Emerging trends in Biometric Authentication Christophe Rosenberger GREYC Laboratory – France ENSICAEN – University of Caen - CNRS
SHPCS’09
PLAN
Authentication
Trends in biometric authentication Achievements
Conclusion & perspectives
SHPCS’09
Authentication Definition: Authentication Process whose objective is to guarantee the identity of a user or a service given a set level of confidence. User Authentication
Definition: Authentication factors An authentication factor is an authenticator element: what we know (password),
what we own (smartcard), What we are or how we behave (biometrics).
SHPCS’09
Authentication Biometric modalities : Biological analysis: Odour, blood, DNA… Behavioural analysis: Keystroke dynamics, voice, gait, signature dynamics... Morphological analysis: Fingerprint, iris, palmprint, finger veins, face, ear…
SHPCS’09
Authentication Properties: A biometric information must respect the following properties:
Universality: All individuals can be characterized by this information ; Uniqueness: The biometric information must be as dissimilar as possible for two different individuals ; Permanency: It must subsist during all individual’s life ; Collectability: The biometric information must be easily computed ; Acceptability: Users must be ready to give this information.
SHPCS’09
Authentication Enrolment Individual’s checkin in the biometric system Unique login
Association
+ Biometric reference
System
Biometric data Sensor Storage
SHPCS’09
Authentication Verification Comparison between the capture and the reference
Reference
login + biometric data
Comparison
System
Sensor
Result
SHPCS’09
Authentication Decision criterion Repartition
Low threshold : no problem for genuine users but impostors might be authenticated Threshold
High threshold: no impostor but genuine will be disturbed
genuine
impostors
True rejected
Wrong accepted
SHPCS’09
Similarity
Authentication Performance evaluation
FAR : False Acceptation Rate FRR : False Rejection Rate
FAR
EER : Equal Error Rate EER
ROC curve: FAR vs FRR
FRR
SHPCS’09
PLAN
Authentication
Trends in biometric authentication Achievements
Conclusion & perspectives
SHPCS’09
Trends in Biometric Authentication Biometric technology could be more deployed for logical and physical access control applications.
Needs: High performance biometric systems ;
Embedded device Low memory Quick verification Correctness
SHPCS’09
Trends in Biometric Authentication Definition of unconstrained biometric systems ; One capture enrolment systems Easiness of use Evaluation of biometric systems ; Performance, acceptability, security Definition of privacy preserving systems… No storage of the biometric reference
SHPCS’09
PLAN
Authentication
Trends in biometric authentication Achievements
Conclusion & perspectives
SHPCS’09
Achievements Facial Authentication Enrolment: only one image unconstrained acquisition (face detection) extraction of local face characteristics
SHPCS’09
Achievements Verification :
C. Rosenberger, L. Brun "Similarity-Based Matching for Face Authentication", IEEE International Conference on Pattern Recognition (ICPR), 2008.
SHPCS’09
Achievements Results Faces94 benchmark: 152 individuals, 20 images per individual. EER = 0.14% AR benchmark: 120 individuals: 65 men, 55 women, 26 images per individual. EER = 9%
SHPCS’09
Achievements Palm veins authentication Enrolment with a single image EER = 0% on a benchmark composed of 24 individuals
P.-O. Ladoux, C. Rosenberger, B. Dorizzi, "Hand Vein Verification System based on SIFT matching", The 3rd IAPR/IEEE International Conference on Biometrics (ICB), 2009.
SHPCS’09
Achievements Keystroke dynamics authentication Use of release, press and inter keys times during the typing of a password. 5 captures for the enrolment EER = 6% on a database of 100 individuals
R. Giot, M. El-Abed, C. Rosenberger, "Keystroke Dynamics Authentication For Collaborative Systems", The IEEE International Symposium on Collaborative Technologies and Systems (CTS), 2009.
SHPCS’09
Achievements 4 captures of the same password
SHPCS’09
Achievements Evaluation of biometric systems Analysis the perception of users
F. Cherifi, B. Hemery, R. Giot, M. Pasquet, C. Rosenberger, "Performance Evaluation Of Behavioural Biometric Systems", Book on Behavioural Biometrics for Human Identification: Intelligent Applications, 21 pages, 2009.
SHPCS’09
Achievements Biohashing Storage of a biocode computed given a random number and a biometric template
R. Belguechi, C. Rosenberger, "Study on the Convergence of FingerHashing and a Secured Biometric System", Proceedings of the International conference CIIA, 2009.
SHPCS’09
PLAN
Authentication
Trends in biometric authentication Achievements
Conclusion & perspectives
SHPCS’09
Conclusion Biometrics Benefit: Close relationship between the client and its authenticator drawbacks: performance (EER>0%) acceptability of users Perspectives: increase the computing performance improve algorithms (performance, robustness…) respect the privacy of users
SHPCS’09
Questions
[email protected] http://www.ecole.ensicaen.fr/~rosenber/
SHPCS’09