119891 (120579 minus 1) lowast (120579 minus 1)
119891 (120579 + 1) + 119891 (120579) + 119891 (120579 minus 1)
(5)
Here119898 is the sign of the angle
10 Mathematical Problems in Engineering
12057221205721
1205720
1205723
Thick scale marksy
x
Figure 12 The location of the thick scale marks
(6) Calculate specific value 119875 according to the followingformula
119875 =anglemean (119898)
anglemean (119898 + 1) (6)
If 119875 is within the 09ndash11 range anglemean(119898) is the angleof the thick scale mark All the angles sorted out successivelyare taken as 120572
0 1205721 1205722 which correspond to the thick
scale marks Therefore the thick scale marks are located onFigure 12
414 Improved Hough Transform-Based Pointer DetectionThe edge detection is implemented in the pointer gaugeimage using the Canny operator to obtain the edge imageof the gauge Then the Hough transform method is adoptedto fit the straight line in the image to obtain all the straightlines among which the pointer border is included Finallyonly the pointer border can be determined by excluding theother straight edges
(1) The Hough transform method is adopted to fit thestraight line and find all the straight lines in the image Inaddition each straight line described in the image coordinatesystem by 119910 = 119896119909 + 119887 is calculated
(2)The distance 119889 from the circle center (1199090 1199100) to the
straight line is calculated according to the following formula
119889 =
10038161003816100381610038161199100 minus 119896 lowast 1199090 minus 1198871003816100381610038161003816
radic1 + 1198962 (7)
Based on prior knowledge we know that the distancefrom the circle center to the straight line of the pointer ismuch less than the diameter of the dial all the straight lines
Figure 13 Schematic of pointer fitting and its direction
conforming to 119889 lt 211990310 are picked out Here 119903 is thediameter of the dial
(3) The angle of each straight line in (1) is calculatedaccording to formula (8) and stored into an array angle[]
120593 =arctan (119896) lowast 180
120587 (8)
(4) As the angles of the pointer direction are acute onesgenerally in the range of 2∘ sim 10
∘ the angle Δ120593 is calculatedsubject to the following formula
Δ120593 =1003816100381610038161003816angle [119899] minus angle [119899 + 1]1003816100381610038161003816 (9)
If it is in the situation where 2 lt Δ120593 lt 10 the straight lineshould be selected as one side of the pointer Therefore bothlines where the two sides of the pointer lie can be found
(5) According to the straight lines of the two sides thepointer direction can be determined
42 The Pointer Gauge Indicating Value Recognition As isshown in Figure 11 the intersection point 119866 of the twostraight lines of the pointer border is calculated and thenlinked to the circle center 119874(119909
0 1199100) The angle of the
line 119874119866 is calculated In accordance with the position ofthe intersection point 119866 the angle 120573 in the rectangularcoordinate system is computed according to formula (1)(Figure 13)
According to the angle 120573 the readings 1198630 1198631 119863
119899
corresponding to each thick scale mark are obtained Basedon the angle that corresponds to the thick scale markand its number 119899 which is detected in Section 414 thepointer reading can be determined according to formula (10)
Mathematical Problems in Engineering 11
Table 1 Experimental samples
Sample name Image resolution Sample character The purpose of sample selectionThe pressure gauge with manylabels and a lot of writing on thedial
1000750A lot of writing on the dial which mayinfluence the location of the dial centerand the pointer segmentation
Verify the robustness of thealgorithm and the capability ofresisting interference
Flow gauge with the unevenscale 500500 Uneven distribution of the scale marks Verify the universality of the
algorithm
The scale within the adjacent thick scale marks is divided bylinear interpolation
119878indicating value
=
10038161003816100381610038161003816100381610038161003816
1205720minus 120573
1205720minus 1205721
10038161003816100381610038161003816100381610038161003816
(1198631minus 1198630) 120572
0lt 120573 lt 120572
1
10038161003816100381610038161003816100381610038161003816
1205721minus 120573
1205721minus 1205722
10038161003816100381610038161003816100381610038161003816
(1198632minus 1198631) + 1198631 120572
1lt 120573 le 120572
2
10038161003816100381610038161003816100381610038161003816
120572119899minus1
minus 120573
120572119899minus1
minus 120572119899
10038161003816100381610038161003816100381610038161003816
(119863119899minus 119863119899minus1
) + 119863119899minus1
120572119899minus1
lt 120573 le 120572119899
(10)
Here 1205720 1205721 1205722 120572
119899are the angles that correspond to the
thick scale marks
5 Experimental Results and Analysis
To verify the effectiveness of the proposed approach a seriesof experiments are conducted to test its performance
51 Experiment Design The experiments include two partsFirst the images of the different kinds of pointer gauges areselected to conduct the simulation experiments and verifythe effectiveness and applicability of the approach Thenthe pressure gauge calibration system which is composedof a float gauge high precise pressure gauge test pressuregauge camera and computer is established to verify theactual performance of this approach in the pointer gaugecalibration
52 Part One of the Experiments Reading Identification ofthe Various Pointer Gauges
521 Samples Selection As can be seen in Table 1 two groupsof representative samples are selected for the experimentThey include the pressure gauge with a lot of writing on thedial and the flow gauge with an uneven scale
522 Experimental Results and Analysis In this part ofthe experiment we adopt the proposed approach to theautomatic recognition of the indicating value of the abovementioned two types of gauges To verify the effectivenessof the proposed approach Figures 14 and 15 illustrate theexperimental results of each step
Figure 14(a) is the original pointer gauge image FromFigure 14(a) we can see that the pointer and scale marks
cannot be directly detected by segmentation methods sincethere is a lot of writing on the dial Therefore the proposedapproach applies an idea that conducts a coarse-to-finesearch The dial and scale regions are located first since theyare more easily detected than the pointer and scale marks Asshown in Figures 14(c) and 14(d) the dial region is locatedusing the adaptive-region-growing method and the scaleregion is detected under the polar coordinate system Thenin both regions the pointer and scale marks are accuratelydetected As shown in Figures 14(e) and 14(f) the improvedHough transform method is adopted to detect the pointerand the improved central projectionmethod is used to detectthe scale marks The location of the thick scale marks isdetermined according to the angle histogram generated fromthe central projection of the scale marks Their indicatingvalues are given according to the location of the thick scalemarks The scale values within any two adjacent thick marksare determined by linear division According to the directionof the pointer and the location of the scale marks the anglemethod which is adopted in the adjacent thick scale marksis used to indicate the pointer gauge value
The proposed approach detects the pointer direction andthe scale marks location respectively In Figures 15(e) and15(f) it can be seen that the location of the scale marks isprecisely obtained by the angle histogram and the pointer isdetected in the dial region Based on the location of the scalemarks and the pointer direction the distance between thepointer and the adjacent scale marks can be calculated usingthe distance method then the pointer indicating value canbe estimatedTherefore whether or not the distribution of thescale is even the proposed approach is able to precisely detectthe gaugersquos indicating value while the most current existingmethods cannot
53 Part Two of the Experiments The Recognition of the FloatPointer Gauge Indicating Value
531 The System and Procedure of Experiments To verifythe performance of the practical application of the proposedapproach in practical application the float gauge is used as thepressure gauge verification system The type of float gauge isY-047 The main technical parameters of the float gauge aredescribed as follows The range of output pressure is 001ndash025MPa the accuracy class is plusmn005 the rated pressureis 05MPa and the standard weights are from 001MPato 25MPa A high-precision gauge is used as the standardpressure gauge whose accuracy class is 15 The test gaugeis a gauge with general accuracy A visual detection systemis established in front of the float gauge An optical lens
12 Mathematical Problems in Engineering
(a) The original gauge image (b) The dial region of the manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 1205723 12057241205720 1205725 1205726 1205727 1205728 1205729 12057210
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 14 The detection results of the pointer gauge with a lot of writing on the dial
whose focal length can be adjustable from 10 to 20mm anda 500-megapixel HD CMOS camera with a 60 fps framerate are applied as image sensors The system is equippedwith a coaxial light source which is placed on the cameralens The hardware configuration of the computer used inthis experiment is Inter Pentium(R) Dual CPU E2200 and
22GHz The programming environment is VS2010 Theverification system of the pointer gauge that is based oncomputer vision is shown in Figure 16 The camera used inthe experiments takes images of the high-precision standardgauge and the test gauge The proposed approach is adoptedto detect the indicating value of both standard and test gauges
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 3
the image especially for the images extracted from industrialfields Image preprocessing improves the image resolutionThus it is easier to detect the features relating to pointer andscale marks by means of image analyzing
Image preprocessing of the pointer gauge mainly com-prises two parts First image preprocessing is used to filterout interference and noise that result from the change ofenvironment For example in [6] the homomorphic filteringmethod was adopted to process the gauge image that hasa large area of glare and reflection Second image prepro-cessing is used to correct an image distortion caused bycamera imaging or different shooting angles For instance[7] discussed how to reduce the reading error when thegauge dial was uneven with the camera lens surface Thepapers [8 9] proposed an intelligent method of gauge valuereading to avoid the influence of imaging distortion thismethod was able to correct the angle between the pointer andthe zero-scale marks and the angle between the zero-scalemarks and the other longer scale marks It also adopted theNewton interpolation method to approximate the functionalrelationships between the pointer angle and its indicatingvalue
After image preprocessing image segmentation shouldbe implemented to obtain a binarization image wherein thethreshold-segmentation method is always applied [10ndash13] Inthe following step image thinning needs to be conducted inthe binarization image to achieve a one-pixel-wide pointerimageThen straight-line fitting is used to detect the pointerSo the key point of the automatic recognition of the gaugeindicating value is the detection of the pointer and scalemarks In the existing literature several methods for pointerdetection such as subtractionmethod the templatematchingmethod the Hough transform method and the central pro-jectionmethodwere proposed In the subtractionmethod [414 15] two gauge images with roughly the same backgroundswere taken by camera while the pointer pointed toward adifferent scaleWhen one image was subtracted from anotherone there were only two pointers left in the difference imageNext the difference image was segmented to detect thepointer Finally the gaugersquos indicating value could be obtainedby calculating the position of the pointer and scale marksThe subtraction method was easily implemented by meansof image subtraction However the different image not onlyreserves the pointer target but also leaves out a lot of noise dueto the influence of illuminationTherefore in many cases thesubtraction method does not produce reliable results
In the template feature method [16] the features thatrepresent the pointer and its direction such as shape areaand gray distribution formed amultiple parameters templateused to match the pointer in the gauge image for pointerdetection Since the template comprises multiple cues of thepointer and its direction the template feature method couldavoid the influence of an uneven illumination and otherenvironmental changing factors However the process of atemplate match is too complicated to accurately detect thepointer
The Hough transform method has often been used inpointer detection In [17 18] after edge detection and imagethinning in the gauge image the Hough transform method
or its improved version was applied to detect the pointer byfitting the pointer edge The Hough transform method couldachieve good results for the pointer gauge that has manywords and symbols on the dial In [19 20] a novel pointerdetection method called central projection was describedThemain idea of the central projection method was to searchfor the pointer location bymeans of calculating the projectionfrom the feature points to the known central point Everyprojection value corresponded to the angle of the straightline that connected the projection and central points Theremust be an angle that corresponds to the largest numberof projection points and represents the pointer Hence thecentral projection method was always used to determine thelocation of the gauge pointer
The least squares method [21] was also a straight-linefitting method just like the Hough transformmethod whichwas used to detect the pointer Compared with the Houghtransform method the least squares method required fewercalculations
After the pointer was detected angle and distance meth-ods [22]were always used to determine the indicating value ofthe point gauge The angle method recognized the indicatingvalue by calculating the deflection angle between the detectedpointer and the zero-scale mark according to the formula119872 times 120573120572 Here 119872 was the range 120572 was the premeasuredangle between the maximum- and zero-scale mark and 120573
was the angle between the current pointer and zero-scalemarkThe distancemethod [23] obtained the indicating valueby calculating the distance between the pointer and nearestscale mark In this method each scale reticule was locatedand labeled with the corresponding scale number in a two-dimensional coordinate system If the distance between thepointer and each reticule is as follows
119889119894=
1003816100381610038161003816119860119909119894 + 119861119910119894 + 1198621003816100381610038161003816
radic1198602 + 1198612 (1)
then the indicating value of the pointer is
119878119899= 119878119871+
1198891
1198891+ 1198890
times (119878119867minus 119878119871) (2)
Here (119909119894 119910119894) is the coordinate of the points in the scale marks
119889119894is the distance between the point and the line 119860119909 + 119861119910 +
119862 = 0 119878119899represents the indicating value of the pointer 119878
119867
and 119878119871are the values of the scale marks which are located
on both sides of the pointer and 1198891and 119889
0are the distance
between the pointer and the nearest scale marks on bothsides respectively The angle method is simpler than thedistance method but it is unsuitable for the pointer gauge inwhich the scale is uneven
3 Overview of Our Work
As mentioned above the current methods for gauge readingrecognition are limited in terms of their practical applicationmainly due to two deficiencies (1)Theexistingmethods wereall implemented in the global gauge images Therefore thepointer and scale marks segmentation cannot be accurately
4 Mathematical Problems in Engineering
conducted when there is some interference or where thereis some complex writing on the gauge dial (2) In manystudies the scale marks are not precisely detected Hence theindicating value of the pointer gauge with an uneven scalecannot be recognized
This study aims to overcome the shortcomings of theexisting methods As its contextual application it also aimsto improve the calibration of the pressure gauge and developan automatic recognition method of pointer gauge readingsbased on machine vision This method implements a schemethat takes detection from coarse to fine in terms of thepointer and scale marks detection First the pointer gaugersquosdial region is located in a global image Then the dialrsquosannular scale region is located In the following steps thescale marks are detected in the annular scale region and thepointer is detected in the dial region Finally based on theresults of the detection of the pointer and scale marks theimproved distance method is used to recognize the gaugersquosvalue Since the improved distance method is implementedfrom the coarse region to the fine target it has a strongability for suppressing interference Since the indicating valueis obtained from the pointer and scale marks detection themethod is also suitable for reading the gauge whose scalemarks are unevenly distributed
31 Machine Vision Based Pressure Gauge Verification SystemMetrological departments usually calibrate pressure gaugeswith piston manometers In this case the standard pressureoutput of piston manometers given by manually addedweights is passed on to the tested pressure gauge Then thequalification of the pressure gauge is determined throughreading analysis and comparison of the indicating value Ingeneral verification personnel manually handle this processthrough observation by recording and analyzing the indicat-ing value of the gauge panel Many problems still exist withthis verification method They include (1) low accuracy andrepeatability in human visual observations and (2) poor workefficiency due to the large workload of verification personnelHence development of an automatic verification methodfor the pressure pointer gauge will boost the advancementof the technology used in their production and verificationcapabilities The system composition for pointer pressuregauge verification is shown in Figure 3 We can see thatimages of pointer gauges are obtained through a camera andprocessed and analyzed by computer wherein the indicatingvalues of the standard pressure gauge and the test gaugeare detected and recognized The final results of the gaugereading will be displayed on amonitor to operators and reacha verification conclusion based on the analyses of the abovereadingsThe detection software also has suchmultifunctionsas the storage of historical data graphical display and imageswitching to realize the automatic verification of pointergauges
32 Introduction of OurWork In this study a new automaticrecognition method of the pointer gauge that indicates valueis proposed Figure 4 shows a block diagram of this method
In general the computer vision based automatic recog-nition method of indicating value comprises two parts One
Piston gaugeCalibrated gauge
Standard gauge
Display PC
Camera
Image capture card
Figure 3 Schematic of a machine vision based pressure gaugeverification system
part is the detection of the indicating value of the pointergauge This mainly includes the scale detection and thepointer detection which refer to the location of the dialregion and the dial center the determination of the annularscale region in the dial the scale detection and the pointerdetection The other is the automatic recognition of thereading Namely it makes the final judgment of the gaugereading according to the detected direction of the pointer andthe distribution of the scale marks As seen in Figure 4 theautomatic recognition of the indicating value proposed inthis paper falls into five steps
(1) The dial region and its center location first in theentire image the dial region can be determined bysemi-interaction (using the mouse to drop downthe box in the image) The pixel point with thebiggest gray value is searched out in the dial regionand taken as the seed point of the region growingmethod used to produce the circular region in thedial In the process of the region growing methodthe similarity threshold value between the pixel pointsis automatically determined Finally according to theresults of the region growing method the radius andcenter of the circular dial are obtained
(2) Determination of the scale region in the dial theimage coordinate is first converted into the polarcoordinate with the dial center as its origin whichis used to express the coordinate of each pixel in theimage Then by comparing the pixel sum of the lowgray value with that of the high gray value in theannular region the annular scale region in the dial isdetermined And finally it is copied into a new imageas the annular scale region
(3) The scale marks are detected using the central pro-jection method In the newly generated annular scaleregion image first the scale mark segmentation isconducted by theOSTUmethod to form the binariza-tion scale imageThen the angle of each line that linksthe black pixel with the circle center is calculatedand the number of each angle is obtained The angleswith large numbers (connecting line) in a narrowangle range correspond to the thick scalemarks (mainscales) on the dial Thus the angles of the scale marksin the scope of 0∘ndash360∘ are obtained
Mathematical Problems in Engineering 5
Scale detection by central projection method
Segment the image in annular scaleregion by OSTU to obtain the
binarization scale image
Traverse the black pixel points in theimage link each black point to the
circle center and then calculate the angleof the connecting line to get that of
each coarse scale mark
Detection of gauge pointer by Houghtransform
Utilize the Canny operator to performborder detection on the dial image
Use the adaptive threshold value toperform the Hough transform on dial
border image and fit the gauge pointercontour
Calculate the pointer direction within
The indicating value reading
Calculate the indicating valueaccording to the pointer direction and
the dial scale distribution
Dial region and center location
Search the pixel point with maximumgray in the dial region as the seed
Conduct region growing from the seedpoint and adaptively determine thesimilarity principle between the pixels
Obtain the center and radius ofcircular dial region by region growing
Determination of annular scale region in the dial
Convert the image into the polarcoordinates with the dial center as its
origin center
Determine the annular scale region bycomparison between the pixel sum oflow gray value and that of high gray
value
Copy the circular scale regionestablished into a new image
the scope of 0ndash Obtain the scale in the scope of 0ndash360degrees
360 degrees
Figure 4 The flow chart of the pointer gauge indicating value recognition
(4) Pointer detection first the Canny operator is usedto implement the border detection Then the Houghtransform method is used to fit the detected edgeand obtain the pointer contour Finally the pointerdirection is obtained according to the center positionand the pointer contour
(5) The indicating value is calculated according to thepointer direction and the dial scale distribution
4 Indicating Value Recognition Approach ofPointer Gauge
41 Pointer Gauge Indicating Value Detection As mentionedabove the detection of the indicating value is achievedthrough the scale and pointer detection As for the scaledetection first the gauge center and dial region should belocated followed by the extraction of the annular scale region
6 Mathematical Problems in Engineering
(a) Typical pointer gauge (b) The results of the dial region growing method
Figure 5 The diagram of the dial region growing results
and the scale mark detection Finally the detection of thepointer direction is performed
411 Dial Region and Its Center Location As for the detectionof the indicating value first the gauge pointer and its directionshould be detected to determine its rotation center In generalthe gaugersquos dial center is the pointerrsquos rotation center Hencefirst the dial center should be detected In this study thecircular dial region is detected for determining the dial centerAs is shown in Figure 5(a) typically there are the labelswords and scale marks in the circular dial Therefore it isdifficult to confirm the circular region of the dial by suchsegment methods as the border and threshold values Inview of the above this paper introduces the adaptive-region-growing method to locate the gaugersquos circular region andfurther to obtain its dial center The specific steps are asfollows
(1) The placement of the seed [24] since the dial isusually white the region with a high gray level orthe pixel point with the highest gray level in thegauge is certainly located at the part where the dialis Therefore the latter is selected as the seed point ofthe region growing
(2) Subject to the similarity of the pixel gray level theexpansion of the dial region is conducted from theseed point However it is hard to fix an appropriatesimilarity threshold value of the gray level betweenpixels As is shown in Figure 5(b) the circular dialregion cannot be completely produced with a smallthreshold value while an oversized dial region will begenerated with a large threshold value
This study proposes an adaptive method to determinethe similarity threshold value of the gray level the concreteprocess is as follows
(A) A small similarity threshold value 119879 of the graylevel is empirically chosen and then the dial regiongrowing method is conducted from the seed point
(B) The ratio of the circumscribed rectanglersquos length andwidth of the newly generated dial region is calculatedto detect whether it is in the 095ndash105 range
(C) If the ratio of the length and width fails to meetthe conditions in (B) the similarity threshold valuewould be increased with a certain step length (119878)and the first step (A) would continue to conductthe region growing from the seed point Howeverif it meets the conditions in (B) the growing regionwill be the circular dial region the center of thecircumscribed rectangle will be the circle center andthe smaller value between the length and width willbe the diameter of the circular region (Figure 6)
412 Polar Coordinate Based Scale Region Location
(1) Image in Polar Coordinates In general in light of thecircular permutation the pointer gauge scale marks aredistributed around the center Therefore to correctly andexpediently show where the scale marks in the image arethe polar coordinate system is utilized to position the pixelsin the image In this study a conversion from the imagecoordinates into the polar coordinates is performed Thismeans that the pixel coordinates expressed by the rectangularimage coordinate system are converted into those expressedby polar coordinates The polar coordinates defined in thisstudy are shown in Figure 7 the circle center being its origin
(2) Detection of the Scale Region in the Dial The procedurefor scale region detection based on the adaptive thresholdmethod is as follows
(A) The circular dial radius detected in Section 411 istaken as the initial radius 119877 its decreasing step lengthbeing 119887 and the initial 119899 = 1 In the gauge dial thegray value difference between the scale marks anddial background is generally rather large Thereforethe adaptive threshold value method can be adoptedto segment the scale In the polar coordinate theannular binarization image is gained by employing
Mathematical Problems in Engineering 7
(a) (b)
(c) (d)
Figure 6 The region growing results under different pixel similarity threshold values
90∘
60∘
210∘
150∘
120∘
180∘
240∘
270∘300∘
330∘
30∘
0∘
Figure 7 Schematic of the polar coordinate
the OSTU method [25] to perform the binarizationsegmentation in the annular region
(B) The ratio between the number of black andwhite pixelpoints in the annular binarization region between theradius 119877 and 119877minus119899lowast119887 is calculated and denoted by 119903
1
After reducing the inner radius according to the steplength 119887 the ratio between the number of black and
white pixel points in the region between 119877minus(119899+1)lowast119887and 119877 is computed and denoted by 119903
2
(C) If 1199031is obviously different from 119903
2 the inner radius
and the outer radius of the scale region would be 119877 minus(119899 + 1) lowast 119887 and 119877 respectively
(D) If 1199031is approximated to 119903
2 the annulus between 119877 minus
(119899 + 1) lowast 119887 and 119877 would still be in the circular scaleregion At thismoment let 119899 = 119899+1 the ratio betweenthe number of black and white pixel points in theannulus between 119877 minus (119899 + 1) lowast 119887 and 119877 is computedand is still denoted by 119903
2 Then go to (B) (Figure 8)
After obtaining the annular scale region in the abovepolar coordinate the scale region is selected on the basis ofangle and radius and copied into a new image called the scalemark region image119872 (Figure 9)
413 Scale Mark Detection Based on the Improved CentralProjection Method In the newly generated image119872 whichis the annular scale region the OSTU method is adopted tosegment the scale marks to form the binarized scale imageIn this image the pixel value of the scale mark is 0 (black)and those in other places are 255 (white) Each black pixelpoint and the circle center are linked and the slope 119870 ofthe connecting line is calculated according to the formula119896 = tan120572 wherein 120572 is the angle between the 119909-axis and
8 Mathematical Problems in Engineering
(a) (b)
(c)
Figure 8 Schematic of scale mark region detection
(a) (b)
Figure 9 The scale mark region diagram and its binarized image
Mathematical Problems in Engineering 9
x
y
0
(a) Schematic of the image coordinate system
x
y
1
1
2
2
3
3
4
4
minus1
minus1
minus2
minus2
0minus3
minus3
minus4
minus4
(b) The rectangular coordinate system
Figure 10 Schematic of the image coordinate system and the rectangular coordinate system
the connecting line between the black pixel point and thecircular center Due to the origin of the image coordinatebeing in the top left corner and the119910-axis running downwardit is inconvenient to denote the central angle within thescope of 0∘ndash360∘ Therefore in this study the rectangularcoordinate system with the image center as its origin isintroduced to express the angle between the black pixelpoint on the scale mark and the circle center instead of theimage coordinate system As is shown in Figure 10 the anglesbetween the black pixel point and the circle center are in the0∘ndash360∘ range
After calculating the number for angle 120572 the angle witha large number (connecting lines) in a small angle range isthat of the thick scale mark (main scale) on the dialThe stepsto detect the scale mark by central projection are detailed asfollows
(1) The black pixel point and the image central pointthat is the circle center (119909
0 1199100) in the new image are linked
to determine the slope 119896 which is converted to an angleaccording to the following formula
120572 = arctan (119896) lowast 180
120587 (3)
(2) As shown in Figure 10 the image coordinate systemis transformed to a rectangular one The angle 120572 followingformula (4) is converted into an angle in the rectangularcoordinate
120579 = minus120572 119894 lt 119903 119895 gt 119903 the first quadrant
120579 = 180 minus 120572 119894 lt 119903 119895 lt 119903 the second quadrant
120579 = 180 minus 120572 119894 gt 119903 119895 lt 119903 the third quadrant
120579 = 360 minus 120572 119894 gt 119903 119895 gt 119903 the forth quadrant
(4)
Occ
urre
nce t
imes
Angle (deg)1205791 1205792 1205793 12057940
Figure 11 The angle histogram
(3) An angle histogram is generated by recording theoccurrence frequency of each angle Figure 11 shows the anglehistogram
(4) The angle with the most occurrence times in thehistogram is selected and denoted by max all the angleswhose occurrence times are more thanmax 2 are foundTheoccurrence times of the angles are denoted by 119891(120579) Here 120579represents the angles that occur more than max 2
(5) For every angle 120579 we calculate the weighted averagevalue anglemean(119898) according the following to formula
anglemean (119898)
=119891 (120579 + 1) lowast (120579 + 1) + 119891 (120579) lowast (120579) + 119891 (120579 minus 1) lowast (120579 minus 1)
119891 (120579 + 1) + 119891 (120579) + 119891 (120579 minus 1)
(5)
Here119898 is the sign of the angle
10 Mathematical Problems in Engineering
12057221205721
1205720
1205723
Thick scale marksy
x
Figure 12 The location of the thick scale marks
(6) Calculate specific value 119875 according to the followingformula
119875 =anglemean (119898)
anglemean (119898 + 1) (6)
If 119875 is within the 09ndash11 range anglemean(119898) is the angleof the thick scale mark All the angles sorted out successivelyare taken as 120572
0 1205721 1205722 which correspond to the thick
scale marks Therefore the thick scale marks are located onFigure 12
414 Improved Hough Transform-Based Pointer DetectionThe edge detection is implemented in the pointer gaugeimage using the Canny operator to obtain the edge imageof the gauge Then the Hough transform method is adoptedto fit the straight line in the image to obtain all the straightlines among which the pointer border is included Finallyonly the pointer border can be determined by excluding theother straight edges
(1) The Hough transform method is adopted to fit thestraight line and find all the straight lines in the image Inaddition each straight line described in the image coordinatesystem by 119910 = 119896119909 + 119887 is calculated
(2)The distance 119889 from the circle center (1199090 1199100) to the
straight line is calculated according to the following formula
119889 =
10038161003816100381610038161199100 minus 119896 lowast 1199090 minus 1198871003816100381610038161003816
radic1 + 1198962 (7)
Based on prior knowledge we know that the distancefrom the circle center to the straight line of the pointer ismuch less than the diameter of the dial all the straight lines
Figure 13 Schematic of pointer fitting and its direction
conforming to 119889 lt 211990310 are picked out Here 119903 is thediameter of the dial
(3) The angle of each straight line in (1) is calculatedaccording to formula (8) and stored into an array angle[]
120593 =arctan (119896) lowast 180
120587 (8)
(4) As the angles of the pointer direction are acute onesgenerally in the range of 2∘ sim 10
∘ the angle Δ120593 is calculatedsubject to the following formula
Δ120593 =1003816100381610038161003816angle [119899] minus angle [119899 + 1]1003816100381610038161003816 (9)
If it is in the situation where 2 lt Δ120593 lt 10 the straight lineshould be selected as one side of the pointer Therefore bothlines where the two sides of the pointer lie can be found
(5) According to the straight lines of the two sides thepointer direction can be determined
42 The Pointer Gauge Indicating Value Recognition As isshown in Figure 11 the intersection point 119866 of the twostraight lines of the pointer border is calculated and thenlinked to the circle center 119874(119909
0 1199100) The angle of the
line 119874119866 is calculated In accordance with the position ofthe intersection point 119866 the angle 120573 in the rectangularcoordinate system is computed according to formula (1)(Figure 13)
According to the angle 120573 the readings 1198630 1198631 119863
119899
corresponding to each thick scale mark are obtained Basedon the angle that corresponds to the thick scale markand its number 119899 which is detected in Section 414 thepointer reading can be determined according to formula (10)
Mathematical Problems in Engineering 11
Table 1 Experimental samples
Sample name Image resolution Sample character The purpose of sample selectionThe pressure gauge with manylabels and a lot of writing on thedial
1000750A lot of writing on the dial which mayinfluence the location of the dial centerand the pointer segmentation
Verify the robustness of thealgorithm and the capability ofresisting interference
Flow gauge with the unevenscale 500500 Uneven distribution of the scale marks Verify the universality of the
algorithm
The scale within the adjacent thick scale marks is divided bylinear interpolation
119878indicating value
=
10038161003816100381610038161003816100381610038161003816
1205720minus 120573
1205720minus 1205721
10038161003816100381610038161003816100381610038161003816
(1198631minus 1198630) 120572
0lt 120573 lt 120572
1
10038161003816100381610038161003816100381610038161003816
1205721minus 120573
1205721minus 1205722
10038161003816100381610038161003816100381610038161003816
(1198632minus 1198631) + 1198631 120572
1lt 120573 le 120572
2
10038161003816100381610038161003816100381610038161003816
120572119899minus1
minus 120573
120572119899minus1
minus 120572119899
10038161003816100381610038161003816100381610038161003816
(119863119899minus 119863119899minus1
) + 119863119899minus1
120572119899minus1
lt 120573 le 120572119899
(10)
Here 1205720 1205721 1205722 120572
119899are the angles that correspond to the
thick scale marks
5 Experimental Results and Analysis
To verify the effectiveness of the proposed approach a seriesof experiments are conducted to test its performance
51 Experiment Design The experiments include two partsFirst the images of the different kinds of pointer gauges areselected to conduct the simulation experiments and verifythe effectiveness and applicability of the approach Thenthe pressure gauge calibration system which is composedof a float gauge high precise pressure gauge test pressuregauge camera and computer is established to verify theactual performance of this approach in the pointer gaugecalibration
52 Part One of the Experiments Reading Identification ofthe Various Pointer Gauges
521 Samples Selection As can be seen in Table 1 two groupsof representative samples are selected for the experimentThey include the pressure gauge with a lot of writing on thedial and the flow gauge with an uneven scale
522 Experimental Results and Analysis In this part ofthe experiment we adopt the proposed approach to theautomatic recognition of the indicating value of the abovementioned two types of gauges To verify the effectivenessof the proposed approach Figures 14 and 15 illustrate theexperimental results of each step
Figure 14(a) is the original pointer gauge image FromFigure 14(a) we can see that the pointer and scale marks
cannot be directly detected by segmentation methods sincethere is a lot of writing on the dial Therefore the proposedapproach applies an idea that conducts a coarse-to-finesearch The dial and scale regions are located first since theyare more easily detected than the pointer and scale marks Asshown in Figures 14(c) and 14(d) the dial region is locatedusing the adaptive-region-growing method and the scaleregion is detected under the polar coordinate system Thenin both regions the pointer and scale marks are accuratelydetected As shown in Figures 14(e) and 14(f) the improvedHough transform method is adopted to detect the pointerand the improved central projectionmethod is used to detectthe scale marks The location of the thick scale marks isdetermined according to the angle histogram generated fromthe central projection of the scale marks Their indicatingvalues are given according to the location of the thick scalemarks The scale values within any two adjacent thick marksare determined by linear division According to the directionof the pointer and the location of the scale marks the anglemethod which is adopted in the adjacent thick scale marksis used to indicate the pointer gauge value
The proposed approach detects the pointer direction andthe scale marks location respectively In Figures 15(e) and15(f) it can be seen that the location of the scale marks isprecisely obtained by the angle histogram and the pointer isdetected in the dial region Based on the location of the scalemarks and the pointer direction the distance between thepointer and the adjacent scale marks can be calculated usingthe distance method then the pointer indicating value canbe estimatedTherefore whether or not the distribution of thescale is even the proposed approach is able to precisely detectthe gaugersquos indicating value while the most current existingmethods cannot
53 Part Two of the Experiments The Recognition of the FloatPointer Gauge Indicating Value
531 The System and Procedure of Experiments To verifythe performance of the practical application of the proposedapproach in practical application the float gauge is used as thepressure gauge verification system The type of float gauge isY-047 The main technical parameters of the float gauge aredescribed as follows The range of output pressure is 001ndash025MPa the accuracy class is plusmn005 the rated pressureis 05MPa and the standard weights are from 001MPato 25MPa A high-precision gauge is used as the standardpressure gauge whose accuracy class is 15 The test gaugeis a gauge with general accuracy A visual detection systemis established in front of the float gauge An optical lens
12 Mathematical Problems in Engineering
(a) The original gauge image (b) The dial region of the manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 1205723 12057241205720 1205725 1205726 1205727 1205728 1205729 12057210
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 14 The detection results of the pointer gauge with a lot of writing on the dial
whose focal length can be adjustable from 10 to 20mm anda 500-megapixel HD CMOS camera with a 60 fps framerate are applied as image sensors The system is equippedwith a coaxial light source which is placed on the cameralens The hardware configuration of the computer used inthis experiment is Inter Pentium(R) Dual CPU E2200 and
22GHz The programming environment is VS2010 Theverification system of the pointer gauge that is based oncomputer vision is shown in Figure 16 The camera used inthe experiments takes images of the high-precision standardgauge and the test gauge The proposed approach is adoptedto detect the indicating value of both standard and test gauges
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
4 Mathematical Problems in Engineering
conducted when there is some interference or where thereis some complex writing on the gauge dial (2) In manystudies the scale marks are not precisely detected Hence theindicating value of the pointer gauge with an uneven scalecannot be recognized
This study aims to overcome the shortcomings of theexisting methods As its contextual application it also aimsto improve the calibration of the pressure gauge and developan automatic recognition method of pointer gauge readingsbased on machine vision This method implements a schemethat takes detection from coarse to fine in terms of thepointer and scale marks detection First the pointer gaugersquosdial region is located in a global image Then the dialrsquosannular scale region is located In the following steps thescale marks are detected in the annular scale region and thepointer is detected in the dial region Finally based on theresults of the detection of the pointer and scale marks theimproved distance method is used to recognize the gaugersquosvalue Since the improved distance method is implementedfrom the coarse region to the fine target it has a strongability for suppressing interference Since the indicating valueis obtained from the pointer and scale marks detection themethod is also suitable for reading the gauge whose scalemarks are unevenly distributed
31 Machine Vision Based Pressure Gauge Verification SystemMetrological departments usually calibrate pressure gaugeswith piston manometers In this case the standard pressureoutput of piston manometers given by manually addedweights is passed on to the tested pressure gauge Then thequalification of the pressure gauge is determined throughreading analysis and comparison of the indicating value Ingeneral verification personnel manually handle this processthrough observation by recording and analyzing the indicat-ing value of the gauge panel Many problems still exist withthis verification method They include (1) low accuracy andrepeatability in human visual observations and (2) poor workefficiency due to the large workload of verification personnelHence development of an automatic verification methodfor the pressure pointer gauge will boost the advancementof the technology used in their production and verificationcapabilities The system composition for pointer pressuregauge verification is shown in Figure 3 We can see thatimages of pointer gauges are obtained through a camera andprocessed and analyzed by computer wherein the indicatingvalues of the standard pressure gauge and the test gaugeare detected and recognized The final results of the gaugereading will be displayed on amonitor to operators and reacha verification conclusion based on the analyses of the abovereadingsThe detection software also has suchmultifunctionsas the storage of historical data graphical display and imageswitching to realize the automatic verification of pointergauges
32 Introduction of OurWork In this study a new automaticrecognition method of the pointer gauge that indicates valueis proposed Figure 4 shows a block diagram of this method
In general the computer vision based automatic recog-nition method of indicating value comprises two parts One
Piston gaugeCalibrated gauge
Standard gauge
Display PC
Camera
Image capture card
Figure 3 Schematic of a machine vision based pressure gaugeverification system
part is the detection of the indicating value of the pointergauge This mainly includes the scale detection and thepointer detection which refer to the location of the dialregion and the dial center the determination of the annularscale region in the dial the scale detection and the pointerdetection The other is the automatic recognition of thereading Namely it makes the final judgment of the gaugereading according to the detected direction of the pointer andthe distribution of the scale marks As seen in Figure 4 theautomatic recognition of the indicating value proposed inthis paper falls into five steps
(1) The dial region and its center location first in theentire image the dial region can be determined bysemi-interaction (using the mouse to drop downthe box in the image) The pixel point with thebiggest gray value is searched out in the dial regionand taken as the seed point of the region growingmethod used to produce the circular region in thedial In the process of the region growing methodthe similarity threshold value between the pixel pointsis automatically determined Finally according to theresults of the region growing method the radius andcenter of the circular dial are obtained
(2) Determination of the scale region in the dial theimage coordinate is first converted into the polarcoordinate with the dial center as its origin whichis used to express the coordinate of each pixel in theimage Then by comparing the pixel sum of the lowgray value with that of the high gray value in theannular region the annular scale region in the dial isdetermined And finally it is copied into a new imageas the annular scale region
(3) The scale marks are detected using the central pro-jection method In the newly generated annular scaleregion image first the scale mark segmentation isconducted by theOSTUmethod to form the binariza-tion scale imageThen the angle of each line that linksthe black pixel with the circle center is calculatedand the number of each angle is obtained The angleswith large numbers (connecting line) in a narrowangle range correspond to the thick scalemarks (mainscales) on the dial Thus the angles of the scale marksin the scope of 0∘ndash360∘ are obtained
Mathematical Problems in Engineering 5
Scale detection by central projection method
Segment the image in annular scaleregion by OSTU to obtain the
binarization scale image
Traverse the black pixel points in theimage link each black point to the
circle center and then calculate the angleof the connecting line to get that of
each coarse scale mark
Detection of gauge pointer by Houghtransform
Utilize the Canny operator to performborder detection on the dial image
Use the adaptive threshold value toperform the Hough transform on dial
border image and fit the gauge pointercontour
Calculate the pointer direction within
The indicating value reading
Calculate the indicating valueaccording to the pointer direction and
the dial scale distribution
Dial region and center location
Search the pixel point with maximumgray in the dial region as the seed
Conduct region growing from the seedpoint and adaptively determine thesimilarity principle between the pixels
Obtain the center and radius ofcircular dial region by region growing
Determination of annular scale region in the dial
Convert the image into the polarcoordinates with the dial center as its
origin center
Determine the annular scale region bycomparison between the pixel sum oflow gray value and that of high gray
value
Copy the circular scale regionestablished into a new image
the scope of 0ndash Obtain the scale in the scope of 0ndash360degrees
360 degrees
Figure 4 The flow chart of the pointer gauge indicating value recognition
(4) Pointer detection first the Canny operator is usedto implement the border detection Then the Houghtransform method is used to fit the detected edgeand obtain the pointer contour Finally the pointerdirection is obtained according to the center positionand the pointer contour
(5) The indicating value is calculated according to thepointer direction and the dial scale distribution
4 Indicating Value Recognition Approach ofPointer Gauge
41 Pointer Gauge Indicating Value Detection As mentionedabove the detection of the indicating value is achievedthrough the scale and pointer detection As for the scaledetection first the gauge center and dial region should belocated followed by the extraction of the annular scale region
6 Mathematical Problems in Engineering
(a) Typical pointer gauge (b) The results of the dial region growing method
Figure 5 The diagram of the dial region growing results
and the scale mark detection Finally the detection of thepointer direction is performed
411 Dial Region and Its Center Location As for the detectionof the indicating value first the gauge pointer and its directionshould be detected to determine its rotation center In generalthe gaugersquos dial center is the pointerrsquos rotation center Hencefirst the dial center should be detected In this study thecircular dial region is detected for determining the dial centerAs is shown in Figure 5(a) typically there are the labelswords and scale marks in the circular dial Therefore it isdifficult to confirm the circular region of the dial by suchsegment methods as the border and threshold values Inview of the above this paper introduces the adaptive-region-growing method to locate the gaugersquos circular region andfurther to obtain its dial center The specific steps are asfollows
(1) The placement of the seed [24] since the dial isusually white the region with a high gray level orthe pixel point with the highest gray level in thegauge is certainly located at the part where the dialis Therefore the latter is selected as the seed point ofthe region growing
(2) Subject to the similarity of the pixel gray level theexpansion of the dial region is conducted from theseed point However it is hard to fix an appropriatesimilarity threshold value of the gray level betweenpixels As is shown in Figure 5(b) the circular dialregion cannot be completely produced with a smallthreshold value while an oversized dial region will begenerated with a large threshold value
This study proposes an adaptive method to determinethe similarity threshold value of the gray level the concreteprocess is as follows
(A) A small similarity threshold value 119879 of the graylevel is empirically chosen and then the dial regiongrowing method is conducted from the seed point
(B) The ratio of the circumscribed rectanglersquos length andwidth of the newly generated dial region is calculatedto detect whether it is in the 095ndash105 range
(C) If the ratio of the length and width fails to meetthe conditions in (B) the similarity threshold valuewould be increased with a certain step length (119878)and the first step (A) would continue to conductthe region growing from the seed point Howeverif it meets the conditions in (B) the growing regionwill be the circular dial region the center of thecircumscribed rectangle will be the circle center andthe smaller value between the length and width willbe the diameter of the circular region (Figure 6)
412 Polar Coordinate Based Scale Region Location
(1) Image in Polar Coordinates In general in light of thecircular permutation the pointer gauge scale marks aredistributed around the center Therefore to correctly andexpediently show where the scale marks in the image arethe polar coordinate system is utilized to position the pixelsin the image In this study a conversion from the imagecoordinates into the polar coordinates is performed Thismeans that the pixel coordinates expressed by the rectangularimage coordinate system are converted into those expressedby polar coordinates The polar coordinates defined in thisstudy are shown in Figure 7 the circle center being its origin
(2) Detection of the Scale Region in the Dial The procedurefor scale region detection based on the adaptive thresholdmethod is as follows
(A) The circular dial radius detected in Section 411 istaken as the initial radius 119877 its decreasing step lengthbeing 119887 and the initial 119899 = 1 In the gauge dial thegray value difference between the scale marks anddial background is generally rather large Thereforethe adaptive threshold value method can be adoptedto segment the scale In the polar coordinate theannular binarization image is gained by employing
Mathematical Problems in Engineering 7
(a) (b)
(c) (d)
Figure 6 The region growing results under different pixel similarity threshold values
90∘
60∘
210∘
150∘
120∘
180∘
240∘
270∘300∘
330∘
30∘
0∘
Figure 7 Schematic of the polar coordinate
the OSTU method [25] to perform the binarizationsegmentation in the annular region
(B) The ratio between the number of black andwhite pixelpoints in the annular binarization region between theradius 119877 and 119877minus119899lowast119887 is calculated and denoted by 119903
1
After reducing the inner radius according to the steplength 119887 the ratio between the number of black and
white pixel points in the region between 119877minus(119899+1)lowast119887and 119877 is computed and denoted by 119903
2
(C) If 1199031is obviously different from 119903
2 the inner radius
and the outer radius of the scale region would be 119877 minus(119899 + 1) lowast 119887 and 119877 respectively
(D) If 1199031is approximated to 119903
2 the annulus between 119877 minus
(119899 + 1) lowast 119887 and 119877 would still be in the circular scaleregion At thismoment let 119899 = 119899+1 the ratio betweenthe number of black and white pixel points in theannulus between 119877 minus (119899 + 1) lowast 119887 and 119877 is computedand is still denoted by 119903
2 Then go to (B) (Figure 8)
After obtaining the annular scale region in the abovepolar coordinate the scale region is selected on the basis ofangle and radius and copied into a new image called the scalemark region image119872 (Figure 9)
413 Scale Mark Detection Based on the Improved CentralProjection Method In the newly generated image119872 whichis the annular scale region the OSTU method is adopted tosegment the scale marks to form the binarized scale imageIn this image the pixel value of the scale mark is 0 (black)and those in other places are 255 (white) Each black pixelpoint and the circle center are linked and the slope 119870 ofthe connecting line is calculated according to the formula119896 = tan120572 wherein 120572 is the angle between the 119909-axis and
8 Mathematical Problems in Engineering
(a) (b)
(c)
Figure 8 Schematic of scale mark region detection
(a) (b)
Figure 9 The scale mark region diagram and its binarized image
Mathematical Problems in Engineering 9
x
y
0
(a) Schematic of the image coordinate system
x
y
1
1
2
2
3
3
4
4
minus1
minus1
minus2
minus2
0minus3
minus3
minus4
minus4
(b) The rectangular coordinate system
Figure 10 Schematic of the image coordinate system and the rectangular coordinate system
the connecting line between the black pixel point and thecircular center Due to the origin of the image coordinatebeing in the top left corner and the119910-axis running downwardit is inconvenient to denote the central angle within thescope of 0∘ndash360∘ Therefore in this study the rectangularcoordinate system with the image center as its origin isintroduced to express the angle between the black pixelpoint on the scale mark and the circle center instead of theimage coordinate system As is shown in Figure 10 the anglesbetween the black pixel point and the circle center are in the0∘ndash360∘ range
After calculating the number for angle 120572 the angle witha large number (connecting lines) in a small angle range isthat of the thick scale mark (main scale) on the dialThe stepsto detect the scale mark by central projection are detailed asfollows
(1) The black pixel point and the image central pointthat is the circle center (119909
0 1199100) in the new image are linked
to determine the slope 119896 which is converted to an angleaccording to the following formula
120572 = arctan (119896) lowast 180
120587 (3)
(2) As shown in Figure 10 the image coordinate systemis transformed to a rectangular one The angle 120572 followingformula (4) is converted into an angle in the rectangularcoordinate
120579 = minus120572 119894 lt 119903 119895 gt 119903 the first quadrant
120579 = 180 minus 120572 119894 lt 119903 119895 lt 119903 the second quadrant
120579 = 180 minus 120572 119894 gt 119903 119895 lt 119903 the third quadrant
120579 = 360 minus 120572 119894 gt 119903 119895 gt 119903 the forth quadrant
(4)
Occ
urre
nce t
imes
Angle (deg)1205791 1205792 1205793 12057940
Figure 11 The angle histogram
(3) An angle histogram is generated by recording theoccurrence frequency of each angle Figure 11 shows the anglehistogram
(4) The angle with the most occurrence times in thehistogram is selected and denoted by max all the angleswhose occurrence times are more thanmax 2 are foundTheoccurrence times of the angles are denoted by 119891(120579) Here 120579represents the angles that occur more than max 2
(5) For every angle 120579 we calculate the weighted averagevalue anglemean(119898) according the following to formula
anglemean (119898)
=119891 (120579 + 1) lowast (120579 + 1) + 119891 (120579) lowast (120579) + 119891 (120579 minus 1) lowast (120579 minus 1)
119891 (120579 + 1) + 119891 (120579) + 119891 (120579 minus 1)
(5)
Here119898 is the sign of the angle
10 Mathematical Problems in Engineering
12057221205721
1205720
1205723
Thick scale marksy
x
Figure 12 The location of the thick scale marks
(6) Calculate specific value 119875 according to the followingformula
119875 =anglemean (119898)
anglemean (119898 + 1) (6)
If 119875 is within the 09ndash11 range anglemean(119898) is the angleof the thick scale mark All the angles sorted out successivelyare taken as 120572
0 1205721 1205722 which correspond to the thick
scale marks Therefore the thick scale marks are located onFigure 12
414 Improved Hough Transform-Based Pointer DetectionThe edge detection is implemented in the pointer gaugeimage using the Canny operator to obtain the edge imageof the gauge Then the Hough transform method is adoptedto fit the straight line in the image to obtain all the straightlines among which the pointer border is included Finallyonly the pointer border can be determined by excluding theother straight edges
(1) The Hough transform method is adopted to fit thestraight line and find all the straight lines in the image Inaddition each straight line described in the image coordinatesystem by 119910 = 119896119909 + 119887 is calculated
(2)The distance 119889 from the circle center (1199090 1199100) to the
straight line is calculated according to the following formula
119889 =
10038161003816100381610038161199100 minus 119896 lowast 1199090 minus 1198871003816100381610038161003816
radic1 + 1198962 (7)
Based on prior knowledge we know that the distancefrom the circle center to the straight line of the pointer ismuch less than the diameter of the dial all the straight lines
Figure 13 Schematic of pointer fitting and its direction
conforming to 119889 lt 211990310 are picked out Here 119903 is thediameter of the dial
(3) The angle of each straight line in (1) is calculatedaccording to formula (8) and stored into an array angle[]
120593 =arctan (119896) lowast 180
120587 (8)
(4) As the angles of the pointer direction are acute onesgenerally in the range of 2∘ sim 10
∘ the angle Δ120593 is calculatedsubject to the following formula
Δ120593 =1003816100381610038161003816angle [119899] minus angle [119899 + 1]1003816100381610038161003816 (9)
If it is in the situation where 2 lt Δ120593 lt 10 the straight lineshould be selected as one side of the pointer Therefore bothlines where the two sides of the pointer lie can be found
(5) According to the straight lines of the two sides thepointer direction can be determined
42 The Pointer Gauge Indicating Value Recognition As isshown in Figure 11 the intersection point 119866 of the twostraight lines of the pointer border is calculated and thenlinked to the circle center 119874(119909
0 1199100) The angle of the
line 119874119866 is calculated In accordance with the position ofthe intersection point 119866 the angle 120573 in the rectangularcoordinate system is computed according to formula (1)(Figure 13)
According to the angle 120573 the readings 1198630 1198631 119863
119899
corresponding to each thick scale mark are obtained Basedon the angle that corresponds to the thick scale markand its number 119899 which is detected in Section 414 thepointer reading can be determined according to formula (10)
Mathematical Problems in Engineering 11
Table 1 Experimental samples
Sample name Image resolution Sample character The purpose of sample selectionThe pressure gauge with manylabels and a lot of writing on thedial
1000750A lot of writing on the dial which mayinfluence the location of the dial centerand the pointer segmentation
Verify the robustness of thealgorithm and the capability ofresisting interference
Flow gauge with the unevenscale 500500 Uneven distribution of the scale marks Verify the universality of the
algorithm
The scale within the adjacent thick scale marks is divided bylinear interpolation
119878indicating value
=
10038161003816100381610038161003816100381610038161003816
1205720minus 120573
1205720minus 1205721
10038161003816100381610038161003816100381610038161003816
(1198631minus 1198630) 120572
0lt 120573 lt 120572
1
10038161003816100381610038161003816100381610038161003816
1205721minus 120573
1205721minus 1205722
10038161003816100381610038161003816100381610038161003816
(1198632minus 1198631) + 1198631 120572
1lt 120573 le 120572
2
10038161003816100381610038161003816100381610038161003816
120572119899minus1
minus 120573
120572119899minus1
minus 120572119899
10038161003816100381610038161003816100381610038161003816
(119863119899minus 119863119899minus1
) + 119863119899minus1
120572119899minus1
lt 120573 le 120572119899
(10)
Here 1205720 1205721 1205722 120572
119899are the angles that correspond to the
thick scale marks
5 Experimental Results and Analysis
To verify the effectiveness of the proposed approach a seriesof experiments are conducted to test its performance
51 Experiment Design The experiments include two partsFirst the images of the different kinds of pointer gauges areselected to conduct the simulation experiments and verifythe effectiveness and applicability of the approach Thenthe pressure gauge calibration system which is composedof a float gauge high precise pressure gauge test pressuregauge camera and computer is established to verify theactual performance of this approach in the pointer gaugecalibration
52 Part One of the Experiments Reading Identification ofthe Various Pointer Gauges
521 Samples Selection As can be seen in Table 1 two groupsof representative samples are selected for the experimentThey include the pressure gauge with a lot of writing on thedial and the flow gauge with an uneven scale
522 Experimental Results and Analysis In this part ofthe experiment we adopt the proposed approach to theautomatic recognition of the indicating value of the abovementioned two types of gauges To verify the effectivenessof the proposed approach Figures 14 and 15 illustrate theexperimental results of each step
Figure 14(a) is the original pointer gauge image FromFigure 14(a) we can see that the pointer and scale marks
cannot be directly detected by segmentation methods sincethere is a lot of writing on the dial Therefore the proposedapproach applies an idea that conducts a coarse-to-finesearch The dial and scale regions are located first since theyare more easily detected than the pointer and scale marks Asshown in Figures 14(c) and 14(d) the dial region is locatedusing the adaptive-region-growing method and the scaleregion is detected under the polar coordinate system Thenin both regions the pointer and scale marks are accuratelydetected As shown in Figures 14(e) and 14(f) the improvedHough transform method is adopted to detect the pointerand the improved central projectionmethod is used to detectthe scale marks The location of the thick scale marks isdetermined according to the angle histogram generated fromthe central projection of the scale marks Their indicatingvalues are given according to the location of the thick scalemarks The scale values within any two adjacent thick marksare determined by linear division According to the directionof the pointer and the location of the scale marks the anglemethod which is adopted in the adjacent thick scale marksis used to indicate the pointer gauge value
The proposed approach detects the pointer direction andthe scale marks location respectively In Figures 15(e) and15(f) it can be seen that the location of the scale marks isprecisely obtained by the angle histogram and the pointer isdetected in the dial region Based on the location of the scalemarks and the pointer direction the distance between thepointer and the adjacent scale marks can be calculated usingthe distance method then the pointer indicating value canbe estimatedTherefore whether or not the distribution of thescale is even the proposed approach is able to precisely detectthe gaugersquos indicating value while the most current existingmethods cannot
53 Part Two of the Experiments The Recognition of the FloatPointer Gauge Indicating Value
531 The System and Procedure of Experiments To verifythe performance of the practical application of the proposedapproach in practical application the float gauge is used as thepressure gauge verification system The type of float gauge isY-047 The main technical parameters of the float gauge aredescribed as follows The range of output pressure is 001ndash025MPa the accuracy class is plusmn005 the rated pressureis 05MPa and the standard weights are from 001MPato 25MPa A high-precision gauge is used as the standardpressure gauge whose accuracy class is 15 The test gaugeis a gauge with general accuracy A visual detection systemis established in front of the float gauge An optical lens
12 Mathematical Problems in Engineering
(a) The original gauge image (b) The dial region of the manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 1205723 12057241205720 1205725 1205726 1205727 1205728 1205729 12057210
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 14 The detection results of the pointer gauge with a lot of writing on the dial
whose focal length can be adjustable from 10 to 20mm anda 500-megapixel HD CMOS camera with a 60 fps framerate are applied as image sensors The system is equippedwith a coaxial light source which is placed on the cameralens The hardware configuration of the computer used inthis experiment is Inter Pentium(R) Dual CPU E2200 and
22GHz The programming environment is VS2010 Theverification system of the pointer gauge that is based oncomputer vision is shown in Figure 16 The camera used inthe experiments takes images of the high-precision standardgauge and the test gauge The proposed approach is adoptedto detect the indicating value of both standard and test gauges
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 5
Scale detection by central projection method
Segment the image in annular scaleregion by OSTU to obtain the
binarization scale image
Traverse the black pixel points in theimage link each black point to the
circle center and then calculate the angleof the connecting line to get that of
each coarse scale mark
Detection of gauge pointer by Houghtransform
Utilize the Canny operator to performborder detection on the dial image
Use the adaptive threshold value toperform the Hough transform on dial
border image and fit the gauge pointercontour
Calculate the pointer direction within
The indicating value reading
Calculate the indicating valueaccording to the pointer direction and
the dial scale distribution
Dial region and center location
Search the pixel point with maximumgray in the dial region as the seed
Conduct region growing from the seedpoint and adaptively determine thesimilarity principle between the pixels
Obtain the center and radius ofcircular dial region by region growing
Determination of annular scale region in the dial
Convert the image into the polarcoordinates with the dial center as its
origin center
Determine the annular scale region bycomparison between the pixel sum oflow gray value and that of high gray
value
Copy the circular scale regionestablished into a new image
the scope of 0ndash Obtain the scale in the scope of 0ndash360degrees
360 degrees
Figure 4 The flow chart of the pointer gauge indicating value recognition
(4) Pointer detection first the Canny operator is usedto implement the border detection Then the Houghtransform method is used to fit the detected edgeand obtain the pointer contour Finally the pointerdirection is obtained according to the center positionand the pointer contour
(5) The indicating value is calculated according to thepointer direction and the dial scale distribution
4 Indicating Value Recognition Approach ofPointer Gauge
41 Pointer Gauge Indicating Value Detection As mentionedabove the detection of the indicating value is achievedthrough the scale and pointer detection As for the scaledetection first the gauge center and dial region should belocated followed by the extraction of the annular scale region
6 Mathematical Problems in Engineering
(a) Typical pointer gauge (b) The results of the dial region growing method
Figure 5 The diagram of the dial region growing results
and the scale mark detection Finally the detection of thepointer direction is performed
411 Dial Region and Its Center Location As for the detectionof the indicating value first the gauge pointer and its directionshould be detected to determine its rotation center In generalthe gaugersquos dial center is the pointerrsquos rotation center Hencefirst the dial center should be detected In this study thecircular dial region is detected for determining the dial centerAs is shown in Figure 5(a) typically there are the labelswords and scale marks in the circular dial Therefore it isdifficult to confirm the circular region of the dial by suchsegment methods as the border and threshold values Inview of the above this paper introduces the adaptive-region-growing method to locate the gaugersquos circular region andfurther to obtain its dial center The specific steps are asfollows
(1) The placement of the seed [24] since the dial isusually white the region with a high gray level orthe pixel point with the highest gray level in thegauge is certainly located at the part where the dialis Therefore the latter is selected as the seed point ofthe region growing
(2) Subject to the similarity of the pixel gray level theexpansion of the dial region is conducted from theseed point However it is hard to fix an appropriatesimilarity threshold value of the gray level betweenpixels As is shown in Figure 5(b) the circular dialregion cannot be completely produced with a smallthreshold value while an oversized dial region will begenerated with a large threshold value
This study proposes an adaptive method to determinethe similarity threshold value of the gray level the concreteprocess is as follows
(A) A small similarity threshold value 119879 of the graylevel is empirically chosen and then the dial regiongrowing method is conducted from the seed point
(B) The ratio of the circumscribed rectanglersquos length andwidth of the newly generated dial region is calculatedto detect whether it is in the 095ndash105 range
(C) If the ratio of the length and width fails to meetthe conditions in (B) the similarity threshold valuewould be increased with a certain step length (119878)and the first step (A) would continue to conductthe region growing from the seed point Howeverif it meets the conditions in (B) the growing regionwill be the circular dial region the center of thecircumscribed rectangle will be the circle center andthe smaller value between the length and width willbe the diameter of the circular region (Figure 6)
412 Polar Coordinate Based Scale Region Location
(1) Image in Polar Coordinates In general in light of thecircular permutation the pointer gauge scale marks aredistributed around the center Therefore to correctly andexpediently show where the scale marks in the image arethe polar coordinate system is utilized to position the pixelsin the image In this study a conversion from the imagecoordinates into the polar coordinates is performed Thismeans that the pixel coordinates expressed by the rectangularimage coordinate system are converted into those expressedby polar coordinates The polar coordinates defined in thisstudy are shown in Figure 7 the circle center being its origin
(2) Detection of the Scale Region in the Dial The procedurefor scale region detection based on the adaptive thresholdmethod is as follows
(A) The circular dial radius detected in Section 411 istaken as the initial radius 119877 its decreasing step lengthbeing 119887 and the initial 119899 = 1 In the gauge dial thegray value difference between the scale marks anddial background is generally rather large Thereforethe adaptive threshold value method can be adoptedto segment the scale In the polar coordinate theannular binarization image is gained by employing
Mathematical Problems in Engineering 7
(a) (b)
(c) (d)
Figure 6 The region growing results under different pixel similarity threshold values
90∘
60∘
210∘
150∘
120∘
180∘
240∘
270∘300∘
330∘
30∘
0∘
Figure 7 Schematic of the polar coordinate
the OSTU method [25] to perform the binarizationsegmentation in the annular region
(B) The ratio between the number of black andwhite pixelpoints in the annular binarization region between theradius 119877 and 119877minus119899lowast119887 is calculated and denoted by 119903
1
After reducing the inner radius according to the steplength 119887 the ratio between the number of black and
white pixel points in the region between 119877minus(119899+1)lowast119887and 119877 is computed and denoted by 119903
2
(C) If 1199031is obviously different from 119903
2 the inner radius
and the outer radius of the scale region would be 119877 minus(119899 + 1) lowast 119887 and 119877 respectively
(D) If 1199031is approximated to 119903
2 the annulus between 119877 minus
(119899 + 1) lowast 119887 and 119877 would still be in the circular scaleregion At thismoment let 119899 = 119899+1 the ratio betweenthe number of black and white pixel points in theannulus between 119877 minus (119899 + 1) lowast 119887 and 119877 is computedand is still denoted by 119903
2 Then go to (B) (Figure 8)
After obtaining the annular scale region in the abovepolar coordinate the scale region is selected on the basis ofangle and radius and copied into a new image called the scalemark region image119872 (Figure 9)
413 Scale Mark Detection Based on the Improved CentralProjection Method In the newly generated image119872 whichis the annular scale region the OSTU method is adopted tosegment the scale marks to form the binarized scale imageIn this image the pixel value of the scale mark is 0 (black)and those in other places are 255 (white) Each black pixelpoint and the circle center are linked and the slope 119870 ofthe connecting line is calculated according to the formula119896 = tan120572 wherein 120572 is the angle between the 119909-axis and
8 Mathematical Problems in Engineering
(a) (b)
(c)
Figure 8 Schematic of scale mark region detection
(a) (b)
Figure 9 The scale mark region diagram and its binarized image
Mathematical Problems in Engineering 9
x
y
0
(a) Schematic of the image coordinate system
x
y
1
1
2
2
3
3
4
4
minus1
minus1
minus2
minus2
0minus3
minus3
minus4
minus4
(b) The rectangular coordinate system
Figure 10 Schematic of the image coordinate system and the rectangular coordinate system
the connecting line between the black pixel point and thecircular center Due to the origin of the image coordinatebeing in the top left corner and the119910-axis running downwardit is inconvenient to denote the central angle within thescope of 0∘ndash360∘ Therefore in this study the rectangularcoordinate system with the image center as its origin isintroduced to express the angle between the black pixelpoint on the scale mark and the circle center instead of theimage coordinate system As is shown in Figure 10 the anglesbetween the black pixel point and the circle center are in the0∘ndash360∘ range
After calculating the number for angle 120572 the angle witha large number (connecting lines) in a small angle range isthat of the thick scale mark (main scale) on the dialThe stepsto detect the scale mark by central projection are detailed asfollows
(1) The black pixel point and the image central pointthat is the circle center (119909
0 1199100) in the new image are linked
to determine the slope 119896 which is converted to an angleaccording to the following formula
120572 = arctan (119896) lowast 180
120587 (3)
(2) As shown in Figure 10 the image coordinate systemis transformed to a rectangular one The angle 120572 followingformula (4) is converted into an angle in the rectangularcoordinate
120579 = minus120572 119894 lt 119903 119895 gt 119903 the first quadrant
120579 = 180 minus 120572 119894 lt 119903 119895 lt 119903 the second quadrant
120579 = 180 minus 120572 119894 gt 119903 119895 lt 119903 the third quadrant
120579 = 360 minus 120572 119894 gt 119903 119895 gt 119903 the forth quadrant
(4)
Occ
urre
nce t
imes
Angle (deg)1205791 1205792 1205793 12057940
Figure 11 The angle histogram
(3) An angle histogram is generated by recording theoccurrence frequency of each angle Figure 11 shows the anglehistogram
(4) The angle with the most occurrence times in thehistogram is selected and denoted by max all the angleswhose occurrence times are more thanmax 2 are foundTheoccurrence times of the angles are denoted by 119891(120579) Here 120579represents the angles that occur more than max 2
(5) For every angle 120579 we calculate the weighted averagevalue anglemean(119898) according the following to formula
anglemean (119898)
=119891 (120579 + 1) lowast (120579 + 1) + 119891 (120579) lowast (120579) + 119891 (120579 minus 1) lowast (120579 minus 1)
119891 (120579 + 1) + 119891 (120579) + 119891 (120579 minus 1)
(5)
Here119898 is the sign of the angle
10 Mathematical Problems in Engineering
12057221205721
1205720
1205723
Thick scale marksy
x
Figure 12 The location of the thick scale marks
(6) Calculate specific value 119875 according to the followingformula
119875 =anglemean (119898)
anglemean (119898 + 1) (6)
If 119875 is within the 09ndash11 range anglemean(119898) is the angleof the thick scale mark All the angles sorted out successivelyare taken as 120572
0 1205721 1205722 which correspond to the thick
scale marks Therefore the thick scale marks are located onFigure 12
414 Improved Hough Transform-Based Pointer DetectionThe edge detection is implemented in the pointer gaugeimage using the Canny operator to obtain the edge imageof the gauge Then the Hough transform method is adoptedto fit the straight line in the image to obtain all the straightlines among which the pointer border is included Finallyonly the pointer border can be determined by excluding theother straight edges
(1) The Hough transform method is adopted to fit thestraight line and find all the straight lines in the image Inaddition each straight line described in the image coordinatesystem by 119910 = 119896119909 + 119887 is calculated
(2)The distance 119889 from the circle center (1199090 1199100) to the
straight line is calculated according to the following formula
119889 =
10038161003816100381610038161199100 minus 119896 lowast 1199090 minus 1198871003816100381610038161003816
radic1 + 1198962 (7)
Based on prior knowledge we know that the distancefrom the circle center to the straight line of the pointer ismuch less than the diameter of the dial all the straight lines
Figure 13 Schematic of pointer fitting and its direction
conforming to 119889 lt 211990310 are picked out Here 119903 is thediameter of the dial
(3) The angle of each straight line in (1) is calculatedaccording to formula (8) and stored into an array angle[]
120593 =arctan (119896) lowast 180
120587 (8)
(4) As the angles of the pointer direction are acute onesgenerally in the range of 2∘ sim 10
∘ the angle Δ120593 is calculatedsubject to the following formula
Δ120593 =1003816100381610038161003816angle [119899] minus angle [119899 + 1]1003816100381610038161003816 (9)
If it is in the situation where 2 lt Δ120593 lt 10 the straight lineshould be selected as one side of the pointer Therefore bothlines where the two sides of the pointer lie can be found
(5) According to the straight lines of the two sides thepointer direction can be determined
42 The Pointer Gauge Indicating Value Recognition As isshown in Figure 11 the intersection point 119866 of the twostraight lines of the pointer border is calculated and thenlinked to the circle center 119874(119909
0 1199100) The angle of the
line 119874119866 is calculated In accordance with the position ofthe intersection point 119866 the angle 120573 in the rectangularcoordinate system is computed according to formula (1)(Figure 13)
According to the angle 120573 the readings 1198630 1198631 119863
119899
corresponding to each thick scale mark are obtained Basedon the angle that corresponds to the thick scale markand its number 119899 which is detected in Section 414 thepointer reading can be determined according to formula (10)
Mathematical Problems in Engineering 11
Table 1 Experimental samples
Sample name Image resolution Sample character The purpose of sample selectionThe pressure gauge with manylabels and a lot of writing on thedial
1000750A lot of writing on the dial which mayinfluence the location of the dial centerand the pointer segmentation
Verify the robustness of thealgorithm and the capability ofresisting interference
Flow gauge with the unevenscale 500500 Uneven distribution of the scale marks Verify the universality of the
algorithm
The scale within the adjacent thick scale marks is divided bylinear interpolation
119878indicating value
=
10038161003816100381610038161003816100381610038161003816
1205720minus 120573
1205720minus 1205721
10038161003816100381610038161003816100381610038161003816
(1198631minus 1198630) 120572
0lt 120573 lt 120572
1
10038161003816100381610038161003816100381610038161003816
1205721minus 120573
1205721minus 1205722
10038161003816100381610038161003816100381610038161003816
(1198632minus 1198631) + 1198631 120572
1lt 120573 le 120572
2
10038161003816100381610038161003816100381610038161003816
120572119899minus1
minus 120573
120572119899minus1
minus 120572119899
10038161003816100381610038161003816100381610038161003816
(119863119899minus 119863119899minus1
) + 119863119899minus1
120572119899minus1
lt 120573 le 120572119899
(10)
Here 1205720 1205721 1205722 120572
119899are the angles that correspond to the
thick scale marks
5 Experimental Results and Analysis
To verify the effectiveness of the proposed approach a seriesof experiments are conducted to test its performance
51 Experiment Design The experiments include two partsFirst the images of the different kinds of pointer gauges areselected to conduct the simulation experiments and verifythe effectiveness and applicability of the approach Thenthe pressure gauge calibration system which is composedof a float gauge high precise pressure gauge test pressuregauge camera and computer is established to verify theactual performance of this approach in the pointer gaugecalibration
52 Part One of the Experiments Reading Identification ofthe Various Pointer Gauges
521 Samples Selection As can be seen in Table 1 two groupsof representative samples are selected for the experimentThey include the pressure gauge with a lot of writing on thedial and the flow gauge with an uneven scale
522 Experimental Results and Analysis In this part ofthe experiment we adopt the proposed approach to theautomatic recognition of the indicating value of the abovementioned two types of gauges To verify the effectivenessof the proposed approach Figures 14 and 15 illustrate theexperimental results of each step
Figure 14(a) is the original pointer gauge image FromFigure 14(a) we can see that the pointer and scale marks
cannot be directly detected by segmentation methods sincethere is a lot of writing on the dial Therefore the proposedapproach applies an idea that conducts a coarse-to-finesearch The dial and scale regions are located first since theyare more easily detected than the pointer and scale marks Asshown in Figures 14(c) and 14(d) the dial region is locatedusing the adaptive-region-growing method and the scaleregion is detected under the polar coordinate system Thenin both regions the pointer and scale marks are accuratelydetected As shown in Figures 14(e) and 14(f) the improvedHough transform method is adopted to detect the pointerand the improved central projectionmethod is used to detectthe scale marks The location of the thick scale marks isdetermined according to the angle histogram generated fromthe central projection of the scale marks Their indicatingvalues are given according to the location of the thick scalemarks The scale values within any two adjacent thick marksare determined by linear division According to the directionof the pointer and the location of the scale marks the anglemethod which is adopted in the adjacent thick scale marksis used to indicate the pointer gauge value
The proposed approach detects the pointer direction andthe scale marks location respectively In Figures 15(e) and15(f) it can be seen that the location of the scale marks isprecisely obtained by the angle histogram and the pointer isdetected in the dial region Based on the location of the scalemarks and the pointer direction the distance between thepointer and the adjacent scale marks can be calculated usingthe distance method then the pointer indicating value canbe estimatedTherefore whether or not the distribution of thescale is even the proposed approach is able to precisely detectthe gaugersquos indicating value while the most current existingmethods cannot
53 Part Two of the Experiments The Recognition of the FloatPointer Gauge Indicating Value
531 The System and Procedure of Experiments To verifythe performance of the practical application of the proposedapproach in practical application the float gauge is used as thepressure gauge verification system The type of float gauge isY-047 The main technical parameters of the float gauge aredescribed as follows The range of output pressure is 001ndash025MPa the accuracy class is plusmn005 the rated pressureis 05MPa and the standard weights are from 001MPato 25MPa A high-precision gauge is used as the standardpressure gauge whose accuracy class is 15 The test gaugeis a gauge with general accuracy A visual detection systemis established in front of the float gauge An optical lens
12 Mathematical Problems in Engineering
(a) The original gauge image (b) The dial region of the manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 1205723 12057241205720 1205725 1205726 1205727 1205728 1205729 12057210
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 14 The detection results of the pointer gauge with a lot of writing on the dial
whose focal length can be adjustable from 10 to 20mm anda 500-megapixel HD CMOS camera with a 60 fps framerate are applied as image sensors The system is equippedwith a coaxial light source which is placed on the cameralens The hardware configuration of the computer used inthis experiment is Inter Pentium(R) Dual CPU E2200 and
22GHz The programming environment is VS2010 Theverification system of the pointer gauge that is based oncomputer vision is shown in Figure 16 The camera used inthe experiments takes images of the high-precision standardgauge and the test gauge The proposed approach is adoptedto detect the indicating value of both standard and test gauges
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
6 Mathematical Problems in Engineering
(a) Typical pointer gauge (b) The results of the dial region growing method
Figure 5 The diagram of the dial region growing results
and the scale mark detection Finally the detection of thepointer direction is performed
411 Dial Region and Its Center Location As for the detectionof the indicating value first the gauge pointer and its directionshould be detected to determine its rotation center In generalthe gaugersquos dial center is the pointerrsquos rotation center Hencefirst the dial center should be detected In this study thecircular dial region is detected for determining the dial centerAs is shown in Figure 5(a) typically there are the labelswords and scale marks in the circular dial Therefore it isdifficult to confirm the circular region of the dial by suchsegment methods as the border and threshold values Inview of the above this paper introduces the adaptive-region-growing method to locate the gaugersquos circular region andfurther to obtain its dial center The specific steps are asfollows
(1) The placement of the seed [24] since the dial isusually white the region with a high gray level orthe pixel point with the highest gray level in thegauge is certainly located at the part where the dialis Therefore the latter is selected as the seed point ofthe region growing
(2) Subject to the similarity of the pixel gray level theexpansion of the dial region is conducted from theseed point However it is hard to fix an appropriatesimilarity threshold value of the gray level betweenpixels As is shown in Figure 5(b) the circular dialregion cannot be completely produced with a smallthreshold value while an oversized dial region will begenerated with a large threshold value
This study proposes an adaptive method to determinethe similarity threshold value of the gray level the concreteprocess is as follows
(A) A small similarity threshold value 119879 of the graylevel is empirically chosen and then the dial regiongrowing method is conducted from the seed point
(B) The ratio of the circumscribed rectanglersquos length andwidth of the newly generated dial region is calculatedto detect whether it is in the 095ndash105 range
(C) If the ratio of the length and width fails to meetthe conditions in (B) the similarity threshold valuewould be increased with a certain step length (119878)and the first step (A) would continue to conductthe region growing from the seed point Howeverif it meets the conditions in (B) the growing regionwill be the circular dial region the center of thecircumscribed rectangle will be the circle center andthe smaller value between the length and width willbe the diameter of the circular region (Figure 6)
412 Polar Coordinate Based Scale Region Location
(1) Image in Polar Coordinates In general in light of thecircular permutation the pointer gauge scale marks aredistributed around the center Therefore to correctly andexpediently show where the scale marks in the image arethe polar coordinate system is utilized to position the pixelsin the image In this study a conversion from the imagecoordinates into the polar coordinates is performed Thismeans that the pixel coordinates expressed by the rectangularimage coordinate system are converted into those expressedby polar coordinates The polar coordinates defined in thisstudy are shown in Figure 7 the circle center being its origin
(2) Detection of the Scale Region in the Dial The procedurefor scale region detection based on the adaptive thresholdmethod is as follows
(A) The circular dial radius detected in Section 411 istaken as the initial radius 119877 its decreasing step lengthbeing 119887 and the initial 119899 = 1 In the gauge dial thegray value difference between the scale marks anddial background is generally rather large Thereforethe adaptive threshold value method can be adoptedto segment the scale In the polar coordinate theannular binarization image is gained by employing
Mathematical Problems in Engineering 7
(a) (b)
(c) (d)
Figure 6 The region growing results under different pixel similarity threshold values
90∘
60∘
210∘
150∘
120∘
180∘
240∘
270∘300∘
330∘
30∘
0∘
Figure 7 Schematic of the polar coordinate
the OSTU method [25] to perform the binarizationsegmentation in the annular region
(B) The ratio between the number of black andwhite pixelpoints in the annular binarization region between theradius 119877 and 119877minus119899lowast119887 is calculated and denoted by 119903
1
After reducing the inner radius according to the steplength 119887 the ratio between the number of black and
white pixel points in the region between 119877minus(119899+1)lowast119887and 119877 is computed and denoted by 119903
2
(C) If 1199031is obviously different from 119903
2 the inner radius
and the outer radius of the scale region would be 119877 minus(119899 + 1) lowast 119887 and 119877 respectively
(D) If 1199031is approximated to 119903
2 the annulus between 119877 minus
(119899 + 1) lowast 119887 and 119877 would still be in the circular scaleregion At thismoment let 119899 = 119899+1 the ratio betweenthe number of black and white pixel points in theannulus between 119877 minus (119899 + 1) lowast 119887 and 119877 is computedand is still denoted by 119903
2 Then go to (B) (Figure 8)
After obtaining the annular scale region in the abovepolar coordinate the scale region is selected on the basis ofangle and radius and copied into a new image called the scalemark region image119872 (Figure 9)
413 Scale Mark Detection Based on the Improved CentralProjection Method In the newly generated image119872 whichis the annular scale region the OSTU method is adopted tosegment the scale marks to form the binarized scale imageIn this image the pixel value of the scale mark is 0 (black)and those in other places are 255 (white) Each black pixelpoint and the circle center are linked and the slope 119870 ofthe connecting line is calculated according to the formula119896 = tan120572 wherein 120572 is the angle between the 119909-axis and
8 Mathematical Problems in Engineering
(a) (b)
(c)
Figure 8 Schematic of scale mark region detection
(a) (b)
Figure 9 The scale mark region diagram and its binarized image
Mathematical Problems in Engineering 9
x
y
0
(a) Schematic of the image coordinate system
x
y
1
1
2
2
3
3
4
4
minus1
minus1
minus2
minus2
0minus3
minus3
minus4
minus4
(b) The rectangular coordinate system
Figure 10 Schematic of the image coordinate system and the rectangular coordinate system
the connecting line between the black pixel point and thecircular center Due to the origin of the image coordinatebeing in the top left corner and the119910-axis running downwardit is inconvenient to denote the central angle within thescope of 0∘ndash360∘ Therefore in this study the rectangularcoordinate system with the image center as its origin isintroduced to express the angle between the black pixelpoint on the scale mark and the circle center instead of theimage coordinate system As is shown in Figure 10 the anglesbetween the black pixel point and the circle center are in the0∘ndash360∘ range
After calculating the number for angle 120572 the angle witha large number (connecting lines) in a small angle range isthat of the thick scale mark (main scale) on the dialThe stepsto detect the scale mark by central projection are detailed asfollows
(1) The black pixel point and the image central pointthat is the circle center (119909
0 1199100) in the new image are linked
to determine the slope 119896 which is converted to an angleaccording to the following formula
120572 = arctan (119896) lowast 180
120587 (3)
(2) As shown in Figure 10 the image coordinate systemis transformed to a rectangular one The angle 120572 followingformula (4) is converted into an angle in the rectangularcoordinate
120579 = minus120572 119894 lt 119903 119895 gt 119903 the first quadrant
120579 = 180 minus 120572 119894 lt 119903 119895 lt 119903 the second quadrant
120579 = 180 minus 120572 119894 gt 119903 119895 lt 119903 the third quadrant
120579 = 360 minus 120572 119894 gt 119903 119895 gt 119903 the forth quadrant
(4)
Occ
urre
nce t
imes
Angle (deg)1205791 1205792 1205793 12057940
Figure 11 The angle histogram
(3) An angle histogram is generated by recording theoccurrence frequency of each angle Figure 11 shows the anglehistogram
(4) The angle with the most occurrence times in thehistogram is selected and denoted by max all the angleswhose occurrence times are more thanmax 2 are foundTheoccurrence times of the angles are denoted by 119891(120579) Here 120579represents the angles that occur more than max 2
(5) For every angle 120579 we calculate the weighted averagevalue anglemean(119898) according the following to formula
anglemean (119898)
=119891 (120579 + 1) lowast (120579 + 1) + 119891 (120579) lowast (120579) + 119891 (120579 minus 1) lowast (120579 minus 1)
119891 (120579 + 1) + 119891 (120579) + 119891 (120579 minus 1)
(5)
Here119898 is the sign of the angle
10 Mathematical Problems in Engineering
12057221205721
1205720
1205723
Thick scale marksy
x
Figure 12 The location of the thick scale marks
(6) Calculate specific value 119875 according to the followingformula
119875 =anglemean (119898)
anglemean (119898 + 1) (6)
If 119875 is within the 09ndash11 range anglemean(119898) is the angleof the thick scale mark All the angles sorted out successivelyare taken as 120572
0 1205721 1205722 which correspond to the thick
scale marks Therefore the thick scale marks are located onFigure 12
414 Improved Hough Transform-Based Pointer DetectionThe edge detection is implemented in the pointer gaugeimage using the Canny operator to obtain the edge imageof the gauge Then the Hough transform method is adoptedto fit the straight line in the image to obtain all the straightlines among which the pointer border is included Finallyonly the pointer border can be determined by excluding theother straight edges
(1) The Hough transform method is adopted to fit thestraight line and find all the straight lines in the image Inaddition each straight line described in the image coordinatesystem by 119910 = 119896119909 + 119887 is calculated
(2)The distance 119889 from the circle center (1199090 1199100) to the
straight line is calculated according to the following formula
119889 =
10038161003816100381610038161199100 minus 119896 lowast 1199090 minus 1198871003816100381610038161003816
radic1 + 1198962 (7)
Based on prior knowledge we know that the distancefrom the circle center to the straight line of the pointer ismuch less than the diameter of the dial all the straight lines
Figure 13 Schematic of pointer fitting and its direction
conforming to 119889 lt 211990310 are picked out Here 119903 is thediameter of the dial
(3) The angle of each straight line in (1) is calculatedaccording to formula (8) and stored into an array angle[]
120593 =arctan (119896) lowast 180
120587 (8)
(4) As the angles of the pointer direction are acute onesgenerally in the range of 2∘ sim 10
∘ the angle Δ120593 is calculatedsubject to the following formula
Δ120593 =1003816100381610038161003816angle [119899] minus angle [119899 + 1]1003816100381610038161003816 (9)
If it is in the situation where 2 lt Δ120593 lt 10 the straight lineshould be selected as one side of the pointer Therefore bothlines where the two sides of the pointer lie can be found
(5) According to the straight lines of the two sides thepointer direction can be determined
42 The Pointer Gauge Indicating Value Recognition As isshown in Figure 11 the intersection point 119866 of the twostraight lines of the pointer border is calculated and thenlinked to the circle center 119874(119909
0 1199100) The angle of the
line 119874119866 is calculated In accordance with the position ofthe intersection point 119866 the angle 120573 in the rectangularcoordinate system is computed according to formula (1)(Figure 13)
According to the angle 120573 the readings 1198630 1198631 119863
119899
corresponding to each thick scale mark are obtained Basedon the angle that corresponds to the thick scale markand its number 119899 which is detected in Section 414 thepointer reading can be determined according to formula (10)
Mathematical Problems in Engineering 11
Table 1 Experimental samples
Sample name Image resolution Sample character The purpose of sample selectionThe pressure gauge with manylabels and a lot of writing on thedial
1000750A lot of writing on the dial which mayinfluence the location of the dial centerand the pointer segmentation
Verify the robustness of thealgorithm and the capability ofresisting interference
Flow gauge with the unevenscale 500500 Uneven distribution of the scale marks Verify the universality of the
algorithm
The scale within the adjacent thick scale marks is divided bylinear interpolation
119878indicating value
=
10038161003816100381610038161003816100381610038161003816
1205720minus 120573
1205720minus 1205721
10038161003816100381610038161003816100381610038161003816
(1198631minus 1198630) 120572
0lt 120573 lt 120572
1
10038161003816100381610038161003816100381610038161003816
1205721minus 120573
1205721minus 1205722
10038161003816100381610038161003816100381610038161003816
(1198632minus 1198631) + 1198631 120572
1lt 120573 le 120572
2
10038161003816100381610038161003816100381610038161003816
120572119899minus1
minus 120573
120572119899minus1
minus 120572119899
10038161003816100381610038161003816100381610038161003816
(119863119899minus 119863119899minus1
) + 119863119899minus1
120572119899minus1
lt 120573 le 120572119899
(10)
Here 1205720 1205721 1205722 120572
119899are the angles that correspond to the
thick scale marks
5 Experimental Results and Analysis
To verify the effectiveness of the proposed approach a seriesof experiments are conducted to test its performance
51 Experiment Design The experiments include two partsFirst the images of the different kinds of pointer gauges areselected to conduct the simulation experiments and verifythe effectiveness and applicability of the approach Thenthe pressure gauge calibration system which is composedof a float gauge high precise pressure gauge test pressuregauge camera and computer is established to verify theactual performance of this approach in the pointer gaugecalibration
52 Part One of the Experiments Reading Identification ofthe Various Pointer Gauges
521 Samples Selection As can be seen in Table 1 two groupsof representative samples are selected for the experimentThey include the pressure gauge with a lot of writing on thedial and the flow gauge with an uneven scale
522 Experimental Results and Analysis In this part ofthe experiment we adopt the proposed approach to theautomatic recognition of the indicating value of the abovementioned two types of gauges To verify the effectivenessof the proposed approach Figures 14 and 15 illustrate theexperimental results of each step
Figure 14(a) is the original pointer gauge image FromFigure 14(a) we can see that the pointer and scale marks
cannot be directly detected by segmentation methods sincethere is a lot of writing on the dial Therefore the proposedapproach applies an idea that conducts a coarse-to-finesearch The dial and scale regions are located first since theyare more easily detected than the pointer and scale marks Asshown in Figures 14(c) and 14(d) the dial region is locatedusing the adaptive-region-growing method and the scaleregion is detected under the polar coordinate system Thenin both regions the pointer and scale marks are accuratelydetected As shown in Figures 14(e) and 14(f) the improvedHough transform method is adopted to detect the pointerand the improved central projectionmethod is used to detectthe scale marks The location of the thick scale marks isdetermined according to the angle histogram generated fromthe central projection of the scale marks Their indicatingvalues are given according to the location of the thick scalemarks The scale values within any two adjacent thick marksare determined by linear division According to the directionof the pointer and the location of the scale marks the anglemethod which is adopted in the adjacent thick scale marksis used to indicate the pointer gauge value
The proposed approach detects the pointer direction andthe scale marks location respectively In Figures 15(e) and15(f) it can be seen that the location of the scale marks isprecisely obtained by the angle histogram and the pointer isdetected in the dial region Based on the location of the scalemarks and the pointer direction the distance between thepointer and the adjacent scale marks can be calculated usingthe distance method then the pointer indicating value canbe estimatedTherefore whether or not the distribution of thescale is even the proposed approach is able to precisely detectthe gaugersquos indicating value while the most current existingmethods cannot
53 Part Two of the Experiments The Recognition of the FloatPointer Gauge Indicating Value
531 The System and Procedure of Experiments To verifythe performance of the practical application of the proposedapproach in practical application the float gauge is used as thepressure gauge verification system The type of float gauge isY-047 The main technical parameters of the float gauge aredescribed as follows The range of output pressure is 001ndash025MPa the accuracy class is plusmn005 the rated pressureis 05MPa and the standard weights are from 001MPato 25MPa A high-precision gauge is used as the standardpressure gauge whose accuracy class is 15 The test gaugeis a gauge with general accuracy A visual detection systemis established in front of the float gauge An optical lens
12 Mathematical Problems in Engineering
(a) The original gauge image (b) The dial region of the manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 1205723 12057241205720 1205725 1205726 1205727 1205728 1205729 12057210
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 14 The detection results of the pointer gauge with a lot of writing on the dial
whose focal length can be adjustable from 10 to 20mm anda 500-megapixel HD CMOS camera with a 60 fps framerate are applied as image sensors The system is equippedwith a coaxial light source which is placed on the cameralens The hardware configuration of the computer used inthis experiment is Inter Pentium(R) Dual CPU E2200 and
22GHz The programming environment is VS2010 Theverification system of the pointer gauge that is based oncomputer vision is shown in Figure 16 The camera used inthe experiments takes images of the high-precision standardgauge and the test gauge The proposed approach is adoptedto detect the indicating value of both standard and test gauges
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 7
(a) (b)
(c) (d)
Figure 6 The region growing results under different pixel similarity threshold values
90∘
60∘
210∘
150∘
120∘
180∘
240∘
270∘300∘
330∘
30∘
0∘
Figure 7 Schematic of the polar coordinate
the OSTU method [25] to perform the binarizationsegmentation in the annular region
(B) The ratio between the number of black andwhite pixelpoints in the annular binarization region between theradius 119877 and 119877minus119899lowast119887 is calculated and denoted by 119903
1
After reducing the inner radius according to the steplength 119887 the ratio between the number of black and
white pixel points in the region between 119877minus(119899+1)lowast119887and 119877 is computed and denoted by 119903
2
(C) If 1199031is obviously different from 119903
2 the inner radius
and the outer radius of the scale region would be 119877 minus(119899 + 1) lowast 119887 and 119877 respectively
(D) If 1199031is approximated to 119903
2 the annulus between 119877 minus
(119899 + 1) lowast 119887 and 119877 would still be in the circular scaleregion At thismoment let 119899 = 119899+1 the ratio betweenthe number of black and white pixel points in theannulus between 119877 minus (119899 + 1) lowast 119887 and 119877 is computedand is still denoted by 119903
2 Then go to (B) (Figure 8)
After obtaining the annular scale region in the abovepolar coordinate the scale region is selected on the basis ofangle and radius and copied into a new image called the scalemark region image119872 (Figure 9)
413 Scale Mark Detection Based on the Improved CentralProjection Method In the newly generated image119872 whichis the annular scale region the OSTU method is adopted tosegment the scale marks to form the binarized scale imageIn this image the pixel value of the scale mark is 0 (black)and those in other places are 255 (white) Each black pixelpoint and the circle center are linked and the slope 119870 ofthe connecting line is calculated according to the formula119896 = tan120572 wherein 120572 is the angle between the 119909-axis and
8 Mathematical Problems in Engineering
(a) (b)
(c)
Figure 8 Schematic of scale mark region detection
(a) (b)
Figure 9 The scale mark region diagram and its binarized image
Mathematical Problems in Engineering 9
x
y
0
(a) Schematic of the image coordinate system
x
y
1
1
2
2
3
3
4
4
minus1
minus1
minus2
minus2
0minus3
minus3
minus4
minus4
(b) The rectangular coordinate system
Figure 10 Schematic of the image coordinate system and the rectangular coordinate system
the connecting line between the black pixel point and thecircular center Due to the origin of the image coordinatebeing in the top left corner and the119910-axis running downwardit is inconvenient to denote the central angle within thescope of 0∘ndash360∘ Therefore in this study the rectangularcoordinate system with the image center as its origin isintroduced to express the angle between the black pixelpoint on the scale mark and the circle center instead of theimage coordinate system As is shown in Figure 10 the anglesbetween the black pixel point and the circle center are in the0∘ndash360∘ range
After calculating the number for angle 120572 the angle witha large number (connecting lines) in a small angle range isthat of the thick scale mark (main scale) on the dialThe stepsto detect the scale mark by central projection are detailed asfollows
(1) The black pixel point and the image central pointthat is the circle center (119909
0 1199100) in the new image are linked
to determine the slope 119896 which is converted to an angleaccording to the following formula
120572 = arctan (119896) lowast 180
120587 (3)
(2) As shown in Figure 10 the image coordinate systemis transformed to a rectangular one The angle 120572 followingformula (4) is converted into an angle in the rectangularcoordinate
120579 = minus120572 119894 lt 119903 119895 gt 119903 the first quadrant
120579 = 180 minus 120572 119894 lt 119903 119895 lt 119903 the second quadrant
120579 = 180 minus 120572 119894 gt 119903 119895 lt 119903 the third quadrant
120579 = 360 minus 120572 119894 gt 119903 119895 gt 119903 the forth quadrant
(4)
Occ
urre
nce t
imes
Angle (deg)1205791 1205792 1205793 12057940
Figure 11 The angle histogram
(3) An angle histogram is generated by recording theoccurrence frequency of each angle Figure 11 shows the anglehistogram
(4) The angle with the most occurrence times in thehistogram is selected and denoted by max all the angleswhose occurrence times are more thanmax 2 are foundTheoccurrence times of the angles are denoted by 119891(120579) Here 120579represents the angles that occur more than max 2
(5) For every angle 120579 we calculate the weighted averagevalue anglemean(119898) according the following to formula
anglemean (119898)
=119891 (120579 + 1) lowast (120579 + 1) + 119891 (120579) lowast (120579) + 119891 (120579 minus 1) lowast (120579 minus 1)
119891 (120579 + 1) + 119891 (120579) + 119891 (120579 minus 1)
(5)
Here119898 is the sign of the angle
10 Mathematical Problems in Engineering
12057221205721
1205720
1205723
Thick scale marksy
x
Figure 12 The location of the thick scale marks
(6) Calculate specific value 119875 according to the followingformula
119875 =anglemean (119898)
anglemean (119898 + 1) (6)
If 119875 is within the 09ndash11 range anglemean(119898) is the angleof the thick scale mark All the angles sorted out successivelyare taken as 120572
0 1205721 1205722 which correspond to the thick
scale marks Therefore the thick scale marks are located onFigure 12
414 Improved Hough Transform-Based Pointer DetectionThe edge detection is implemented in the pointer gaugeimage using the Canny operator to obtain the edge imageof the gauge Then the Hough transform method is adoptedto fit the straight line in the image to obtain all the straightlines among which the pointer border is included Finallyonly the pointer border can be determined by excluding theother straight edges
(1) The Hough transform method is adopted to fit thestraight line and find all the straight lines in the image Inaddition each straight line described in the image coordinatesystem by 119910 = 119896119909 + 119887 is calculated
(2)The distance 119889 from the circle center (1199090 1199100) to the
straight line is calculated according to the following formula
119889 =
10038161003816100381610038161199100 minus 119896 lowast 1199090 minus 1198871003816100381610038161003816
radic1 + 1198962 (7)
Based on prior knowledge we know that the distancefrom the circle center to the straight line of the pointer ismuch less than the diameter of the dial all the straight lines
Figure 13 Schematic of pointer fitting and its direction
conforming to 119889 lt 211990310 are picked out Here 119903 is thediameter of the dial
(3) The angle of each straight line in (1) is calculatedaccording to formula (8) and stored into an array angle[]
120593 =arctan (119896) lowast 180
120587 (8)
(4) As the angles of the pointer direction are acute onesgenerally in the range of 2∘ sim 10
∘ the angle Δ120593 is calculatedsubject to the following formula
Δ120593 =1003816100381610038161003816angle [119899] minus angle [119899 + 1]1003816100381610038161003816 (9)
If it is in the situation where 2 lt Δ120593 lt 10 the straight lineshould be selected as one side of the pointer Therefore bothlines where the two sides of the pointer lie can be found
(5) According to the straight lines of the two sides thepointer direction can be determined
42 The Pointer Gauge Indicating Value Recognition As isshown in Figure 11 the intersection point 119866 of the twostraight lines of the pointer border is calculated and thenlinked to the circle center 119874(119909
0 1199100) The angle of the
line 119874119866 is calculated In accordance with the position ofthe intersection point 119866 the angle 120573 in the rectangularcoordinate system is computed according to formula (1)(Figure 13)
According to the angle 120573 the readings 1198630 1198631 119863
119899
corresponding to each thick scale mark are obtained Basedon the angle that corresponds to the thick scale markand its number 119899 which is detected in Section 414 thepointer reading can be determined according to formula (10)
Mathematical Problems in Engineering 11
Table 1 Experimental samples
Sample name Image resolution Sample character The purpose of sample selectionThe pressure gauge with manylabels and a lot of writing on thedial
1000750A lot of writing on the dial which mayinfluence the location of the dial centerand the pointer segmentation
Verify the robustness of thealgorithm and the capability ofresisting interference
Flow gauge with the unevenscale 500500 Uneven distribution of the scale marks Verify the universality of the
algorithm
The scale within the adjacent thick scale marks is divided bylinear interpolation
119878indicating value
=
10038161003816100381610038161003816100381610038161003816
1205720minus 120573
1205720minus 1205721
10038161003816100381610038161003816100381610038161003816
(1198631minus 1198630) 120572
0lt 120573 lt 120572
1
10038161003816100381610038161003816100381610038161003816
1205721minus 120573
1205721minus 1205722
10038161003816100381610038161003816100381610038161003816
(1198632minus 1198631) + 1198631 120572
1lt 120573 le 120572
2
10038161003816100381610038161003816100381610038161003816
120572119899minus1
minus 120573
120572119899minus1
minus 120572119899
10038161003816100381610038161003816100381610038161003816
(119863119899minus 119863119899minus1
) + 119863119899minus1
120572119899minus1
lt 120573 le 120572119899
(10)
Here 1205720 1205721 1205722 120572
119899are the angles that correspond to the
thick scale marks
5 Experimental Results and Analysis
To verify the effectiveness of the proposed approach a seriesof experiments are conducted to test its performance
51 Experiment Design The experiments include two partsFirst the images of the different kinds of pointer gauges areselected to conduct the simulation experiments and verifythe effectiveness and applicability of the approach Thenthe pressure gauge calibration system which is composedof a float gauge high precise pressure gauge test pressuregauge camera and computer is established to verify theactual performance of this approach in the pointer gaugecalibration
52 Part One of the Experiments Reading Identification ofthe Various Pointer Gauges
521 Samples Selection As can be seen in Table 1 two groupsof representative samples are selected for the experimentThey include the pressure gauge with a lot of writing on thedial and the flow gauge with an uneven scale
522 Experimental Results and Analysis In this part ofthe experiment we adopt the proposed approach to theautomatic recognition of the indicating value of the abovementioned two types of gauges To verify the effectivenessof the proposed approach Figures 14 and 15 illustrate theexperimental results of each step
Figure 14(a) is the original pointer gauge image FromFigure 14(a) we can see that the pointer and scale marks
cannot be directly detected by segmentation methods sincethere is a lot of writing on the dial Therefore the proposedapproach applies an idea that conducts a coarse-to-finesearch The dial and scale regions are located first since theyare more easily detected than the pointer and scale marks Asshown in Figures 14(c) and 14(d) the dial region is locatedusing the adaptive-region-growing method and the scaleregion is detected under the polar coordinate system Thenin both regions the pointer and scale marks are accuratelydetected As shown in Figures 14(e) and 14(f) the improvedHough transform method is adopted to detect the pointerand the improved central projectionmethod is used to detectthe scale marks The location of the thick scale marks isdetermined according to the angle histogram generated fromthe central projection of the scale marks Their indicatingvalues are given according to the location of the thick scalemarks The scale values within any two adjacent thick marksare determined by linear division According to the directionof the pointer and the location of the scale marks the anglemethod which is adopted in the adjacent thick scale marksis used to indicate the pointer gauge value
The proposed approach detects the pointer direction andthe scale marks location respectively In Figures 15(e) and15(f) it can be seen that the location of the scale marks isprecisely obtained by the angle histogram and the pointer isdetected in the dial region Based on the location of the scalemarks and the pointer direction the distance between thepointer and the adjacent scale marks can be calculated usingthe distance method then the pointer indicating value canbe estimatedTherefore whether or not the distribution of thescale is even the proposed approach is able to precisely detectthe gaugersquos indicating value while the most current existingmethods cannot
53 Part Two of the Experiments The Recognition of the FloatPointer Gauge Indicating Value
531 The System and Procedure of Experiments To verifythe performance of the practical application of the proposedapproach in practical application the float gauge is used as thepressure gauge verification system The type of float gauge isY-047 The main technical parameters of the float gauge aredescribed as follows The range of output pressure is 001ndash025MPa the accuracy class is plusmn005 the rated pressureis 05MPa and the standard weights are from 001MPato 25MPa A high-precision gauge is used as the standardpressure gauge whose accuracy class is 15 The test gaugeis a gauge with general accuracy A visual detection systemis established in front of the float gauge An optical lens
12 Mathematical Problems in Engineering
(a) The original gauge image (b) The dial region of the manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 1205723 12057241205720 1205725 1205726 1205727 1205728 1205729 12057210
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 14 The detection results of the pointer gauge with a lot of writing on the dial
whose focal length can be adjustable from 10 to 20mm anda 500-megapixel HD CMOS camera with a 60 fps framerate are applied as image sensors The system is equippedwith a coaxial light source which is placed on the cameralens The hardware configuration of the computer used inthis experiment is Inter Pentium(R) Dual CPU E2200 and
22GHz The programming environment is VS2010 Theverification system of the pointer gauge that is based oncomputer vision is shown in Figure 16 The camera used inthe experiments takes images of the high-precision standardgauge and the test gauge The proposed approach is adoptedto detect the indicating value of both standard and test gauges
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
8 Mathematical Problems in Engineering
(a) (b)
(c)
Figure 8 Schematic of scale mark region detection
(a) (b)
Figure 9 The scale mark region diagram and its binarized image
Mathematical Problems in Engineering 9
x
y
0
(a) Schematic of the image coordinate system
x
y
1
1
2
2
3
3
4
4
minus1
minus1
minus2
minus2
0minus3
minus3
minus4
minus4
(b) The rectangular coordinate system
Figure 10 Schematic of the image coordinate system and the rectangular coordinate system
the connecting line between the black pixel point and thecircular center Due to the origin of the image coordinatebeing in the top left corner and the119910-axis running downwardit is inconvenient to denote the central angle within thescope of 0∘ndash360∘ Therefore in this study the rectangularcoordinate system with the image center as its origin isintroduced to express the angle between the black pixelpoint on the scale mark and the circle center instead of theimage coordinate system As is shown in Figure 10 the anglesbetween the black pixel point and the circle center are in the0∘ndash360∘ range
After calculating the number for angle 120572 the angle witha large number (connecting lines) in a small angle range isthat of the thick scale mark (main scale) on the dialThe stepsto detect the scale mark by central projection are detailed asfollows
(1) The black pixel point and the image central pointthat is the circle center (119909
0 1199100) in the new image are linked
to determine the slope 119896 which is converted to an angleaccording to the following formula
120572 = arctan (119896) lowast 180
120587 (3)
(2) As shown in Figure 10 the image coordinate systemis transformed to a rectangular one The angle 120572 followingformula (4) is converted into an angle in the rectangularcoordinate
120579 = minus120572 119894 lt 119903 119895 gt 119903 the first quadrant
120579 = 180 minus 120572 119894 lt 119903 119895 lt 119903 the second quadrant
120579 = 180 minus 120572 119894 gt 119903 119895 lt 119903 the third quadrant
120579 = 360 minus 120572 119894 gt 119903 119895 gt 119903 the forth quadrant
(4)
Occ
urre
nce t
imes
Angle (deg)1205791 1205792 1205793 12057940
Figure 11 The angle histogram
(3) An angle histogram is generated by recording theoccurrence frequency of each angle Figure 11 shows the anglehistogram
(4) The angle with the most occurrence times in thehistogram is selected and denoted by max all the angleswhose occurrence times are more thanmax 2 are foundTheoccurrence times of the angles are denoted by 119891(120579) Here 120579represents the angles that occur more than max 2
(5) For every angle 120579 we calculate the weighted averagevalue anglemean(119898) according the following to formula
anglemean (119898)
=119891 (120579 + 1) lowast (120579 + 1) + 119891 (120579) lowast (120579) + 119891 (120579 minus 1) lowast (120579 minus 1)
119891 (120579 + 1) + 119891 (120579) + 119891 (120579 minus 1)
(5)
Here119898 is the sign of the angle
10 Mathematical Problems in Engineering
12057221205721
1205720
1205723
Thick scale marksy
x
Figure 12 The location of the thick scale marks
(6) Calculate specific value 119875 according to the followingformula
119875 =anglemean (119898)
anglemean (119898 + 1) (6)
If 119875 is within the 09ndash11 range anglemean(119898) is the angleof the thick scale mark All the angles sorted out successivelyare taken as 120572
0 1205721 1205722 which correspond to the thick
scale marks Therefore the thick scale marks are located onFigure 12
414 Improved Hough Transform-Based Pointer DetectionThe edge detection is implemented in the pointer gaugeimage using the Canny operator to obtain the edge imageof the gauge Then the Hough transform method is adoptedto fit the straight line in the image to obtain all the straightlines among which the pointer border is included Finallyonly the pointer border can be determined by excluding theother straight edges
(1) The Hough transform method is adopted to fit thestraight line and find all the straight lines in the image Inaddition each straight line described in the image coordinatesystem by 119910 = 119896119909 + 119887 is calculated
(2)The distance 119889 from the circle center (1199090 1199100) to the
straight line is calculated according to the following formula
119889 =
10038161003816100381610038161199100 minus 119896 lowast 1199090 minus 1198871003816100381610038161003816
radic1 + 1198962 (7)
Based on prior knowledge we know that the distancefrom the circle center to the straight line of the pointer ismuch less than the diameter of the dial all the straight lines
Figure 13 Schematic of pointer fitting and its direction
conforming to 119889 lt 211990310 are picked out Here 119903 is thediameter of the dial
(3) The angle of each straight line in (1) is calculatedaccording to formula (8) and stored into an array angle[]
120593 =arctan (119896) lowast 180
120587 (8)
(4) As the angles of the pointer direction are acute onesgenerally in the range of 2∘ sim 10
∘ the angle Δ120593 is calculatedsubject to the following formula
Δ120593 =1003816100381610038161003816angle [119899] minus angle [119899 + 1]1003816100381610038161003816 (9)
If it is in the situation where 2 lt Δ120593 lt 10 the straight lineshould be selected as one side of the pointer Therefore bothlines where the two sides of the pointer lie can be found
(5) According to the straight lines of the two sides thepointer direction can be determined
42 The Pointer Gauge Indicating Value Recognition As isshown in Figure 11 the intersection point 119866 of the twostraight lines of the pointer border is calculated and thenlinked to the circle center 119874(119909
0 1199100) The angle of the
line 119874119866 is calculated In accordance with the position ofthe intersection point 119866 the angle 120573 in the rectangularcoordinate system is computed according to formula (1)(Figure 13)
According to the angle 120573 the readings 1198630 1198631 119863
119899
corresponding to each thick scale mark are obtained Basedon the angle that corresponds to the thick scale markand its number 119899 which is detected in Section 414 thepointer reading can be determined according to formula (10)
Mathematical Problems in Engineering 11
Table 1 Experimental samples
Sample name Image resolution Sample character The purpose of sample selectionThe pressure gauge with manylabels and a lot of writing on thedial
1000750A lot of writing on the dial which mayinfluence the location of the dial centerand the pointer segmentation
Verify the robustness of thealgorithm and the capability ofresisting interference
Flow gauge with the unevenscale 500500 Uneven distribution of the scale marks Verify the universality of the
algorithm
The scale within the adjacent thick scale marks is divided bylinear interpolation
119878indicating value
=
10038161003816100381610038161003816100381610038161003816
1205720minus 120573
1205720minus 1205721
10038161003816100381610038161003816100381610038161003816
(1198631minus 1198630) 120572
0lt 120573 lt 120572
1
10038161003816100381610038161003816100381610038161003816
1205721minus 120573
1205721minus 1205722
10038161003816100381610038161003816100381610038161003816
(1198632minus 1198631) + 1198631 120572
1lt 120573 le 120572
2
10038161003816100381610038161003816100381610038161003816
120572119899minus1
minus 120573
120572119899minus1
minus 120572119899
10038161003816100381610038161003816100381610038161003816
(119863119899minus 119863119899minus1
) + 119863119899minus1
120572119899minus1
lt 120573 le 120572119899
(10)
Here 1205720 1205721 1205722 120572
119899are the angles that correspond to the
thick scale marks
5 Experimental Results and Analysis
To verify the effectiveness of the proposed approach a seriesof experiments are conducted to test its performance
51 Experiment Design The experiments include two partsFirst the images of the different kinds of pointer gauges areselected to conduct the simulation experiments and verifythe effectiveness and applicability of the approach Thenthe pressure gauge calibration system which is composedof a float gauge high precise pressure gauge test pressuregauge camera and computer is established to verify theactual performance of this approach in the pointer gaugecalibration
52 Part One of the Experiments Reading Identification ofthe Various Pointer Gauges
521 Samples Selection As can be seen in Table 1 two groupsof representative samples are selected for the experimentThey include the pressure gauge with a lot of writing on thedial and the flow gauge with an uneven scale
522 Experimental Results and Analysis In this part ofthe experiment we adopt the proposed approach to theautomatic recognition of the indicating value of the abovementioned two types of gauges To verify the effectivenessof the proposed approach Figures 14 and 15 illustrate theexperimental results of each step
Figure 14(a) is the original pointer gauge image FromFigure 14(a) we can see that the pointer and scale marks
cannot be directly detected by segmentation methods sincethere is a lot of writing on the dial Therefore the proposedapproach applies an idea that conducts a coarse-to-finesearch The dial and scale regions are located first since theyare more easily detected than the pointer and scale marks Asshown in Figures 14(c) and 14(d) the dial region is locatedusing the adaptive-region-growing method and the scaleregion is detected under the polar coordinate system Thenin both regions the pointer and scale marks are accuratelydetected As shown in Figures 14(e) and 14(f) the improvedHough transform method is adopted to detect the pointerand the improved central projectionmethod is used to detectthe scale marks The location of the thick scale marks isdetermined according to the angle histogram generated fromthe central projection of the scale marks Their indicatingvalues are given according to the location of the thick scalemarks The scale values within any two adjacent thick marksare determined by linear division According to the directionof the pointer and the location of the scale marks the anglemethod which is adopted in the adjacent thick scale marksis used to indicate the pointer gauge value
The proposed approach detects the pointer direction andthe scale marks location respectively In Figures 15(e) and15(f) it can be seen that the location of the scale marks isprecisely obtained by the angle histogram and the pointer isdetected in the dial region Based on the location of the scalemarks and the pointer direction the distance between thepointer and the adjacent scale marks can be calculated usingthe distance method then the pointer indicating value canbe estimatedTherefore whether or not the distribution of thescale is even the proposed approach is able to precisely detectthe gaugersquos indicating value while the most current existingmethods cannot
53 Part Two of the Experiments The Recognition of the FloatPointer Gauge Indicating Value
531 The System and Procedure of Experiments To verifythe performance of the practical application of the proposedapproach in practical application the float gauge is used as thepressure gauge verification system The type of float gauge isY-047 The main technical parameters of the float gauge aredescribed as follows The range of output pressure is 001ndash025MPa the accuracy class is plusmn005 the rated pressureis 05MPa and the standard weights are from 001MPato 25MPa A high-precision gauge is used as the standardpressure gauge whose accuracy class is 15 The test gaugeis a gauge with general accuracy A visual detection systemis established in front of the float gauge An optical lens
12 Mathematical Problems in Engineering
(a) The original gauge image (b) The dial region of the manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 1205723 12057241205720 1205725 1205726 1205727 1205728 1205729 12057210
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 14 The detection results of the pointer gauge with a lot of writing on the dial
whose focal length can be adjustable from 10 to 20mm anda 500-megapixel HD CMOS camera with a 60 fps framerate are applied as image sensors The system is equippedwith a coaxial light source which is placed on the cameralens The hardware configuration of the computer used inthis experiment is Inter Pentium(R) Dual CPU E2200 and
22GHz The programming environment is VS2010 Theverification system of the pointer gauge that is based oncomputer vision is shown in Figure 16 The camera used inthe experiments takes images of the high-precision standardgauge and the test gauge The proposed approach is adoptedto detect the indicating value of both standard and test gauges
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 9
x
y
0
(a) Schematic of the image coordinate system
x
y
1
1
2
2
3
3
4
4
minus1
minus1
minus2
minus2
0minus3
minus3
minus4
minus4
(b) The rectangular coordinate system
Figure 10 Schematic of the image coordinate system and the rectangular coordinate system
the connecting line between the black pixel point and thecircular center Due to the origin of the image coordinatebeing in the top left corner and the119910-axis running downwardit is inconvenient to denote the central angle within thescope of 0∘ndash360∘ Therefore in this study the rectangularcoordinate system with the image center as its origin isintroduced to express the angle between the black pixelpoint on the scale mark and the circle center instead of theimage coordinate system As is shown in Figure 10 the anglesbetween the black pixel point and the circle center are in the0∘ndash360∘ range
After calculating the number for angle 120572 the angle witha large number (connecting lines) in a small angle range isthat of the thick scale mark (main scale) on the dialThe stepsto detect the scale mark by central projection are detailed asfollows
(1) The black pixel point and the image central pointthat is the circle center (119909
0 1199100) in the new image are linked
to determine the slope 119896 which is converted to an angleaccording to the following formula
120572 = arctan (119896) lowast 180
120587 (3)
(2) As shown in Figure 10 the image coordinate systemis transformed to a rectangular one The angle 120572 followingformula (4) is converted into an angle in the rectangularcoordinate
120579 = minus120572 119894 lt 119903 119895 gt 119903 the first quadrant
120579 = 180 minus 120572 119894 lt 119903 119895 lt 119903 the second quadrant
120579 = 180 minus 120572 119894 gt 119903 119895 lt 119903 the third quadrant
120579 = 360 minus 120572 119894 gt 119903 119895 gt 119903 the forth quadrant
(4)
Occ
urre
nce t
imes
Angle (deg)1205791 1205792 1205793 12057940
Figure 11 The angle histogram
(3) An angle histogram is generated by recording theoccurrence frequency of each angle Figure 11 shows the anglehistogram
(4) The angle with the most occurrence times in thehistogram is selected and denoted by max all the angleswhose occurrence times are more thanmax 2 are foundTheoccurrence times of the angles are denoted by 119891(120579) Here 120579represents the angles that occur more than max 2
(5) For every angle 120579 we calculate the weighted averagevalue anglemean(119898) according the following to formula
anglemean (119898)
=119891 (120579 + 1) lowast (120579 + 1) + 119891 (120579) lowast (120579) + 119891 (120579 minus 1) lowast (120579 minus 1)
119891 (120579 + 1) + 119891 (120579) + 119891 (120579 minus 1)
(5)
Here119898 is the sign of the angle
10 Mathematical Problems in Engineering
12057221205721
1205720
1205723
Thick scale marksy
x
Figure 12 The location of the thick scale marks
(6) Calculate specific value 119875 according to the followingformula
119875 =anglemean (119898)
anglemean (119898 + 1) (6)
If 119875 is within the 09ndash11 range anglemean(119898) is the angleof the thick scale mark All the angles sorted out successivelyare taken as 120572
0 1205721 1205722 which correspond to the thick
scale marks Therefore the thick scale marks are located onFigure 12
414 Improved Hough Transform-Based Pointer DetectionThe edge detection is implemented in the pointer gaugeimage using the Canny operator to obtain the edge imageof the gauge Then the Hough transform method is adoptedto fit the straight line in the image to obtain all the straightlines among which the pointer border is included Finallyonly the pointer border can be determined by excluding theother straight edges
(1) The Hough transform method is adopted to fit thestraight line and find all the straight lines in the image Inaddition each straight line described in the image coordinatesystem by 119910 = 119896119909 + 119887 is calculated
(2)The distance 119889 from the circle center (1199090 1199100) to the
straight line is calculated according to the following formula
119889 =
10038161003816100381610038161199100 minus 119896 lowast 1199090 minus 1198871003816100381610038161003816
radic1 + 1198962 (7)
Based on prior knowledge we know that the distancefrom the circle center to the straight line of the pointer ismuch less than the diameter of the dial all the straight lines
Figure 13 Schematic of pointer fitting and its direction
conforming to 119889 lt 211990310 are picked out Here 119903 is thediameter of the dial
(3) The angle of each straight line in (1) is calculatedaccording to formula (8) and stored into an array angle[]
120593 =arctan (119896) lowast 180
120587 (8)
(4) As the angles of the pointer direction are acute onesgenerally in the range of 2∘ sim 10
∘ the angle Δ120593 is calculatedsubject to the following formula
Δ120593 =1003816100381610038161003816angle [119899] minus angle [119899 + 1]1003816100381610038161003816 (9)
If it is in the situation where 2 lt Δ120593 lt 10 the straight lineshould be selected as one side of the pointer Therefore bothlines where the two sides of the pointer lie can be found
(5) According to the straight lines of the two sides thepointer direction can be determined
42 The Pointer Gauge Indicating Value Recognition As isshown in Figure 11 the intersection point 119866 of the twostraight lines of the pointer border is calculated and thenlinked to the circle center 119874(119909
0 1199100) The angle of the
line 119874119866 is calculated In accordance with the position ofthe intersection point 119866 the angle 120573 in the rectangularcoordinate system is computed according to formula (1)(Figure 13)
According to the angle 120573 the readings 1198630 1198631 119863
119899
corresponding to each thick scale mark are obtained Basedon the angle that corresponds to the thick scale markand its number 119899 which is detected in Section 414 thepointer reading can be determined according to formula (10)
Mathematical Problems in Engineering 11
Table 1 Experimental samples
Sample name Image resolution Sample character The purpose of sample selectionThe pressure gauge with manylabels and a lot of writing on thedial
1000750A lot of writing on the dial which mayinfluence the location of the dial centerand the pointer segmentation
Verify the robustness of thealgorithm and the capability ofresisting interference
Flow gauge with the unevenscale 500500 Uneven distribution of the scale marks Verify the universality of the
algorithm
The scale within the adjacent thick scale marks is divided bylinear interpolation
119878indicating value
=
10038161003816100381610038161003816100381610038161003816
1205720minus 120573
1205720minus 1205721
10038161003816100381610038161003816100381610038161003816
(1198631minus 1198630) 120572
0lt 120573 lt 120572
1
10038161003816100381610038161003816100381610038161003816
1205721minus 120573
1205721minus 1205722
10038161003816100381610038161003816100381610038161003816
(1198632minus 1198631) + 1198631 120572
1lt 120573 le 120572
2
10038161003816100381610038161003816100381610038161003816
120572119899minus1
minus 120573
120572119899minus1
minus 120572119899
10038161003816100381610038161003816100381610038161003816
(119863119899minus 119863119899minus1
) + 119863119899minus1
120572119899minus1
lt 120573 le 120572119899
(10)
Here 1205720 1205721 1205722 120572
119899are the angles that correspond to the
thick scale marks
5 Experimental Results and Analysis
To verify the effectiveness of the proposed approach a seriesof experiments are conducted to test its performance
51 Experiment Design The experiments include two partsFirst the images of the different kinds of pointer gauges areselected to conduct the simulation experiments and verifythe effectiveness and applicability of the approach Thenthe pressure gauge calibration system which is composedof a float gauge high precise pressure gauge test pressuregauge camera and computer is established to verify theactual performance of this approach in the pointer gaugecalibration
52 Part One of the Experiments Reading Identification ofthe Various Pointer Gauges
521 Samples Selection As can be seen in Table 1 two groupsof representative samples are selected for the experimentThey include the pressure gauge with a lot of writing on thedial and the flow gauge with an uneven scale
522 Experimental Results and Analysis In this part ofthe experiment we adopt the proposed approach to theautomatic recognition of the indicating value of the abovementioned two types of gauges To verify the effectivenessof the proposed approach Figures 14 and 15 illustrate theexperimental results of each step
Figure 14(a) is the original pointer gauge image FromFigure 14(a) we can see that the pointer and scale marks
cannot be directly detected by segmentation methods sincethere is a lot of writing on the dial Therefore the proposedapproach applies an idea that conducts a coarse-to-finesearch The dial and scale regions are located first since theyare more easily detected than the pointer and scale marks Asshown in Figures 14(c) and 14(d) the dial region is locatedusing the adaptive-region-growing method and the scaleregion is detected under the polar coordinate system Thenin both regions the pointer and scale marks are accuratelydetected As shown in Figures 14(e) and 14(f) the improvedHough transform method is adopted to detect the pointerand the improved central projectionmethod is used to detectthe scale marks The location of the thick scale marks isdetermined according to the angle histogram generated fromthe central projection of the scale marks Their indicatingvalues are given according to the location of the thick scalemarks The scale values within any two adjacent thick marksare determined by linear division According to the directionof the pointer and the location of the scale marks the anglemethod which is adopted in the adjacent thick scale marksis used to indicate the pointer gauge value
The proposed approach detects the pointer direction andthe scale marks location respectively In Figures 15(e) and15(f) it can be seen that the location of the scale marks isprecisely obtained by the angle histogram and the pointer isdetected in the dial region Based on the location of the scalemarks and the pointer direction the distance between thepointer and the adjacent scale marks can be calculated usingthe distance method then the pointer indicating value canbe estimatedTherefore whether or not the distribution of thescale is even the proposed approach is able to precisely detectthe gaugersquos indicating value while the most current existingmethods cannot
53 Part Two of the Experiments The Recognition of the FloatPointer Gauge Indicating Value
531 The System and Procedure of Experiments To verifythe performance of the practical application of the proposedapproach in practical application the float gauge is used as thepressure gauge verification system The type of float gauge isY-047 The main technical parameters of the float gauge aredescribed as follows The range of output pressure is 001ndash025MPa the accuracy class is plusmn005 the rated pressureis 05MPa and the standard weights are from 001MPato 25MPa A high-precision gauge is used as the standardpressure gauge whose accuracy class is 15 The test gaugeis a gauge with general accuracy A visual detection systemis established in front of the float gauge An optical lens
12 Mathematical Problems in Engineering
(a) The original gauge image (b) The dial region of the manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 1205723 12057241205720 1205725 1205726 1205727 1205728 1205729 12057210
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 14 The detection results of the pointer gauge with a lot of writing on the dial
whose focal length can be adjustable from 10 to 20mm anda 500-megapixel HD CMOS camera with a 60 fps framerate are applied as image sensors The system is equippedwith a coaxial light source which is placed on the cameralens The hardware configuration of the computer used inthis experiment is Inter Pentium(R) Dual CPU E2200 and
22GHz The programming environment is VS2010 Theverification system of the pointer gauge that is based oncomputer vision is shown in Figure 16 The camera used inthe experiments takes images of the high-precision standardgauge and the test gauge The proposed approach is adoptedto detect the indicating value of both standard and test gauges
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
10 Mathematical Problems in Engineering
12057221205721
1205720
1205723
Thick scale marksy
x
Figure 12 The location of the thick scale marks
(6) Calculate specific value 119875 according to the followingformula
119875 =anglemean (119898)
anglemean (119898 + 1) (6)
If 119875 is within the 09ndash11 range anglemean(119898) is the angleof the thick scale mark All the angles sorted out successivelyare taken as 120572
0 1205721 1205722 which correspond to the thick
scale marks Therefore the thick scale marks are located onFigure 12
414 Improved Hough Transform-Based Pointer DetectionThe edge detection is implemented in the pointer gaugeimage using the Canny operator to obtain the edge imageof the gauge Then the Hough transform method is adoptedto fit the straight line in the image to obtain all the straightlines among which the pointer border is included Finallyonly the pointer border can be determined by excluding theother straight edges
(1) The Hough transform method is adopted to fit thestraight line and find all the straight lines in the image Inaddition each straight line described in the image coordinatesystem by 119910 = 119896119909 + 119887 is calculated
(2)The distance 119889 from the circle center (1199090 1199100) to the
straight line is calculated according to the following formula
119889 =
10038161003816100381610038161199100 minus 119896 lowast 1199090 minus 1198871003816100381610038161003816
radic1 + 1198962 (7)
Based on prior knowledge we know that the distancefrom the circle center to the straight line of the pointer ismuch less than the diameter of the dial all the straight lines
Figure 13 Schematic of pointer fitting and its direction
conforming to 119889 lt 211990310 are picked out Here 119903 is thediameter of the dial
(3) The angle of each straight line in (1) is calculatedaccording to formula (8) and stored into an array angle[]
120593 =arctan (119896) lowast 180
120587 (8)
(4) As the angles of the pointer direction are acute onesgenerally in the range of 2∘ sim 10
∘ the angle Δ120593 is calculatedsubject to the following formula
Δ120593 =1003816100381610038161003816angle [119899] minus angle [119899 + 1]1003816100381610038161003816 (9)
If it is in the situation where 2 lt Δ120593 lt 10 the straight lineshould be selected as one side of the pointer Therefore bothlines where the two sides of the pointer lie can be found
(5) According to the straight lines of the two sides thepointer direction can be determined
42 The Pointer Gauge Indicating Value Recognition As isshown in Figure 11 the intersection point 119866 of the twostraight lines of the pointer border is calculated and thenlinked to the circle center 119874(119909
0 1199100) The angle of the
line 119874119866 is calculated In accordance with the position ofthe intersection point 119866 the angle 120573 in the rectangularcoordinate system is computed according to formula (1)(Figure 13)
According to the angle 120573 the readings 1198630 1198631 119863
119899
corresponding to each thick scale mark are obtained Basedon the angle that corresponds to the thick scale markand its number 119899 which is detected in Section 414 thepointer reading can be determined according to formula (10)
Mathematical Problems in Engineering 11
Table 1 Experimental samples
Sample name Image resolution Sample character The purpose of sample selectionThe pressure gauge with manylabels and a lot of writing on thedial
1000750A lot of writing on the dial which mayinfluence the location of the dial centerand the pointer segmentation
Verify the robustness of thealgorithm and the capability ofresisting interference
Flow gauge with the unevenscale 500500 Uneven distribution of the scale marks Verify the universality of the
algorithm
The scale within the adjacent thick scale marks is divided bylinear interpolation
119878indicating value
=
10038161003816100381610038161003816100381610038161003816
1205720minus 120573
1205720minus 1205721
10038161003816100381610038161003816100381610038161003816
(1198631minus 1198630) 120572
0lt 120573 lt 120572
1
10038161003816100381610038161003816100381610038161003816
1205721minus 120573
1205721minus 1205722
10038161003816100381610038161003816100381610038161003816
(1198632minus 1198631) + 1198631 120572
1lt 120573 le 120572
2
10038161003816100381610038161003816100381610038161003816
120572119899minus1
minus 120573
120572119899minus1
minus 120572119899
10038161003816100381610038161003816100381610038161003816
(119863119899minus 119863119899minus1
) + 119863119899minus1
120572119899minus1
lt 120573 le 120572119899
(10)
Here 1205720 1205721 1205722 120572
119899are the angles that correspond to the
thick scale marks
5 Experimental Results and Analysis
To verify the effectiveness of the proposed approach a seriesof experiments are conducted to test its performance
51 Experiment Design The experiments include two partsFirst the images of the different kinds of pointer gauges areselected to conduct the simulation experiments and verifythe effectiveness and applicability of the approach Thenthe pressure gauge calibration system which is composedof a float gauge high precise pressure gauge test pressuregauge camera and computer is established to verify theactual performance of this approach in the pointer gaugecalibration
52 Part One of the Experiments Reading Identification ofthe Various Pointer Gauges
521 Samples Selection As can be seen in Table 1 two groupsof representative samples are selected for the experimentThey include the pressure gauge with a lot of writing on thedial and the flow gauge with an uneven scale
522 Experimental Results and Analysis In this part ofthe experiment we adopt the proposed approach to theautomatic recognition of the indicating value of the abovementioned two types of gauges To verify the effectivenessof the proposed approach Figures 14 and 15 illustrate theexperimental results of each step
Figure 14(a) is the original pointer gauge image FromFigure 14(a) we can see that the pointer and scale marks
cannot be directly detected by segmentation methods sincethere is a lot of writing on the dial Therefore the proposedapproach applies an idea that conducts a coarse-to-finesearch The dial and scale regions are located first since theyare more easily detected than the pointer and scale marks Asshown in Figures 14(c) and 14(d) the dial region is locatedusing the adaptive-region-growing method and the scaleregion is detected under the polar coordinate system Thenin both regions the pointer and scale marks are accuratelydetected As shown in Figures 14(e) and 14(f) the improvedHough transform method is adopted to detect the pointerand the improved central projectionmethod is used to detectthe scale marks The location of the thick scale marks isdetermined according to the angle histogram generated fromthe central projection of the scale marks Their indicatingvalues are given according to the location of the thick scalemarks The scale values within any two adjacent thick marksare determined by linear division According to the directionof the pointer and the location of the scale marks the anglemethod which is adopted in the adjacent thick scale marksis used to indicate the pointer gauge value
The proposed approach detects the pointer direction andthe scale marks location respectively In Figures 15(e) and15(f) it can be seen that the location of the scale marks isprecisely obtained by the angle histogram and the pointer isdetected in the dial region Based on the location of the scalemarks and the pointer direction the distance between thepointer and the adjacent scale marks can be calculated usingthe distance method then the pointer indicating value canbe estimatedTherefore whether or not the distribution of thescale is even the proposed approach is able to precisely detectthe gaugersquos indicating value while the most current existingmethods cannot
53 Part Two of the Experiments The Recognition of the FloatPointer Gauge Indicating Value
531 The System and Procedure of Experiments To verifythe performance of the practical application of the proposedapproach in practical application the float gauge is used as thepressure gauge verification system The type of float gauge isY-047 The main technical parameters of the float gauge aredescribed as follows The range of output pressure is 001ndash025MPa the accuracy class is plusmn005 the rated pressureis 05MPa and the standard weights are from 001MPato 25MPa A high-precision gauge is used as the standardpressure gauge whose accuracy class is 15 The test gaugeis a gauge with general accuracy A visual detection systemis established in front of the float gauge An optical lens
12 Mathematical Problems in Engineering
(a) The original gauge image (b) The dial region of the manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 1205723 12057241205720 1205725 1205726 1205727 1205728 1205729 12057210
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 14 The detection results of the pointer gauge with a lot of writing on the dial
whose focal length can be adjustable from 10 to 20mm anda 500-megapixel HD CMOS camera with a 60 fps framerate are applied as image sensors The system is equippedwith a coaxial light source which is placed on the cameralens The hardware configuration of the computer used inthis experiment is Inter Pentium(R) Dual CPU E2200 and
22GHz The programming environment is VS2010 Theverification system of the pointer gauge that is based oncomputer vision is shown in Figure 16 The camera used inthe experiments takes images of the high-precision standardgauge and the test gauge The proposed approach is adoptedto detect the indicating value of both standard and test gauges
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 11
Table 1 Experimental samples
Sample name Image resolution Sample character The purpose of sample selectionThe pressure gauge with manylabels and a lot of writing on thedial
1000750A lot of writing on the dial which mayinfluence the location of the dial centerand the pointer segmentation
Verify the robustness of thealgorithm and the capability ofresisting interference
Flow gauge with the unevenscale 500500 Uneven distribution of the scale marks Verify the universality of the
algorithm
The scale within the adjacent thick scale marks is divided bylinear interpolation
119878indicating value
=
10038161003816100381610038161003816100381610038161003816
1205720minus 120573
1205720minus 1205721
10038161003816100381610038161003816100381610038161003816
(1198631minus 1198630) 120572
0lt 120573 lt 120572
1
10038161003816100381610038161003816100381610038161003816
1205721minus 120573
1205721minus 1205722
10038161003816100381610038161003816100381610038161003816
(1198632minus 1198631) + 1198631 120572
1lt 120573 le 120572
2
10038161003816100381610038161003816100381610038161003816
120572119899minus1
minus 120573
120572119899minus1
minus 120572119899
10038161003816100381610038161003816100381610038161003816
(119863119899minus 119863119899minus1
) + 119863119899minus1
120572119899minus1
lt 120573 le 120572119899
(10)
Here 1205720 1205721 1205722 120572
119899are the angles that correspond to the
thick scale marks
5 Experimental Results and Analysis
To verify the effectiveness of the proposed approach a seriesof experiments are conducted to test its performance
51 Experiment Design The experiments include two partsFirst the images of the different kinds of pointer gauges areselected to conduct the simulation experiments and verifythe effectiveness and applicability of the approach Thenthe pressure gauge calibration system which is composedof a float gauge high precise pressure gauge test pressuregauge camera and computer is established to verify theactual performance of this approach in the pointer gaugecalibration
52 Part One of the Experiments Reading Identification ofthe Various Pointer Gauges
521 Samples Selection As can be seen in Table 1 two groupsof representative samples are selected for the experimentThey include the pressure gauge with a lot of writing on thedial and the flow gauge with an uneven scale
522 Experimental Results and Analysis In this part ofthe experiment we adopt the proposed approach to theautomatic recognition of the indicating value of the abovementioned two types of gauges To verify the effectivenessof the proposed approach Figures 14 and 15 illustrate theexperimental results of each step
Figure 14(a) is the original pointer gauge image FromFigure 14(a) we can see that the pointer and scale marks
cannot be directly detected by segmentation methods sincethere is a lot of writing on the dial Therefore the proposedapproach applies an idea that conducts a coarse-to-finesearch The dial and scale regions are located first since theyare more easily detected than the pointer and scale marks Asshown in Figures 14(c) and 14(d) the dial region is locatedusing the adaptive-region-growing method and the scaleregion is detected under the polar coordinate system Thenin both regions the pointer and scale marks are accuratelydetected As shown in Figures 14(e) and 14(f) the improvedHough transform method is adopted to detect the pointerand the improved central projectionmethod is used to detectthe scale marks The location of the thick scale marks isdetermined according to the angle histogram generated fromthe central projection of the scale marks Their indicatingvalues are given according to the location of the thick scalemarks The scale values within any two adjacent thick marksare determined by linear division According to the directionof the pointer and the location of the scale marks the anglemethod which is adopted in the adjacent thick scale marksis used to indicate the pointer gauge value
The proposed approach detects the pointer direction andthe scale marks location respectively In Figures 15(e) and15(f) it can be seen that the location of the scale marks isprecisely obtained by the angle histogram and the pointer isdetected in the dial region Based on the location of the scalemarks and the pointer direction the distance between thepointer and the adjacent scale marks can be calculated usingthe distance method then the pointer indicating value canbe estimatedTherefore whether or not the distribution of thescale is even the proposed approach is able to precisely detectthe gaugersquos indicating value while the most current existingmethods cannot
53 Part Two of the Experiments The Recognition of the FloatPointer Gauge Indicating Value
531 The System and Procedure of Experiments To verifythe performance of the practical application of the proposedapproach in practical application the float gauge is used as thepressure gauge verification system The type of float gauge isY-047 The main technical parameters of the float gauge aredescribed as follows The range of output pressure is 001ndash025MPa the accuracy class is plusmn005 the rated pressureis 05MPa and the standard weights are from 001MPato 25MPa A high-precision gauge is used as the standardpressure gauge whose accuracy class is 15 The test gaugeis a gauge with general accuracy A visual detection systemis established in front of the float gauge An optical lens
12 Mathematical Problems in Engineering
(a) The original gauge image (b) The dial region of the manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 1205723 12057241205720 1205725 1205726 1205727 1205728 1205729 12057210
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 14 The detection results of the pointer gauge with a lot of writing on the dial
whose focal length can be adjustable from 10 to 20mm anda 500-megapixel HD CMOS camera with a 60 fps framerate are applied as image sensors The system is equippedwith a coaxial light source which is placed on the cameralens The hardware configuration of the computer used inthis experiment is Inter Pentium(R) Dual CPU E2200 and
22GHz The programming environment is VS2010 Theverification system of the pointer gauge that is based oncomputer vision is shown in Figure 16 The camera used inthe experiments takes images of the high-precision standardgauge and the test gauge The proposed approach is adoptedto detect the indicating value of both standard and test gauges
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
12 Mathematical Problems in Engineering
(a) The original gauge image (b) The dial region of the manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 1205723 12057241205720 1205725 1205726 1205727 1205728 1205729 12057210
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 14 The detection results of the pointer gauge with a lot of writing on the dial
whose focal length can be adjustable from 10 to 20mm anda 500-megapixel HD CMOS camera with a 60 fps framerate are applied as image sensors The system is equippedwith a coaxial light source which is placed on the cameralens The hardware configuration of the computer used inthis experiment is Inter Pentium(R) Dual CPU E2200 and
22GHz The programming environment is VS2010 Theverification system of the pointer gauge that is based oncomputer vision is shown in Figure 16 The camera used inthe experiments takes images of the high-precision standardgauge and the test gauge The proposed approach is adoptedto detect the indicating value of both standard and test gauges
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 13
(a) Original gauge image (b) Dial region by manual interception
(c) Detection results of dial center and dial region (d) Detection results of scale region
1205721 1205722 12057231205720
(e) Angle histogram for scale marks detection (f) Results of the pointer and scale marksdetection
Figure 15 The detection results of the pointer gauge with an uneven scale
According to the gauge calibration regulations the exper-iment procedure is as follows First a positive stroke isconducted Here the float gaugersquos output is increased bygradually loading the weights while the gaugersquos indicatingvalue is increased At each indicating value test point thecamera takes images of both the standard pointer and testgauges The proposed approach is to use the images torecognize the indicating values Then a negative stroke isconducted of which the process is that the float gauge outputis reduced from the highest test point by gradually unloadingthe weights while the gaugersquos indicating value is accordinglyreduced An image of each test point is taken to recognize theindicating values
532 Experimental Results and Analysis The experimentalresults are shown in Figures 17 and 18 The gaugersquos upperlimit is 025MPa the lower limit is 0MPa Whatever theexperiment is in the positive or negative stroke the floatgauge output intervals are 0025MPa The detection resultsfor the indicating values of the standard and test gauges areshown in Tables 2 and 3 respectively
As can be seen from the experimental results the pro-posed approach is applicable to the pressure gauge ver-ification The indicating values of the test and standardgauges can be obtained by the proposed approach instead ofbeing obtained through manual observation The test gaugecalibration is completed by comparing the indicating value
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
14 Mathematical Problems in Engineering
(a) (b)
Figure 16 The experimental system of gauge test
Figure 17 The standard gauge results
Figure 18 The test gauge results
of the test gauge with that of the standard gauge From theexperimental results in Tables 2 3 and 4 comparing thevisual data with the manual reading we can see that thedetection results of the indicating value based on computervision are more accurate valid and have more significantdigits and a faster detection speed
54 Experimental Conclusion From the above experimentalresults we can draw the following conclusions
(1) The indicating value of the pointer gauge can beeffectively detected using themethod proposed in this
paper the approach can be used for pressure gaugeverification
(2) The approach proposes a stable frame about thereading identification of the pointer gauge Frombig region to small target the approach first locatesthe scale and the pointer region Then it preciselysegments the pointer and scale marks in the targetarea Finally the indicating value of the gauge isobtained based on the direction of the pointer andthe distribution of the scale marks This frameworkis a general recognition method of a pointer gaugebased on computer vision and has good interferenceimmunity However to achieve more satisfactoryresults in its practical application each step has to beimproved For example in the process of locating ofthe dial region based on the region growing methodmany factors need to be considered to determinewhether the use of this method should be terminated
(3) In its practical application the image preprocessingmethod should be used along with the proposedmethod so as to obtain more satisfactory resultsespecially for some industrial field applications
6 Conclusion
This study proposes a pointer gauge automatic readingapproach based on computer vision The approach aims atovercoming the defects in the current pointer gauge auto-matic reading approach and the application of pressure gaugeverification According to the framework of the proposedapproach which is conducted from big-region segmentationto small-target detection first the target region in the dialis located Then the pointer and scale marks are detected inthis region Finally the indicating gauge value is obtained Assuch the proposed approach has good interference immunityand is applicable for both gauges with and without evenlydistributed scale marks The approach can be widely appliedto the reading identification of different types of pointergauges
The proposed method is robust and precise enough to beused in pointer gauge verificationHowever in some practical
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 15
Table2Th
estand
ardgauged
etectio
nresults
Thep
ressureo
fcheckp
oint
(MPa)
Theind
icatingvalueo
fstand
ardgauge
Thec
itederror
(absolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0024343
0024
000
0657
0001
0025610
0025
000
0610
0026024
0050
004
8727
0049
0001273
0001
0049761
005
000
0240
0051010
0075
0074386
0075
000
0614
00075876
0076
000
0876
0001
025035
0100
01004
080099
000
0408
0001
0100147
01
000
0147
0016006
0125
0125060
0125
000
0060
00125482
0126
000
0482
0001
0024019
0150
0149762
015
000
0238
001506
470151
000
0647
0001
0095026
0175
0174613
0175
000
0387
00174720
0176
000
0280
0001
0155011
0200
0198316
02
0001684
00201540
0201
0001540
0001
0674062
0225
0225053
0224
000
0053
0001
0225149
0225
000
0149
0002006
0250
0249344
0249
000
0656
0001
0249344
0249
000
0660
0001
026026
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
16 Mathematical Problems in Engineering
Table3Th
etestgauge
detectionresults
Thep
ressure
output
ofthe
checkp
oint
(MPa)
Theind
icatingvalueo
ftestedgauge
Thec
itederrors
(thea
bsolutee
rrorrange)
Positives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Negatives
troke
Error1
(visu
aldetection-weights
value)
Error2
(manual
reading-weights
value)
Visual
detection
Manual
reading
Visual
detection
Manual
reading
00
00
00
00
00
0025
0025650
0026
000
0650
0001
0027084
0028
0002084
0003
026083
0050
0050200
0051
000
0200
0001
0051623
0052
0001623
0002
008065
0075
0075470
0076
000
0470
0001
0077778
0078
0002778
0003
019111
0100
0100100
0101
000
0100
0001
0101095
0102
0001095
0002
004
044
0125
01260
000126
0001000
0001
0126106
0127
000110
60002
04044
0150
0150900
0152
000
0900
0002
0151486
0153
0001486
0003
036059
0175
0177626
0173
0002626
0002
0177678
0179
0002678
000
410
510
70200
0202263
0203
0002263
0003
0202445
0204
0002445
000
40900978
0225
0226146
0227
000114
60002
0226223
0228
0001223
0003
046
049
0250
0249057
0252
000
0940
0002
0249057
0252
000
0940
0002
038038
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 17
Table4Com
paris
onsb
etweenmanualreading
andvisualdetection
Thep
ressureo
utpu
tof
thec
heckpo
int
(MPa)
Theind
icatingvalueo
fvisu
aldetection
Theind
icatingvalueo
fmanualreading
Com
paris
on(error
1minuserror2
)Po
sitives
troke
Error1
1205751minus1205752(M
Pa)
Relativ
eerror
1(1205751minus1205752)1205751(
)Po
sitives
troke
Error2
1205901minus1205902(M
Pa)
Relativ
eerror
2(1205901minus1205902)1205901(
)Standard
gauge
1205751(M
Pa)
Teste
dgauge
1205752(M
Pa)
Standard
gauge
1205901(M
Pa)
Teste
dgauge
1205902(M
Pa)
00
00
00
00
00
0025
0024343
0025650
0001310
53700
00024
0026
0002
8333
008202
0050
004
8727
0050200
0001473
302296
0049
0051
0002
4082
003533
0075
0074387
0075470
0001080
145644
0075
0076
0001
1333
001225
0100
01004
080100100
000
0308
030675
0099
0101
0002
202
002051
0125
0125060
01260
00000
0940
075164
0125
0126
0001
08
000706
0150
0149762
0150900
000114
0075987
015
0152
0002
1333
001219
0175
0174613
0177626
0003010
172553
0175
0173
0002
11429
00144
00200
0198316
0202263
0003950
199026
02
0203
0003
15001105
0225
0225053
0226146
0001090
048566
0224
0227
0003
1339
001230
0250
0249344
0249057
000
0287
011510
0249
0252
0003
1205
001233
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
18 Mathematical Problems in Engineering
applications especially in the industrial field it should beused by combining it with image preprocessing methodsso as to achieve better effects The improved approach canalso be used for the pointer gauge whose dial is noncircularTherefore in the future we will conduct the following studiesto complete the digitization work on the pointer gauge so asto develop a general indicating value readingmethod (1)Theimage preprocessing methods will be developed to eliminatethe influence of stray or reflective light or to correct the imagedistortion caused by changes in the relative position betweenthe camera and gauge (2) A general automatic readingmethod will be developed for different types of pointergauges which can be used in various hardware systems andhave a unified standard output format
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
The work is supported by Beijing Key Discipline Develop-ment Program (no XK100080537) the Beijing Natural Sci-ence Foundation (4122050) andOpening Project of Key Lab-oratory of Operation Safety Technology on Transport Vehi-cle Ministry of Transport China
References
[1] L Yanwen G Jinfeng Z Hongwei and R Hui ldquoDetection sys-tem of meter pointer based on computer visionrdquo in Proceedingsof the International Conference on Electronic and MechanicalEngineering and Information Technology (EMEIT rsquo11) pp 908ndash911 August 2011
[2] Z Jun F Haiping and K Ming ldquoAutomatic calibration ofdial gauges based on computer visionrdquo Infrared and LaserEngineering vol 37 pp 333ndash336 2008
[3] MKGellaboina G Swaminathan andVVenkoparao ldquoAnalogdial gauge reader for handheld devicesrdquo in Proceedings of theIEEE 8th Conference on Industrial Electronics and Applications(ICIEA rsquo13) pp 1147ndash1150 June 2013
[4] F C Alegria and A C Serra ldquoAutomatic calibration of analogand digital measuring instruments using computer visionrdquoIEEE Transactions on Instrumentation and Measurement vol49 no 1 pp 94ndash99 2000
[5] B Sun C Zhang Z Liu B Qian and H P Zhang ldquoStudy ofFPGA-based visual detection for pointer dialrdquo in Proceedings ofthe 26th Chinese Control and Decision Conference (CCDC 14)pp 1467ndash1472 Changsha China May-June 2014
[6] G Gui-fa W Ren-huang and W Huan ldquoA pretreatment ofimage based on reformative homomorphic filteringrdquo Journal ofGuangdong University of Technology vol 26 no 3 pp 57ndash592009
[7] D-C Luo S-C Wang H-G Zeng Z-Z Li and X-M LuldquoDesign of recognition system of analog measuring instru-mentsrdquo Laser amp Infrared vol 37 no 4 pp 377ndash380 2007
[8] H Wen Z Teng S Yang and S Liu ldquoIntelligent readingmethod for analog meter based on computer visionrdquo Chinese
Journal of Scientific Instrument vol 28 no 7 pp 1234ndash12392007
[9] W Qi T Xiling D Cheng H Yao and F Yanjun ldquoAutomaticalignment system based on center point recognition of analogmeasuring instruments dialrdquo in Proceedings of the 39th AnnualConference of the IEEE Industrial Electronics Society (IECONrsquo13) pp 5532ndash5536 IEEE Vienna Austria November 2013
[10] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceedings of the Chinese Society of Elec-trical Engineering vol 27 no 7 pp 470ndash475 2007
[11] Z LeThe research of indication recognition for dial instrumentsbased on image processing [MS thesis] North China ElectricPower University Beijing China 2013
[12] S-H Lu Study on automatic test system for high precision pointerinstrumeng based on machine vision [Dissertation for the MasterDegree in Engineering] Harbin Institute of Technology 2013
[13] Q Wang Y Fang W Wang M Wu R Wang and Y FangldquoResearch on automatic reading recognition of index instru-ments based on computer visionrdquo in Proceedings of the 3rdInternational Conference on Computer Science and NetworkTechnology (ICCSNT rsquo13) pp 10ndash13 October 2013
[14] Z Shutao Direct reading instrument calibration method basedon computer vision [PhD thesis] North China Electric PowerUniversity Beijing China 2006
[15] S W Wang and Y W Dai ldquoAutomatic recognizing of multi-pointerrsquo water meter using image processingrdquo Chinese Journalof Scientific Instrument vol 26 no 11 pp 1178ndash1120 2005
[16] F-J Sun T-J An J-Q Fan C-P Yang and Z Xu ldquoStudy on therecognition of pointer position of electric power transformertemperature meterrdquo Proceeding of the CSEE vol 27 no 7 pp70ndash78 2007
[17] B Hemming and H Lehto ldquoCalculation of uncertainty ofmeasurement in machine vision case a system for the cal-ibration of dial indicatorsrdquo in Proceedings of the 18th IEEEInstrumentation and Measurement Technology Conference pp665ndash670 Budapest Hungary May 2001
[18] B Chen and L Jin ldquoA new meter dial indicator detectionmethod based on center point projectionrdquo Application Researchof Computers vol 22 no 1 pp 246ndash248 2005
[19] J Zetao W Shi and L Kewei ldquoAn effective method onnon-contact measurement for gauges with pointerrdquo ComputerApplication and Software vol 26 no 4 pp 281ndash285 2009
[20] J-C Yin J Lou and H-T Zhao ldquoVision system and itsapplication in the auto door latch assembly linerdquo Techniques ofAutomation and Applications no 1 pp 79ndash82 2010
[21] H Zhou H-E Xu and C-G Geng ldquoAutomatic checking ofpointer automotive dashboard based on HIS model and Houghtransformationrdquo Journal of Zhejiang University (EngineeringScience) vol 44 no 6 pp 1108ndash1113 2010
[22] J Han E Li B Tao and M Lv ldquoReading recognition methodof analog measuring instruments based on improved houghtransformrdquo in Proceedings of the IEEE 10th International Con-ference on Electronic Measurement and Instruments (ICEMI rsquo11)pp 337ndash340 August 2011
[23] C Li L Yang Z Liu and K Li ldquoA new automatic seeded regiongrowing algorithmrdquo in Proceedings of the 6th InternationalCongress on Image and Signal Processing (CISP 13) vol 1 pp543ndash549 Hangzhou China December 2013
[24] SHengqiang andWChangji ldquoAnewalgorithmbased on super-green features for ostursquos method using image segmentationrdquo in
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 19
Proceedings of the World Automation Congress (WAC rsquo12) pp1ndash4 June 2012
[25] J Xu X Sun D Zhang and K Fu ldquoAutomatic detection ofinshore ships in high-resolution remote sensingimages usingrobust invariant generalized hough transformrdquo IEEEGeoscienceand Remote Sensing Letters vol 11 no 12 pp 2070ndash2074 2014
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of