1 image pre-processing. 2 digital image processing the process of extracting information from...
TRANSCRIPT
1
Image Pre-Processing
2
Digital Image Processing
• The process of extracting information from digital images obtained from satellites
• Information regarding each pixel is fed into an algorithm and the result of the computation stored for that pixel
• Thus for each image being processed by a particular algorithm there is an input and output image
• Order of processing is important
3
The basic processes• Pre-processing- this lecture
1. Image rectification (geometric correction)
2. Radiometric correction (includes noise removal, DN-to-radiance conversion)
3. Atmospheric correction
• Processing
Image enhancement – contrast enhancement and image filtering (may be only visual)
Image classification
Data merging/data fusion
4
1. Geometric correction• Various geometric distortions:
Random– Variations in altitude, attitude and velocity of the sensor
platform– Atmospheric refraction– Relief displacement– Variable speed of scanning mirrorSystematic– Panoramic distortion– Skew distortion due to earth rotation during sweep of IFOV)– Earth curvature – orbit variation due to ellipsoid
Output is a geometrically accurate image, registered to a ground coordinate system - georeferenced
5
Systematic distortions• Panoramic Distortion
– The ground area imaged is proportional to the tangent of the scan angle rather than to the angle itself. Because data are sampled at regular intervals, this produces along-scan distortion.
• Skew Distortion– Earth rotates as the sensor scans the terrain. This results in a
shift of the ground swath being scanned, causing along-scan distortion.
– deskewing involves offsetting each scan line successively to west
– Skewed parallellogram appearance of images
Change in scale at edge of scan (tangential distortion)
6
Correction of distortions1. Most systematic distortions corrected at
ground station2. Most random distortions are corrected
by analysing GCPs in the image to register the image to the ground co-ordinate system (geo-referencing, registering)
7
Geometric correction using ground control points
•Uses least squares regression •Sum of squared difference between image and true coordinates minimised•Find four least squared coefficients
map x coordinate as function of image c and rmap y coordinate as function of image c and rimage c coordinate as function of map x and yimage r coordinate as function of map x and y
Then: x1 = a0 + a1c1 + a2r1 where a is the coefficient
8
Resampling
• Process of resampling: which cell values to use?– nearest neighbour– bilinear interpolation
(distance weighted average of 4 nearest pixels)
9
2. Radiometric correction– Need to calibrate data radiometrically due to:-
(i) Geometric and atmospheric effects• Scene illumination (time of day, season)• Viewing geometry• Relative position of sensor and illumination
(ii) System calibration effects• systematic differences in the digital numbers eg. striping• conversion of ground radiance to DNs due to differential
sensitivity of detectors to different wavebands• different sensors convert differently to byte scale• Effects of System noise on pixel values
10
Radiometric correction: effects of seasonal change
11
A form of radiometric correction is the conversion of the digital numbers to absolute radiance values
DN-to-Radiance conversion
eg. for LANDSAT L=((Lmax-Lmin)/QCalmax-QCalmin)*(QCal-QCalmin) + Lmin
12
Noise removal
• Noise is the unwanted disturbance in an image that is due to limitations in the sensing, digitisation or data recording process
• The effects of noise range from a degradation to total masking of the true radiometric information content of the digital image
13
Noise removal
• Critical to the subsequent processing and classification of an image
• Done to produce an image that is as close to the original radiometry of the scene as possible
• Noise may either be systematic (banding of multispectral images) to dropped lines or parts of lines
14
The use of moving
windows to average out
random noise
15
Algorithm for removal of random AND systematic noise
16
Stripe noise
• Sixteen-line frequency noise in a LANDSAT TM band 2 – Sumatra coastline
17
Image after scan-line noise
removal
18
Line drop
• Dropped line removed by averaging pixels each side of the line
19
3. Atmospheric correctionThe effects of the atmosphere include• reduction in the amount of energy reaching the ground by absorption
and scattering• increasing the amount of energy reaching the sensor by scattering
the radiation (diffuse radiation)• decrease in thermal due to w. vapour absorption
Atmospheric correction done by• empirical methods• dark pixel method
20
Dark pixel method: eg. for NIR band
21
SPOT images (SPOTs 1-5)
• Spot images available is range of pre-processed levels:– 1A– 1B– 2A– 2B– Ortho
22
Upper air data for empirical correction
Pw representswater vapour
http://envf.ust.hk/dataview/profile/current/
23
Level 1A
• Raw image
• Detector normalisation for each band
• Least amount of processing
• Panoramic effect due to scale change• Cost HK$23,300 XS; HK$29,500 Pan
• Use Radiometric studies, stereoplotting
24
Level 1B
• Radiometric corrections as for 1A• Geometric corrections
• Panoramic effect• Earth rotation and curvature• Orbit altitude variation w.r.t. reference ellipsoid
Cost: HK$23,300 XS; HK$29,500 PanUse: Interpretation, thematic studies,
stereoplotting
25
Level 2A
• Corrections Rectified to given projection and rotated to North
from satellite data
• Accuracy 500m planimetric
• UsesLow accuracy cartographic
• Cost HK$27,500 XS; HK$33,000 Pan
26
Level 2B• Corrections Rectified to control from either
maps or survey. Still has relief displacement
• Accuracy Absolute to 20m RMS
• Use High accuracy cartographic studies, SPOTView products
• Cost HK$42,000 both
Level 2B with relief correction from DTM
Ortho