ONLINE FINGERPRINT VERIFICATION

1
ONLINE FINGERPRINT VERIFICATION

• Sharat Chikkerur
• Center for Unified Biometrics and Sensors
• University at Buffalo
• www.cubs.buffalo.edu
• Advisor A. N. Cartwright
• Committee V. Govindaraju, A. H. Titus, L. Kondi

2
Abstract

• Background
• Traditional password/token based authentication
  schemes are insecure and are being replaced by
  biometric authentication mechanisms
• Fingerprints were one of the first biometrics to
  be widely used
• Despite 40 years of research, fingerprint
  recognition is still an open problem.
• Challenges
• Feature extraction is very unreliable in poor
  quality prints
• Matching fingerprints under non linear distortion
  is difficult
• Contributions
• New fingerprint image enhancement using STFT
  analysis.
• New feature extraction algorithm based on chain
  code contours
• New graph based matching algorithm robust to non
  linear distortion

3
Outline

• Introduction
• Biometrics
• Fingerprints 101
• Fingerprint Image Enhancement
• Minutia Feature Extraction
• Matching Algorithm
• Conclusion
• Software Demos

4
Biometrics

• Definition
• Biometrics is the science of verifying and
  establishing the identity of an individual
  through physiological features or behavioral
  traits
• Examples
• Physical Biometrics
• Fingerprint, Hand Geometry, Face
• Measurement Biometric
• Dependent on environment/interaction
• Behavioral Biometrics
• Handwriting, Signature, Speech, Gait
• Performance/Temporal biometric
• Dependent on state of mind
• Chemical/Biological Biometrics
• Skin spectroscopy
• DNA, blood-glucose

5
Fingerprints as a Biometric

• High Universality
• A majority of the population (gt96) have legible
  fingerprints
• More than the number of people who possess
  passports, license and IDs
• High Distinctiveness
• Even identical twins have different fingerprints
  (most biometrics fail)
• Individuality of fingerprints established through
  empirical evidence
• High Permanence
• Fingerprints are formed in the fetal stage and
  remain structurally unchanged through out life.
• High Performance
• One of the most accurate forms of biometrics
  available
• Best trade off between convenience and security
• High Acceptability
• Fingerprint acquisition is non intrusive.
  Requires no training.

6
Fingerprints 101 Fingerprint Classes

• A fingerprint is made up of system of oriented
  friction ridges
• A fingerprint can be classified based on type the
  ridge flow pattern

• Classification helps in narrowing down possible
  matches
• In reality, the class distribution is skewed
  (gt65 are loops)
• Used only in law enforcement applications

7
Fingerprints 101 Ridge Characteristics

• Fingerprints can be distinguished based on the
  ridge characteristics
• Ridge characteristics mark local discontinuities
  in the ridge flow
• No two individuals have the same pattern of
  ridge characteristics at the same relative
  locations

Global Features
Local Features
8
Prior Related Work Matching Paradigms

• Manual
• Human experts use a combination of visual,
  textural, minutiae cues and experience for
  verification
• Still used in the final stages of law enforcement
  applications
• Image based
• Utilizes only visual appearance.
• Requires the complete image to be stored (large
  template sizes)
• Texture based
• Treats the fingerprint as an oriented texture
  image
• Less accurate than minutiae based matchers since
  most regions in the fingerprints carry low
  textural content
• Minutiae based
• Uses the relative position of the minutiae points
• The most popular and accurate approach for
  verification
• Resembles manual approach very closely.

9
Image Based Matching Optical Correlation

• Advantages
• Image itself is used as the template
• Requires only low resolution images
• Optical correlation makes it extremely fast
  (Choudary and Awwal 99, Lee et al. 99, Roberge
  et al. 99, Baze et al.00)
• Disadvantages
• Image itself is used as the template (template
  size about 30 KB)
• Requires accurate alignment of the two prints
  (unreliable in poor prints)
• Not robust to changes in scale, orientation and
  position.

10
Texture Based Matching Filterbanks

• Advantages
• Uses texture information (lost in optical and
  minutiae based schemes)
• Performs well with poor quality prints
• Features are statistically independent from
  minutiae and can be combined with minutiae
  matchers for higher accuracy (Jain et al. 00,
  Jain et al 01)
• Disadvantages
• Requires accurate alignment of the two prints
  (unreliable in poor prints)
• Not invariant to translation, orientation and
  non-linear distortion.
• Less Accurate than minutiae based matchers

11
Minutiae Based Matching

• Advantages
• Invariant to translation, rotation and scale
  changes
• Very accurate (Ratha et al 96, Jain et al. 97,
  Jian Yau 00, Bazen and Garez 03)
• Disadvantages
• Minutiae extraction is error prone is low quality
  images
• Not robust to non-linear distortion.
• Does not use visual and textural cues

12
General Architecture
13
Outline

• Introduction
• Fingerprint Image Enhancement
• Need for Enhancement
• Prior Related Work
• Proposed Algorithm STFT Analysis
• Experimental Evaluation
• Minutia Feature Extraction
• Matching Algorithm
• Software Demos

14
Need for Enhancement
What you see
What you think you see
15
Reality What you usually get..
High contrast print
Typical dry print
Faint print
Low contrast print
Typical Wet Print
Creases
16
Challenges

• Challenges
• Fingerprint image is non stationary (has dominant
  local orientation and frequency)
• General purpose image processing algorithms are
  not useful
• Traditional operators and filters assume Gaussian
  noise model
• Noise in fingerprint images consists mostly of
  ridge breaks
• Contextual Filters
• Existing techniques are based on contextual
  filtering
• Filter parameters are adapted to each local
  neighborhood
• Filter parameters in unrecoverable regions can
  be interpolated based on its neighbors

17
Prior Related Work Spatial Filtering

• (Yang et.al 1996, Greenberg et. Al 1999) proposed
  local anisotropic filtering
• Filter kernel adapts at each pixel location

• Hong et al, 96/98 proposed the use of Gabor
  filters for enhancement
• Gabor filter has the best joint space-frequency
  localization
• Does not handle high curvature regions well due
  to block wise approach.

Even Symmetric Kernel
Fourier spectrum showing the localization
18
Prior Related Work Fourier Domain Filtering

• Sherlock et al 94, proposed the use of Fourier
  domain filtering
• The image is convolved with a filter bank of
  directionally selective filters
• Image enhanced by selecting a linear combination
  of filter responses
• Watson et al. 94, proposed the use or root
  filtering for enhancement.(Pseudo matched
  filter)
• Does not require the computation of orientation
  images

Root Filtering
Fourier Domain Filtering
19
Traditional Approaches
Local Orientation ?(x,y) Gradient Method
Enhancement Frequency/Spatial
Local Ridge Spacing F(x,y) Projection Based Method
Ratha et al 95
20
Proposed Approach Overview
Region Mask
STFT Analysis
Frequency Image
Fourier domain Enhancement
Orientation Image
Coherence Image
21
STFT Analysis

• Fingerprint image is non stationary, so we
  require both space and frequency resolution time
  frequency analysis
• STFT in 1D
• STFT in 2D

22
Surface Wave Model
Fingerprint ridges can be modeled as an oriented
wave
Surface wave
Local Neighborhoods
Validity of the model
23
Parameter Estimation

• Paradigm The Fourier domain response can be
  viewed as a distribution of surface waves. Each
  term F(r, ?) corresponds to a surface wave of
  frequency 1/r and orientation ?
• We seek to find the most likely surface wave and
  hence estimate the dominant direction and
  frequency
• We can represent the Fourier spectrum in polar
  form as F(r,?) The power spectrum is reduced to a
  joint probability density function using
• The angular and frequency densities are given by
  marginal density functions

24
Ridge Orientation Image
25
Region Mask

• The surface wave approximation does not hold in
  the background region
• The region mask is obtained by simple
  thresholding of the block energy image

26
Frequency Image
Jain et al 00
27
Coherence Image

• Block processing is unreliable in regions of
  high curvature
• Sherlock and Monro 94, relax filter parameters
  near the singular locations
• Estimation of singular point is difficult in
  poor images!
• We use an angular coherence measure proposed by
  Rao and Jain 90

28
Enhancement
29
Additional Enhancement Results
30
Qualitative Comparison
Root Filtering
Original Image
31
Qualitative Comparison(cont.)
Gabor Filter based Enhancement
Proposed Approach
32
Objective Evaluation

• We evaluated the effect of enhancement on 800
  images from FVC2002 DB3
• The evaluation consists of 2800 genuine test and
  4950 impostor tests
• It can be seen that the matcher performance
  improves with enhancement

33
Outline

• Introduction
• Fingerprint Image Enhancement
• Minutia Feature Extraction
• Prior Related Work
• Chain code contour
• Experimental Evaluation
• Matching Algorithm
• Conclusion
• Software Demos

34
Background

• Minutiae represent local discontinuities in ridge
  flow
• Minutiae features are the most widely used
  fingerprint representation
• There are several standards such as CBEFF (file
  format) and ANSI-NIST (interchange format)
  standards for minutiae based fingerprint
  representation

• Minutiae extraction approaches may be broadly
  categorized into
• Binarization based approaches
• Direct gray scale extraction

35
Prior Related Work

• Binarization Approaches
• MINDTCT,NIST NFIS, (Garris et. Al, 02)
• Directionally adaptive binarization
• Template matching is used to detect minutiae
• Adaptive Flow Orientation technique (Ratha et.
  al., 95)
• Binarization is performed by peak detection
• Peak detection leads to false positives in
  regions of poor ridge constrast.
• Direct Gray Scale Ridge Following
• Ridge Following (Maio and Maltoni 97, Jiang and
  Yau 01)
• Based on ridge pursuit
• Has low computational complexity.
• Cannot handle poor contrast prints and images
  with poor ridge structure.
• Relies on a good orientation map for ridge pursuit

36
Binarization Method
Binarization
Acquisition
Thinning
Minutia Detection
37
Proposed Approach Chain Code Contours

• Provides a lossless description of the contour
  and also gives direction and curvature
  information.
• Translation and rotation invariant
• Used in computer vision for encoding object
  boundaries
• Used for character recognition (Madhavanth et. al
  99)

38
Minutiae Detection using Chain Codes

• Minutiae are encountered as points of
  significant turn on the contour
• Left turn Ridge ending
• Right turn Bifurcation

39
Determining Turn Points
40
Results
41
Results (cont.)
42
Experimental Evaluation

• Test Data
• 150 prints from FVC2002(DB1) were randomly
  selected for evaluation.
• Ground truth was established using a semi
  automated truthing tool.
• Results compared using NIST NFIS open source
  software.
• Metrics
• Proposed by Sherlock et. Al 94
• Sensitivity Ability of the algorithm to detect
  true minutiae
• Specificity Ability of the algorithm to avoid
  false positives
• Flipped Minutiae whose type has been exchanged

43
Quantitative Analysis Results

• Examples

44
Results

• Summary results
• Count TP(NIST) gt proposed 40 of 150
• Count E(NIST) lt proposed 40 of 150

Sensitivity distribution
Overall statistics
45
Outline

• Introduction
• Fingerprint Image Enhancement
• Minutia Feature Extraction
• Matching Algorithm
• Prior Related Work
• New Representation K-plet
• Local Matching Dynamic Programming
• Consolidation Coupled BFS
• Experimental Evaluation
• Conclusion
• Software Demos

46
Minutiae Based Matching

• Challenges
• Minutiae extraction is error prone is low quality
  images
• Not robust to non-linear distortion.
• Intra-user variation

47
Challenges Non-linear Distortion
48
Challenges Quality and Intra-user variance
Variation in quality
Intra-user variation
49
Prior Related Work Global Matching

• Global Matching
• Point correspondences not known combinatorial
  problem
• Relaxation approach (Ranade and Rosenfield 93)
• Likelihood of each match is either decreased or
  increased at each iteration based on
  compatibility of rest of the points
• Iterative approach makes it too slow to be
  practical
• Generalized Hough Transform (Ratha et al. 96)
• All possible transformation represented as a
  quantized search space
• Searches for the most optimal transform in the
  search space
• Very fast
• Ridge Alignment (Jain et al. 97)
• Performs explicit alignment before matching
• Each minutiae is associated with its ridge
  (represented by a curve)
• The alignment is based on ridge correspondence
• Global matching is then performed using string
  edit distance

50
Prior Related Work Local Matching

• (Jiang and Yau 00)
• 11 dimensional local features derived from
  reference minutiae and two closest neighbors
• Best match is used only for explicit alignment
• (Jea and Govindaraju 04)
• 5 dimesional features Si (ri0, ri1, fi0, fi1, di)
  derived from two closest neighbors
• Alignment is still required
• (Ratha et al. 00)
• Star representation derived from all minutiae
  within a particular radius
• Consolidation by checking consistency
• (Garris et. al 03 BOZORTH3)
• Line features
• Consolidation by linking consisting matches

51
Proposed Algorithm

• Representation
• K-Plet
• Features invariant to rotation and translation
• Local relationship formally represented by a
  directed graph
• Local Matching
• Posed as a string alignment problem and solved by
  dynamic programming
• Matches all neighbors simultaneously
• Consolidation
• Coupled Breadth First Search
• Breadth first search is used to propagate the
  matches
• Similar to human verification
• No explicit alignment required at any stage

52
Neighborhood Representation K-plet
53
K-plet
T
r
F
54
Local Matching

• All local neighbors have to be matched
  simultaneously. Greedy approach does not work
  when conflicts occur
• These can solved by finding the alignment through
  optimization process such as by solving a string
  alignment problem

• Example of alignment
• S (acbcdb) (ac__bcdb)
• T (cadbd) - (_cadb_d_)
• Trivial solution requires exponential time
• Each match is given a cost. Alignment solved
  through recurrence relation

55
The Graphical View
56
Graphical Matching Coupled BFS
57
Coupled BFS
58
Graphical Matching Coupled BFS
59
Graphical Matching Coupled BFS
60
Graphical Matching Coupled BFS
61
Important Differences

• Traditional Breadth First Search
• Traversal Defined only over a single graph
• All neighbors are considered for expanding the
  path
• Coupled Breadth First Search
• Traversal proceeds in two directed graphs
  simultaneously
• Only matched neighbors are considered for
  expanding the path
• Constant number of neighbors provides a bound for
  the traversal complexity

62
Experimental Evaluation

• 800 prints from FVC2002(DB1)
• 2800 genuine tests,4950 impostor tests
• Compared with BOZORTH 3

Error Rates BOZORTH3 3.6 EER, 5.0
FMR100 Proposed 1.5 EER, 1.65 FMR100
63
Software

• CUBS Truthing Tool
• CUBS Minutiae Truthing Tool
• CUBS Fingerprint Verification Demo
• Matlab code for Fingerprint Enhancement
• Matlab Toolbox for Fingerprint Verification
• http//www.mathworks.com/matlabcentral
• 1179 downloads since 01/30
• 637 downloads since 03/22

64
Conclusion

• Contributions
• New Fingerprint Image Enhancement using STFT
  Analysis.
• Simultaneously estimates all intrinsic images
• Increases recognition rate of existing matchers
• New Feature Extraction Algorithm using Chain code
  Contour
• Obviates need for thinning
• Performs favorably with NIST feature extractor
• New Graph based matching algorithm
• Robust to non linear distortion
• Formal technique for propagating local matches
• Performs better than NIST BOZORTH3 matcher over
  FVC DB1 database

65
Acknowledgements

• Tsai Yang Jea (Alan)
• Chaohang Wu
• Sergey Tulyakov
• Faisal Farooq
• Amit Mhatre
• Karthik Sridharan
• Sankalp Nayak
• Rest of the research group at CUBS

66
Thank You

• http//www.cubs.buffalo.edu