03-Biometrics-Lecture-3-Part2-2008-10-06

03-Biometrics-Lecture-3-Part2-2008-10-06 - Master SC...

Info iconThis preview shows pages 1–11. Sign up to view the full content.

View Full Document Right Arrow Icon
1 Biometrics http://scgwww.epfl.ch/courses Master SC – Information and Communication Security Dr. Andrzej Drygajlo Speech Processing and Biometrics Group Signal Processing Institute Ecole Polytechnique Fédérale de Lausanne (EPFL) Center for Interdisciplinary Studies in Information Security (ISIS)
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 Biometrics - Contents Fundamentals of Biometrics Identity and Biometrics Individuality of Biometric Data Recognition, Verification , Identification and Authentication Analysis, Modeling and Interpretation of Biometric Data Mathematical Tools Sensing and Storage Representation and Feature Extraction Enrollment and Template Creation Biometric System Errors Evaluation of Biometric Systems
Background image of page 2
3 Enrolment phase Recognition phase A/D conversion- End-point detection- Pre-emphasis- Frame blocking- Feature extraction- Feature-domain normalisation- Basic components Speaker modelling Model database Front-end Feature vectors MFCC- LPCC- -Text-dependent: DTW, HMM -Text-independent: VQ, HMM, GMM -Speaker verification -Speaker identification Score computation
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 Biometric Verification Templates or models Feature extraction Sensor Enrollment module Score generation Thresholding or comparison Sensor CLASSIFIER Verification (testing) module 1. SIGNALS 2. FEATURES 3. SCORES 4. DECISIONS (Users) Feature extraction
Background image of page 4
5 Biometric Verification with Iris and Speech Example Assume 10’000 customers are signed up for biometric authentication and 1’000 transactions are done weekly Assume best-case biometric verification error of 1 in 1 million (iris) Assume best-case speaker verification error of 1 in 1 hundred How often are customers falsely billed? Answer On average 10 people are falsely billed each week On average 100 000 people are falsely billed each week
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 References R.M. Bolle, J.H. Connell, S. Pankanti, N.K. Ratha, A.W. Senior, Guide to Biometrics , Chapter 5 “Basic System Errors”, Springer- Verlag, New York, 2004
Background image of page 6
7 Matching B B Sensing and Feature Extraction Matching Sensing and Feature Extraction R R’ S ( R’,R ) Real-world biometrics Features (, ) ((( ) ) ,( ( ) ) tt sR R s f B t f Bt ′′ = Biometric matching makes a decision by computing a measure of the likelihood that the two input samples from two persons are the « same » and hence that the subjects are the same real-world identity.
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 Performance Evaluation (Threshold) Scores Impostors Genuine users Distribution Global threshold S User T FNMR FMR T
Background image of page 8
9 FMR and FNMR Error rate 1 0 Decision threshold False acceptance (FA) False Match Rate (FMR) False rejection (FR) False Non-Match Rate (FNMR) T 1 T 2 T 3 Equal Error Rate (EER)
Background image of page 9

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

View Full DocumentRight Arrow Icon
10
Background image of page 10
Image of page 11
This is the end of the preview. Sign up to access the rest of the document.

Page1 / 39

03-Biometrics-Lecture-3-Part2-2008-10-06 - Master SC...

This preview shows document pages 1 - 11. Sign up to view the full document.

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