03-Biometrics-Lecture3-Day1-2007-14-00-15-30

03-Biometrics-Lecture3-Day1-2007-14-00-15-30 - Centre for...

Info iconThis preview shows pages 1–9. Sign up to view the full content.

View Full Document Right Arrow Icon
1 1 Centre for Vision, Speech and Signal Processing Multimodal Biometrics Josef Kittler Centre for Vision, Speech and Signal Processing University of Surrey, Guildford GU2 7XH [email protected] 2 OUTLINE Introduction and context Fusion architectures Score level fusion: Problem formulation Score normalisation Estimation error Multiple expert paradigm Quality based fusion of biometric modalities
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
2 3 Introduction • Different biometric modalities developed –finger print –iris –face (2D, 3D) –voice –hand –lips dynamics –gait Different traits- different properties •usability •acceptability •performance •robustness in changing environment •reliability •applicability (different scenarios) 4 Introduction Motivation for multiple biometrics To enhance performance To increase population coverage by reducing the failure to enroll rate To improve resilience to spoofing To permit choice of biometric modality for authentication To extend the range of environmental conditions under which authentication can be performed
Background image of page 2
3 5 Fusion architectures Integration of multiple biometric modalities Sensor (data) level fusion Linear/nonlinear combination of registered variables Representation space augmentation Feature level fusion Soft decision level fusion Decision level fusion 6 Integration of Multimodal Biometrics FACE VERIFICATION FROM VIDEO •Lip tracker used to determine mouth status •Face template selected accordingly LIPS DYNAMICS
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
4 7 Levels of Fusion PCA LDA MFCC PLP DCT GMM MLP MSE GMM HMM Fusion Feature Fusion Data Fusion Score Fusion less information to deal w ith threshold score Legend 8 Multiple expert paradigm surgeon internist pathologist radiologist psychiatrist oculist anaesthetist Combined diagnosis exploiting complementary expertise
Background image of page 4
5 9 Decision-level fusion How useful? clients impostors score modality1 score modality2 10 Decision-level fusion Accepted by either modality clients impostors score modality1
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
6 11 Decision-level fusion Accepted by both clients impostors score modality1 score modality2 13 Decision-level fusion clients impostors score modality1 Better performance by adapting the thresholds
Background image of page 6
7 14 Score-level fusion Should improve performance clients impostors score modality1 score modality2 19 Problem formulation Given sensor outputs (features) x 1 ,…, x R Bayes decision rule Assign object to class ω i if P ( ω i | x 1 ,…, x R ) = max P ( ω j | x 1 ,…, x R ) Under the assumption of independence Decision: Assign object to class ω i if
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
8 21 Fixed fusion strategies 22 Product/sum rule in verification
Background image of page 8
Image of page 9
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 06/25/2009 for the course MATH MAT 400 taught by Professor Jamespotvein during the Fall '08 term at University of Toronto.

Page1 / 28

03-Biometrics-Lecture3-Day1-2007-14-00-15-30 - Centre for...

This preview shows document pages 1 - 9. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online