Biometric identification technologies (the automatic recognition of individuals based on physical and/or behavioral characteristics) date back over 50 years to the earliest digital computers. Over the last two decades, biometric identification devices have become faster, cheaper, and more reliable, allowing for a variety of applications. This course looks at the history, theory, algorithms, applications, and standards of biometric recognition, including voice, iris, face, hand, and fingerprint identification.
Complete Details
- To explain the current definition of “biometric recognition” and the distinctives of this form of “biometrics”
- To list currently available technologies and how they work at the algorithmic level
- To present a “system analysis” of the interaction of components in “real-world” applications.
- To define performance metrics and their limitations in describing practical systems
- To present independent test results and their value in performance prediction
- To outline current international standards in the field.
- To present several case studies of successful and unsuccessful applications of this technology, particular in large-scale environments.
Coordinator and Lecturer
James L. Wayman, PhD , has worked continuously in the field of automated human recognition since 1984. He is currently a contractor to the US, UK and AU governments, an IEEE Fellow and “Distinguished Lecturer”, a “core member” of the UK government’s “Biometric Working Group”, an IET Fellow, a Principal UK Expert (PUKE) on ISO/IEC JTC1 SC 37 international standards committee on biometrics, and the editor of both the ISO/IEC 19794-13 Voice Data Format and the ANSI/NIST Type-11 Voice Data Record. He is the Vice Chair of the US Organization of Scientific Area Committees (OSAC) subcommittee on Forensic and Investigative Speaker Recognition. He is co-editor of the Wayman, Jain, Maltoni and Maio, Biometric Systems: Technology, Design, and Performance Evaluation (Springer, London, 2005). He was director of the US National Biometric Test Center under the Clinton administration from 1997-2000 and was a member of the US National Academies of Science/National Research Council committees on “Authentication Technologies and their Implications for Privacy”, “Whither Biometrics” and Panel on Information Technology. He has been a paid biometrics advisor to 11 national governments.
Daily Schedule
Day 1
Introduction
- A very short history of biometrics
- Definitions
- Core concepts
- Taxonomy of applications
- System description
- Performance metrics
Day 2
Mathematical Underpinnings
- Vector spaces
- Distance measures and distributions
- Correlation and covariance matrices
- Eigen-Systems and principal component analysis
- Sampling theory
- Generalized Fourier Transform
- One- and two-dimensional filtering
- Neural nets and support vector machines
- Probability fundamentals
- The binomial distribution
- Doddington’s Rule of 30 and the Rule of 3
- Doddington’s Zoo
- Duhem-Quine Hypothesis and holistic testing
- Bayesian inversion of conditional probabilities (The elephant in the room)
Day 3
The Algorithms
- Voice
– Detection/Segmentation
– Mel-scale cepstrum
– GMM
– HMM
– iVectors and beyond - Face
– Segmentation
– Decomposition methods (eigenface, LFA, ICA)
– Elastic Bunch Graphs
– Local correlation techniques
– ISO//IEC 19794-5 - Fingerprint methods
– Optical
– Transform
– Correlation
– Minutiae extraction
– Collection, transmission, and storage standards
– ISO/IEC 19794-2 - Iris
– Daugman’s 1994 patent
– Segmentation
– Feature extraction
– ISO/IEC 19794-6 - Other Modalities
– Hand geometry
– Retinal scanning
– Heartbeat
Day 4
Performance Testing and Applications International Testing Standards (iSO/IEC 19795)
- NIST Test programs:
– SRE
– FpVTE
– PTT
– FRVT
– MINEX
– MBGC
– MBE
– IREX
– Other NIST results - Case Studies
- EasyPASS, SmartGate, and other border crossing systems
- Social Service Systems
- Vulnerability Assessment
Upon Program Completion
Participants will:
- Understand how “biometrics” in the context of human identification differs from other forms of recognition technology
- Be able to state precisely what functions these systems perform.
- Be able to draw a system-level diagram for any biometric system and discuss its components.
- Know which technologies are currently available, which are now defunct and how the computer algorithms analyze and compare patterns
- Discuss performance metrics and their limitations in predicting “real-world” performance
- Be able to reference on-line independent government test results
- Be able to describe the state of international standards development
- Name and describe several successful and unsuccessful government applications and some of the controversies surrounding them.
For more information contact the Short Course Program Office:
shortcourses@uclaextension.edu | (310) 825-3344 | fax (310) 206-2815