Fingerprints: 75 Billion-Class Recognition Problem Anil Jain Michigan State University October 23, 2018 http://biometrics.cse.msu.edu/
Friction Ridge Patterns Dermatoglyphics. Derma: skin; Glyphs: carving Cumins and Midlo, Finger Prints, Palms and Soles, Dover, 1961 2
Fingerprints 0.73 x 0.84 in. Each finger, including those of identical twins, has a different pattern 3
75-Class Recognition Problem 7.5 billion x 10 = ~75 billion pattern classes 350 K births/day; no. of classes keeps increasing 4
First Encounter with Fingerprints Sun SPARCstation host ~100 MHz CPU; 512 MB RAM 16 Xilinx 4010s as PEs (512 KB memory) Fingerprint matching on Sun SPARC: 70 matches/sec 100-times speedup on Splash 2 FPGA @1 MHz clock Ratha, Rover, Jain, "An FPGA-Based Point pattern Matching Processor with application to Fingerprint Matching", CAMP `95, 5
Fingerprint Formation Ridge formation starts at 1 or 2 focal points and spreads over the fingertip Localized ridge units merge to form ridges at ~10.5 weeks of gestational age Fingerprints possess genotype & phenotype properties L. S. Penrose and P. T. Ohara. The development of the epidermal ridges. Journal of Medical Genetics. 1973 M. Okajima. Development of dermal ridges in the fetus. Journal of Medical Genetics. 1975 6
Fingerprint Milestones Seventeen classes 9/11 terrorist attacks lead to govt. Whorl loopaadhaar (left and right) mandates to use(double biometrics loop), in What is wanted is a means of regulating travel 99% of fingerprints & archintl. cover Supreme Court upholds Habitual Criminals Act Fingerprint a personal classifying the First records ofuse of the constitutional US Congress authorizes DOJ to collect Identimat: commercial Galton Henryas fingerprint system State/ AFIS mark fingerprints biometrics adopted by Scotland Yard and arrest information State AFIS Pay of Aadhaar habitual criminal, such that as Apple validity State AFIS soon as the particulars of the personality of any prisoner (whether description, measurements, marks, or 300 1839 1858 1869 1883 1900 1905 1924 1963 1972 1999 2001 2003 2008 2013 2014 2017 photographs) are received, it B.C. may be possible to ascertain Other operations Criminal booking readily, and with certainty, Forensics FaceID of Bertillonage First use usedeed of fingerprints AEarly Chinese of sale US-VISIT whether his case is FBIininaugurates the TouchID fingerprints in on fingerprint matchingfull invented for Civil applications with a fingerprint Trauring publishes paper in Nature operation Aadhaar gives dignity to the marginalized. Dignity register, and if so, who he is British criminal case (Bengal, India) Goldstein et al. publish face recognition paper in Proc. IEEE (1971) State AFIS IAFIS of IAFIS 2018 to the marginalized outweighs privacy, Justice FBI Next Generation Identification Sikri Delta Core 7
Drivers: Lack of Trust Osmania University (OU) enhanced the exam fee in all the affiliated colleges by Rs. 100 per semester for implementation of biometric attendance system. Times of India, Oct 22, 2018 No end to JNTU-H fake certification verifications. HANS INDIA, Times of India, Oct 12, 2018 8
Enablers: Fingerprint Readers 1892 Juan Vucetich Ink and Paper 1990 Optical sensor 1990 Capacitive sensor 9
Enablers: Processors, RAM, Algorithms 1960s 1989 725K tenprints 15K matches/sec 2017 4M tenprints 1M matches/sec Courtesy: James Blanchard, Michigan State Police
Fingerprint Enhancement Ridge Voting Hong, Wang and Jain, IEEE Trans. PAMI, 1999
Fingerprint Representation Level-1 Level-2 Level-3 Ridge flow and pattern type Minutiae Pores and incipient ridges Singular Points Bifurcation Incipient Ridge Cores Ending Pores Deltas Orientation Field Template: A compact representation of fingerprint features 12
Minutiae Extraction Input Image Ridge Flow Ridge Filter Extracted Minutiae Postprocessing Ridge Thinning Minutiae Detection 13
14 Minutiae Descriptors Ridge Flow-based Descriptor Ridge flow values in the minutiae neighborhood Neighboring minutiae-based Descriptor Set of minutiae in a local neighborhood Minutia Neighborhood Flow-based Descriptor Minutiae-based Descriptor
Fingerprint Comparison Query fingerprint Similarity = 0.9 Enrolled fingerprint 15
How to Align? Query Fingerprint Gallery Fingerprint Jain, et al. An Identity Authentication System Using Fingerprints, Proc. IEEE, 1997 16
Frequency Recognition Performance ROC curve 1 1 Similarity score Threshold determines tradeoff between FAR & FRR 17
System Requirements 100K visitors/day to Disney Park, Orlando Usability Fast verification to maintain throughput Low error rates, especially FRR Day/night operation Robust to finger condition: wet, dry,.. Return on investment Embedded system Template encryption 18
Aadhaar: World s Largest Biometric System 121 crore (1.21 billion) individuals have been enrolled 19
De-duplication: Limited Capacity of Fingerprints
Aadhaar Authentication 21
Daily Authentication Transactions 100% successful authentication NOT possible, UIDAI CEO admits in SC https://uidai.gov.in/aadhaar_dashboard/auth_trend.php
State of the Art Performance Rolled prints Plain prints Latent prints Authentication: TAR of 99.9% @FAR = 0.001% Retrieval (search) Plain prints: 99.3% (100K background gallery) Latent prints: 67.2% (70.2% with image + markup) C. Watson, et al.. Fingerprint Vendor Technology Evaluation, NISTIR, 2012 M. Indovina, et al.. ELFT-EFS Evaluation of Latent Fingerprint Technologies: Extended Feature Sets NISTIR, 2012 23
Sources of Error No. of false minutiae = 0 No. of false minutiae = 7 No. of false minutiae = 27 24
Sources of Error Genuine comparisons 489 368 6 329 77 Imposter comparisons Query 29 11 12 21 20 Intra-finger variations and Inter-finger similarity
What s Next? Scale 10 7 10 5 10 3 10 1 99% 99.999% Unusable Hard to Use 90% 99.99% Accuracy Easy to Use Transparent to User Usability Fingerprint Recognition is not solved!
Scalability Assume one billion users Identification Performance False Negative Identification Rate (FNIR): user is enrolled, but not retrieved False Positive Identification Rate (FPIR): user is not enrolled, but an identity is returned Identification & verification performances are related FPIR = 1 (1 FMR) N ; FNIR = FNMR Suppose for N = 10 9 enrollment, we require FNIR = 0.0001%; FPIR = 0.001% Require a matcher: FMR 10-12 %; FNMR = 0.0001% 27
Capacity & Persistence Uniqueness: How many different individuals can we recognize? 6-digit code:10 6 unique PINs Permanence: Does the recognition performance change over time? PINs do not become stale
29 Prob. of False Correspondence "Two like fingerprints would be found only once every 10 48 years (Sc. Am, 1911) Prob. of a fingerprint with n minutiae and another with m minutiae sharing q minutiae (a) M=52 m=n=q=26 P = 2.40 x 10-30 (b) M=52 m=n=26, q=10 P = 5.49 x 10-4 M = A/C Pankanti, Prabhakar and Jain, On the Individuality of Fingeprints, IEEE PAMI, 2002
Persistence Database: fingerprints of 20K convicts with an average of 8 arrests over a span of 12 years Longitudinal model showed: Fingerprint accuracy (i) is stable over 12 years, (ii) accuracy depends more on fingerprint quality than time gap Yoon and Jain, "Longitudinal Study of Fingerprint Recognition", PNAS, 2015 30
Spoof Attacks Requirements: TAR = 98% @FAR = 0.2% Chugh, Kai and Jain, "Fingerprint Spoof Buster: Use of Minutiae-centered Patches", IEEE TIFS, 2018
Fingerprint Obfuscation Fingerprint of Gus Winkler (1933) before and after alteration 32
Template Protection Similarity Reconstructed Original ISO Score Fingerprint = Fingerprint 460 Template (VeriFinger) Image Image 33
Fingermarks (Latent Prints) 34
Madrid Train Bombing (2004) 35 Partial print on a duffel bag Brandon Mayfield s prints in file
Automated Latent AFIS Reference database Feedback Acquisition Cropping Enhancement Minutiae Comparison Kai and Jain, IEEE PAMI, 2018 290 71 70 48 Candidate list 36
Successful Match Latent Mated Rolled # Matched minutiae = 2 Similarity score = 3 Enhanced Latent Mated Rolled # Matched minutiae = 13 Similarity score = 38
Infant Fingerprints Digital Persona U.are.U (500 ppi) Custom NEC Reader (1270 ppi) MSU Match in Box (1900 ppi) Right thumb of a 3 month old infant captured with 500,1270 & 1900 ppi readers Jain, Arora, Cao, Best-Rowden, Bhatnagar, Fingerprint recognition of young children, IEEE TIFS, 2017 Engelsma, Cao, Jain, Fingerprint Match in Box. IEEE BTAS, 2018 Engelsma, Cao, Jain, Fingerprint Match in Box. IEEE PAMI, 2018
Fingerprint Match in Box (a) (b) (c) A low cost ($200), open source, 1900 ppi compact (10 cm cube) fingerprint reader with embedded spoof detector, feature extractor, and matcher with 1:100K search; thumbprint of 3-month-old baby
Privacy Concerns 40
Security v. Privacy
Summary Fingerprint based transactions are used by hundreds of millions of citizens worldwide Applications: mobile phone unlock, social benefits disbursement, border crossing, forensics; new applications are emerging Challenges: sensor design, image quality, robust & accurate solution, privacy, security It s all about lack of TRUST
Security v. Privacy 44