fingerprint detection
DESCRIPTION
This is a complete report on Bio-metrics, finger print detection. It include what finger print is, how to scan and refin finger print, how the mechanism of its detection work, applications, etcTRANSCRIPT
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
ABSTRACT
“FINGERPRINT RECOGNITION SYSTEM” is a security facility provided to the
users which supports decision to make on the access rights to the authorized
users by authentication.This project extrapolates the necessary fingerprint
data verification and enrollment and requires the users to make decisions
taking high quality image, more responsibility, and accountability and making
comparisons on the ridge patterns of the fingerprints of the users. This
project is based on the fact that each person has a unique pattern of the
fingerprint that differentiates him from others. Fingerprint recognition is a
biometric technique for personal identification. Biometrics based fingerprint
recognition provides one of the promising solutions for the security of the
software and the domain of applying this techniques for security is increasing
day by day. Biometric features also include speech, handwriting, face
identification etc. Face identification is one of the popular techniques for
personal identification, but may fail in certain situations where two people
look very similar. Even the speech and handwriting recognition systems may
fail in certain situations, Fingerprints’ being complex patterns has the
advantage of being a passive, noninvasive system for personal identification
and its success depends on solving the two problems:
Representation of the complex patterns of the fingerprints and
Matching these fingerprint patterns.
This project uses both, algebraic and geometric features to representation
fingerprint images. Here we divide both the existing finger print in the
database and the scanned finger print into frames and compare the pixel
values of the same and the user is authenticated based on the percentage of
values being compared. The constraints of the percentage of fingerprint being
matched can me modified as needed and hence the authentication can be
made as strict as possible based on the criticality of its application.
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
TABLE OF CONTENTS
Chapter no TITLE PAGE NO
ABSTRACT
1. INTRODUCTION 5
1.1 BIOMETRIC SYSTEMS 5
1.1.1 What is biometric system? 5
1.1.2 Working of a Biometric System 6
1.1.3 Issues have to be address. 7
1.1.4 The most common biometrics. 8
1.1.5 What is Fingerprint? 12
2. FINGERPRINT DECTION 13
2.1 Finger Print Detection. 13
2.2 PRE-PROCSSING OF IMAGES 13
2.2.1 Binarization 14
2.2.2 THINNING: 14
2.2.2.1 Erosion: 15
2.2.2.2 Dilation: 15
2.2.3 Final Noise Removal 18
2.3 MINUTAE MATCHING
18
2.3.1 CHARACTERSTICS: 18
2.3.2 MINUTAE EXTRACTION 19
2.3.3 FINDING A RIDGE SUMMIT POINT: 19
2.3.3.1 TRACING A RIDGE:
20 2.4 PATTERN MATCHING
21
2.5 FINGERPRINT MATCHING & AUTHENTICATION 22
3 CONCLUSION 24
3.1 ADVANTAGES & DISADVANTAGES 24
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
3.1.1 DISADVANTAGES OF USING FINGERPRINT 24
3.1.2 ADVANTAGES OF USING FINGERPRINT 24
3.2 APPLICATIONS 25
4 BIBLIOGRAPHY 26
LIST OF FIGURES
S.NO TITLE PAGE NO
1. Some of the biometrics 5
2. Biometric Market Report in the year 2002. 10
3. fingerprint 11
4. Effect of Binarization 13
5. Figure: Effect of Dilation 14
6. Figure: Effect of Block filter 15
7. Noise removal 15
8. Combined image of both the images 15
9. Crossing over 1 16
10. Crossing over 2 16
11. Crossing over 3 17
12. Figure: Effect of Spurs 17
13. Thinned image from block filtering 17
14. Impact of deleting short island segments 18
15. Figure: Ridge endings 18
16. Figure: Ridge bifurcation 18
17. Figure: Ridg ridges 19
18. Figure: Ridge enclosures 19
19. Figure: minutia attributes 19
20. Figure: RIDGE SUMMIT POINT: 20
21. Figure: Ridge tracing 21
22. Process of identification 23
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
Chapter1
INTRODUCTION-
1.1 BIOMETRIC SYSTEMS
1.1.1 What is biometric system ?
A biometric system is essentially a pattern recognition system that
recognizes a person by determining the authenticity of a specific
physiological and/or behavioral characteristic possessed by that person. An
important issue in designing a practical biometric system is to determine
how an individual is recognized. Depending on the application context, a
biometric system may be called either a verification system or an
identification system:·
A verification system authenticates a person’s identity by comparing the
captured biometric characteristic with her own biometric template(s) pre-
stored in the system. It conducts one-to-one comparison to determine
whether the identity claimed by the individual is true. A verification system
either rejects or accepts the submitted claim of identity (Am I whom I claim I
am?);·
An identification system recognizes an individual by searching the entire
template atabasefor a match. It conducts one-to-many comparisons to
establish the identity of the individual. In an identification system, the system
establishes a subject’s identity (or fails if the subject is not enrolled in
thesystem database) without the subject having to claim an identity (Who am
I?).
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
Some of the biometrics are: a)ear, b)face, c)facial thermo gram, d)hand
thermo gram, e)hand vein, f)hand geometry, g)fingerprint, h)iris, i)retina,
j)signature, k)voice.
1.1.2 Working of a Biometric System
The term authentication is also frequently used in the biometric field,
sometimes as a synonym for verification; actually, in the information
technology language, authenticating a user means to let the system know the
user identity regardless of the mode (verification or identification). The
enrollment module is responsible for registering individuals in the biometric
system database (system DB). During the enrollment phase, the biometric
characteristic of an individual is first scanned by a biometric reader to
produce a raw digital representation of the characteristic. A quality check is
generally performed to ensure that the acquired sample can be reliably
processed by successive stages. In order to facilitate matching, the raw digital
representation is usually further processed by a feature extractor to generate
a compact but expressive representation, called a template. Depending on the
application, the template may be stored in the central database of the
biometric system or be recorded on a magnetic card or smartcard issued to
the individual.
The verification task is responsible for verifying individuals at the point of
access. During the operation phase, the user’s name or PIN (Personal
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
Identification Number) is entered through a keyboard (or a keypad); the
biometric reader captures the characteristic of the individual to be
recognized and converts it to a digital format, which is further processed by
the feature extractor to produce a compact digital representation. The
resulting representation is fed to the feature matcher, which compares it
against the template of a single user (retrieved from the system DB based on
the user’s PIN). In the identification task, no PIN is provided and the system
compares the representation of the input biometric against the templates of
all the users in the system database; the output is either the identity of an
enrolled user or an alert message such as “user not identified.” Because
identification in large databases is computationally expensive, classification
and indexing techniques are often deployed to limit the number of templates
that have to be matched against the input.
1.1.3 When choosing a biometric for an application the following issues
have to be address.
Does the application need verification or identification? If an application
requires an identification of a subject from a large database, it needs a
scalable and relatively more distinctive biometric (e.g., fingerprint, iris, or
DNA).
What are the operational modes of the application? For example, whether
the application is attended (semi-automatic) or unattended (fully
automatic), whether the users are habituated (or willing to be habituated)
to the given biometrics, whether the application is covert or overt,
whether subjects are cooperative or non-cooperative, and so on.
What is the storage requirement of the application? For example, an
application that performs the recognition at a remote server may require
a small template size.
How stringent are the performance requirements? For example, an
application that demands very high accuracy needs a more distinctive
biometric.
What types of biometrics are acceptable to the users? Different biometrics
are acceptable in applications deployed in different demographics
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
depending on the cultural, ethical, social, religious, and hygienic standards
of that society.
1.1.4 The most common biometrics.
Ear: It is known that the shape of the ear and the structure of the
cartilaginous tissue of the pinna are distinctive. The features of an ear are
not expected to be unique to an individual. The ear recognition
approaches are based on matching the distance of salient points on the
pinna from a landmark location on the ear.
Face: The face is one of the most acceptable biometrics because it is one
of the most common methods of recognition that humans use in their
visual interactions. In addition, the method of acquiring face images is
nonintrusive. Facial disguise is of concern in unattended recognition
applications. It is very challenging to develop face recognition techniques
that can tolerate the effects of aging, facial expressions, slight variations in
the imaging environment, and variations in the pose of the face with
respect to the camera.
Facial, hand, and hand vein infrared thermograms: The pattern of heat
radiated by the human body is a characteristic of each individual body
and can be captured by an infrared camera in an unobtrusive way much
like a regular (visible spectrum) photograph. The technology could be
used for covert recognition and could distinguish between identical twins.
A thermogrambased system is non-contact and non-invasive but sensing
challenges in uncontrolled environments, where heat-emanating surfaces
in the vicinity of the body, such as, room heaters and vehicle exhaust
pipes, may drastically affect the image acquisition phase. A related
technology using near infrared imaging is used to scan the back of a
clenched fist to determine hand vein structure. Infrared sensors are
prohibitively expensive which a factor inhibiting widespread use of the
thermograms.
Hand and finger geometry: Some features related to a human hand (e.g.,
length of fingers) are relatively invariant and peculiar (although not very
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
distinctive) to an individual. The image acquisition system requires
cooperation of the subject and captures frontal and side view images of
the palm flatly placed on a panel with outstretched fingers. The
representational requirements of the hand are very small (nine bytes in
one of the commercially available products), which is an attractive feature
for bandwidth- and memory-limited systems. Due to its limited
distinctiveness, hand geometry-based systems are typically used for
verification and do not scale well for identification applications. Finger
geometry systems (which measure the geometry of only one or two
fingers) may be preferred because of their compact size.
Iris: Visual texture of the human iris is determined by the chaotic
morphogenetic processes during embryonic development and is posited
to be distinctive for each person and each eye. An iris image is typically
captured using a non-contact imaging process. Capturing an iris image
involves cooperation from the user, both to register the image of iris in
the central imaging area and to ensure that the iris is at a predetermined
distance from the focal plane of the camera. The iris recognition
technology is believed to be extremely accurate and fast.
Retinal scan: The retinal vasculature is rich in structure and is supposed
to be a characteristic of each individual and each eye. It is claimed to be
the most secure biometric since it is not easy to change or replicate the
retinal vasculature. The image capture requires a person to peep into an
eyepiece and focus on a specific spot in the visual field so that a
predetermined part of the retinal vasculature may be imaged. The image
acquisition involves cooperation of the subject, entails contact with the
eyepiece, and requires a conscious effort on the part of the user. All these
factors adversely affect public acceptability of retinal biometrics. Retinal
vasculature can reveal some medical conditions (e.g., hypertension),
which is another factor standing in the way of public acceptance of retinal
scan-based biometrics.
Signature: The way a person signs his name is known to be a
characteristic of that individual. Although signatures require contact and
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
effort with the writing instrument, they seem to be acceptable in many
government, legal, and commercial transactions as a method of
verification. Signatures are a behavioral biometric that change over a
period of time and are influenced by physical and emotional conditions of
the signatories. Signatures of some people vary a lot: even successive
impressions of their signature are significantly different. Furthermore,
professional forgers can reproduce signatures to fool the unskilled eye.
Voice: Voice capture is unobtrusive and voice print is an acceptable
biometric in almost all societies. Voice may be the only feasible biometric
in applications requiring person recognition over a telephone. Voice is not
expected to be sufficiently distinctive to permit identification of an
individual from a large database of identities. Moreover, a voice signal
available for recognition is typically degraded in quality by the
microphone, communication channel, and digitizer characteristics. Voice
is also affected by a person’s health (e.g., cold), stress, emotions, and so
on. Besides, some people seem to be extraordinarily skilled in mimicking
others.
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
Table comparing various biometric technologies High, Medium, and
Low are denoted by H, M, and L, respectively.
Biometric Market Report (International Biometric Group) estimated the
revenue of various biometrics in the year 2002.
1.1.5 What is Fingerprint?
A fingerprint is a textural image containing a large number of ridges that
form groups of almost parallel curves. It has been established that
fingerprint's ridges are individually unique and are unlikely to change during
the whole life.
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
Chapter 2
FINGERPRINT DECTION
2.1 Finger Print Detection.
The use of fingerprints as a biometric is both the oldest mode of computer-
aided, personal identification and the most prevalent in use today. However,
this widespread use of fingerprints has been and still is largely for law
enforcement applications. There is expectation that a recent combination of
factors will favor the use of fingerprints for the much larger market of
personal authentication. These factors include: small and inexpensive
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
fingerprint capture devices, fast computing hardware, recognition rate and
speed to meet the needs of many applications, the explosive growth of
network and Internet transactions, and the heightened awareness of the need
for ease-of-use as an essential component of reliable security. This method
has been widely used in criminal identification, access authority verification,
financial transferring confirmation, and many other civilian applications. In
the old days, fingerprint recognition was done manually by professional
experts. But this task has become more difficult and time consuming.
2.2 PRE-PROCSSING OF IMAGES
Following image capture to obtain the fingerprint image, image processing is
performed. The ultimate objective of image processing is to achieve the best
image by
which to produce the correct match result.
2.2.1 Binarization
Image binarization is the process of turning a grayscale image to a
black and white image.
In a gray-scale image, a pixel can take on 256 different intensity values
while each pixel is assigned to be either black or white in a black and
white image.
This conversion from gray-scale to black and white is performed by
applying a
threshold value to the image.
A critical component in the binarization process is choosing a correct
value for the threshold. The threshold values used in this study were
selected empirically by trial and error.
Steps Of Pre-Processing
THINNINGBINARIZATION NOISE REMOVAL
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
Figure: Effect of Binarization
2.2.2 THINNING:
This thinning method to be done with Block Filtering method attempts to
preserve the outermost pixels along each ridge
This is done with the following steps:
Step One: ridge width reduction
This step involves applying a morphological process to the image to reduce
the width of the ridges. Morphological is a means of changing a stem to adjust
its meaning to fit its syntactic and communicational context Two basic
morphological processes are
Erosion
Dilation
2.2.2.1 Erosion:
Erosion thins objects in a binary image (ridge)In this project we are
using the
2.2.2.2 Dilation:
A dilation process is used to thicken the area of the valleys in the
fingerprint.
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
Original Gray Image After
Level Image Dilation
Figure: Effect of Dilation
Step Two: passage of block filter
The next step involves performing a pixel-by pixel scan for black pixels across
the entire image Note that in MATLAB, image rows are numbered in
increasing order beginning with the very top of the image as row one.
Similarly, columns are numbered in increasing order beginning with the
leftmost side of the left to right scan continues until it covers the entire
image. Next, a similar scan is performed across the image from right to left
beginning at the pixel in row one and the last column.
Original image Image after block filter
Figure: Effect of Block filter
Step Three: removal of isolated noise
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
Image with noise Image after noise removal
Step Four: scan combination
A value of two means that the pixel from each scan was white, while a value
of zero indicates the pixel from each scan was black. Meanwhile, a value of
one means that the pixel from one scan was black while the same pixel from
the other scan was white. As a result, the new matrix needs to be adjusted to
represent a valid binary image containing only zeros and ones. Specifically,
all zeros and ones are assigned a value of zero (black pixel),
Combined image of both the images
Step Five: elimination of one pixel from two-by-two squares of black
Next, a new scan is conducted on the combined image to detect two-by-two
blocks of black pixels which represent a location where a ridge has not been
thinned to a one-pixel width. It is likely that some of these two-by two blocks
were created by the combination of the previous scans. This problem can be
compensated for by changing one pixel within the block from black to white,
which reduces the width at that particular point from two pixels to one. At
the same time, This operation can be performed by analyzing the pixels
touching each individual black pixel. Note that each black pixel touches the
three other black pixels within the two-by-two block. Therefore, there are
only five other pixels that contain useful information.
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
Step Six: removal of unwanted spurs
Crossing over 1
Crossing over 2
Crossing over 3
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
With spurs after its removal
Figure: Effect of Spurs
Step Seven: removal of duplicate horizontal and duplicate vertical lines
Thinned image from block filtering
2.2.3 Final Noise Removal
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
Impact of deleting short island segments
2.3 MINUTAE MATCHING
2.3.1 CHARACTERSTICS:
A fingerprint is a textural image containing a large number of ridges that
form groups of almost parallel curves. It has been established that
fingerprint's ridges are individually unique and are unlikely to change during
the whole life. Although the structure of ridges in a fingerprint is fairly
complex, it is well known that a fingerprint can be identified by its special
features such as:
Ridge endings: The ending of the ridges takes place at the middle as shown.
Ridge bifurcation : The division of the ridges in the middle as shown
S hort ridges : The small lines present in between two ridges as shown
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
Ridge enclosures: These are the loops formed as shown in fig(c).
2.3.2 MINUTAE EXTRACTION
For convenience, we represent a fingerprint image in reverse gray scale. That
is, the dark pixels of the ridges are assigned high values where as the light
pixels of the valleys are given low values. In a fingerprint, each minutia is
represented by its location (x, y) and the local ridge direction Figure 4 shows
the attributes of a fingerprint's minutia. The process of minutiae detection
starts with finding a summit point on a ridge, and then continues by tracing
the ridge until a minutia, which can be either a ridge ending or bifurcation, is
encountered.
2.3.3 FINDING A RIDGE SUMMIT POINT:
To find a summit point on a ridge, we start from a point x = (x1, x2) and
compute the direction angle by using the gradient method. Then theφ
vertical section orthogonal to the direction is constructed. The point in this
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
section with maximum gray level is a summit point on the nearest ridge. The
direction angle at a point φ x
mentioned above is computed as follows. A 9×9 neighborhood around x is
used to determine the trend of gray level change. At each pixel u = (u1, u2) in
this neighborhood, a gradient vector v(u) = (v1(u), v2(u)) is obtained by
applying the operator h = (h1, h2) with
to the gray levels in a neighborhood of u. That is,
Where y runs over the eight neighboring pixels around u and g(y) is the gray
level of pixel y in the image. The angle represents the direction of the unit
vector t that is (almost) orthogonal to all gradient vectors v. That is, t is
chosen so that is minimum.
2.3.3.1 TRACING A RIDGE:
The task of tracing a ridgeline to detect minutiae is described in the following
algorithm. This algorithm also constructs a traced image of the fingerprint.
Every time a new summit point of the ridge is found, its location in the traced
image is assigned a high gray value and the surrounding pixels are given
lower gray levels if they have not been marked.
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
Algorithm 1 (Ridge tracing):
Start from a summit point x of a ridge.
Repeat
Compute the direction angle at x;
Move pixels from ∝ x along the direction to another point y;
Find the next summit point z on the ridge, which is the local maximum of the
section orthogonal to direction at point y; Set x = z;
Until point x is a termination point (i.e. a minutia or off valid area).
Determine if the termination point x is a valid minutia, if so record it.
End Algorithm 1
2.4 PATTERN MATCHING
The more macroscopic approach to matching is called global pattern
matching or simply pattern matching. In this approach, the flow of ridges is
compared at all locations between a pair of fingerprint images. The ridge flow
constitutes a global pattern of the fingerprint. Three fingerprint patterns are
shown in Figure (Different classification schemes can use up to ten or so
pattern classes, but these three are the basic patterns.) Two other features
are sometimes used for matching: core and delta. (Figure) The core can be
thought of as the center of the fingerprint pattern. The delta is a singular
point from which three patterns deviate. The core and delta locations can be
used as landmark locations by which to orient two fingerprints for
subsequent matching – though these features are not present on all
fingerprints. There may be other features of the fingerprint that are used in
matching. For instance, pores can be resolved by some fingerprint sensors
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
and there is a body of work (mainly research at this time) to use the position
of the pores for matching in the same manner that the minutiae are used. Size
of the fingerprint, and average ridge and valley widths can be used for
matching, however these are changeable over time. The positions of scars
and creases can also be used, but are usually not used because they can be
temporary or artificially introduced.
2.5 FINGERPRINT MATCHING AND AUTHENTICATION
Reliably matching fingerprint images is an extremely difficult problem, mainly
due to the large variability in different impressions of the same finger (i.e., large
intra-class variations). The main factors responsible for the intra-class variations
are: displacement, rotation, partial overlap, non-linear distortion, variable
pressure, changing skin condition, noise, and feature extraction errors. Therefore,
fingerprints from the same finger may sometimes look quite different whereas
fingerprints from different fingers may appear quite similar (see Figure 1.14).
Difficulty in fingerprint matching:
Fingerprint look different to an untrained eye but they are impressions of
the same finger.
Fingerprint look similar to an untrained eye but they are from different
fingers.
Human fingerprint examiners, in order to claim that two fingerprints are from the
same finger, evaluate several factors:
i) global pattern configuration agreement, which means that two fingerprints must
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
be of the same type,
ii) qualitative concordance, which requires that the corresponding minute details
must be identical,
iii) quantitative factor, which specifies that at least a certain number (a minimum
of 12 according to the forensic guidelines in the United States) of corresponding
minute details must be found, and
iv) corresponding minute details, which must be identically inter-related. In
practice, complex protocols have been defined for fingerprint matching and a
detailed flowchart is available to guide fingerprint examiners in manually
performing fingerprint matching.
Given below is a figure showing the general method by which fingerprints are
matched. Figure showing a general method as to how the finger print is
matched and compared with an existing fingerprint from the database.
Process of identification
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
CHAPTER 3
CONCLUSION
3.1 ADVANTAGES & DISADVANTAGES OF USING FINGERPRINT
3.1.1 DISADVANTAGES OF USING FINGERPRINT
There are some problems in collecting the second database:
The different age of the persons leads to a different size of the fingerprint
Some of the twins are children so there are scratches in the fingerprints
Some of them did not fully cooperate with the researchers, so most of the
images of their fingerprints do not contain enough features to create an
extraction.
3.1.2 ADVANTAGES OF USING FINGERPRINT
Prevents unauthorized use or access
Adds a higher level of security to an identification process
Eliminates the burden and bulk of carrying ID cards or remembering Pins
Heightens overall confidence of business processes dependent on personal
identification.
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
3.2 APPLICATIONS
Criminal identification
Prison security
ATM
Aviation security
Border crossing controls
Database access
Door-Lock System
Safe Box
Simple Access Controller
Vehicle Control
BIOMETRIC SYSTEM : FRINGUREPRINT DETECTION
BIBLIOGRAPHY
i. www.biometix.com
ii. www.biomet.org
iii. www.owlinvestigation.com
iv. www.ddl.ision.co.uk
v. www.zdnetindia.com/techzone/resources
vi. www.biodata.com.au
vii. http://en.wikipedia.org/wiki/1-Wire
viii. http://fingereprint.blogspot.com/
ix. Baruch,O.Following", Pattern Recognition Letters, Vol. 8 No. 4, 1988, pp.
271-276.
x. Nist image group’s fingerprint research.
http://www.itl.nist.gov/iad/894.03/fing/fing.html. [Online; accessed
25-February-2010].
xi. Fvc2006 the fourth international fingerprint verification competition.
http://bias.csr.unibo.it/fvc2006/results/O res db2 a.asp. [Online; accessed
25-February-2010].
xii. Fvc2004 the third international fingerprint verification competition.
http://bias.csr.unibo.it/fvc2004/results.asp. [Online; accessed 25-
February-2010].
xiii. S.A. Niyogi and E.H. Adelson. Analyzing and recognizing walking
figures in xyt. CVPR, 94:469–474.