ICAPR Tutorial on Automated Biometrics

1. Introduction

   In the modern world, there is an ever-growing need to authenticate and identify individuals automatically. The current technologies of using a PIN or password for these purposes are inadequate because they are disclosable, transferable and hard to remember. Biometric-based authentication and identification methods are emerging as the most reliable. Automated biometrics deal with physiological or behavioral characteristics such as fingerprints, voice and face that can be used to authenticate a person's identity or establish an identity within a database. With rapid progress in electronic and Internet commerce, there is also a growing need to authenticate the identity of a person for secure transaction processing.

   When designing an automated system to handle large population identification, accuracy and reliability of authentication are significant challenges. The biometric systems need to handle variations, distortions and noise in inputs from the real world. Researchers have addressed many issues including sensing, representation, matching and error analysis of automated identification systems. In particular, fingerprint identification and speaker verification have received considerable attention over the last 25 years. Recently, the problems of face recognition and iris-based authentication have been studied extensively. Many kinds of biometric systems are now being deployed successfully in commercial applications.

   This tutorial will address many research as well as practical issues in automated biometrics. The underlying pattern recognition and computer vision techniques will be reviewed. The state of the art in fingerprint, face, iris and speaker identification will be presented. Issues in integrating biometrics and representative applications will also be discussed.

2. Outline

   • Biometrics
     • Introduction
     • Identification vs. Verification
     • Taxonomy of biometrics
     • Application characteristics
     • Pattern Recognition and Computer Vision
       • Statistical pattern recognition
       • Decision making and Bayes decision theory
       • Error estimation
       • System performance evaluation
       • Sensing modalities
       • Representation issues
       • Matching and decision making
     • Issues in biometrics systems:
       • Privacy
       • Acceptability
   • Specific Biometrics
     1. Fingerprint identification
        • Review of Automatic Fingerprint Identification Systems
        • Sensing techniques
        • Features and representation
        • Automatic feature extraction
        • Matching techniques
        • Compression/decompression standard
        • Performance evaluation
     2. Face recognition
        • Review of face recognition
        • Obstacles to face recognition
        • Face finding
        • Facial feature finding
        • Face representations and comparison
        • Face databases
        • Performance evaluation
     3. Speaker identification
        • Taxonomy of speaker ID tasks
        • Obstacles to speaker ID
        • Speech acquisition and representation
        • Speaker identification
        • Speaker databases
        • Performance evaluation
     4. Iris recognition
        • Image acquisition
        • Feature extraction
        • Matching
        • Performance evaluation
   • Future directions in biometrics
     • Integration of biometrics
     • Applications
       • Internet commerce
       • Smartcard authentication
       • Security and biometrics

3. Intended audience

   This tutorial is intended for those with a general computing background, and will cover the basics of biometrics. It is intended for those with little or no knowledge of biometrics, and is meant to be of use to those beginning work in the field and also to those wishing to acquire a general awareness of the current state of the art. The tutorial will present a variety of biometrics (finger, face etc.) as pattern recognition and computer vision problems, and will cover some basic pattern recognition issues common to biometrics problems. Advanced applications of biometrics in the area of electronic commerce, smartcard and security will be discussed.

Slides

• Introduction and overview of biometrics issues
• Fingerprint and iris recognition
• Bibliography for first half
• Face recognition
• Bibliography
• Speaker ID, signature recognition, Multibiometrics and advances in biometrics.
• Bibliography for second half

[ IBM home page | Order | Search | Contact IBM | Help | (C) | (TM) ]