Guide To Biometric Reference Systems And Performance Evaluation (Hb)

With the pioneering studies of Galton and Henry on ?ngerprint-based human id- ti?cation over a hundred years ago, biometrics recognition is now permeating many sectors of our society. Even though the origins and the most frequent use of b- metrics, primarily ?ngerprints, have been in forensics and law enforcement app- cations, biometric recognition is now needed to address a number of societal pr- lems related to security risks, terror threats and ?nancial fraud. In some situations, for example, detecting duplicate enrolments (for issuing of?cial documents such as passports and driver licenses), biometrics is the only method to con?rm that an - plicant has not been previously enrolled in the database. In addition to ?ngerprints, new anatomical and behavioral traits, viz. , face, iris, voice and hand shape have also become popular. Biometrics recognition is an excellent proving ground for pattern recognition. While pattern recognition systems have been available for document imaging, speech recognition, medical image analysis, remote sensing and variety of insp- tion tasks, no other application domain offers as many challenges as the biometrics domain.

Our daily interactions are based on recognizing people. As information technology advances, our interactions will be accomplished through newly designed interfaces, with a vital element being the automated process of recognition of a person’s real identity: Biometrics. In recent years, biometrics has moved from using fingerprints, to many different methods of assessing human physical and behavioural traits.

This useful guide introduces a new performance evaluation framework designed to provide complete coverage of performance evaluation of major biometric systems. Two key areas are presented: The first gives a unique snapshot of the performance evaluation of various biometrics within a common evaluation framework (databases and protocols). The second presents a state-of-the-art benchmarking evaluation framework for use in future performance evaluations – comprised of open-source reference systems.

Features and benefits:

• Includes a Foreword by renowned biometric expert Prof. Anil K. Jain, Michigan State University, USA

• Introduces the concept of reproducible results and comparative evaluations

• Presents an extensive benchmarking methodology

• Describes reference algorithms in detail, with a link to source code

• Biometrics Reference and Evaluation Toolkit provided via http://share.int-evry.fr/svnview-eph/

• Implements this methodology throughout all chapters

• Offers readers a tool for a better evaluation of their algorithms by giving them the instruments to fully reproduce the benchmarking (reference) experiments (software, database, experimental protocols)

• Provides a global perspective with a common evaluation scheme and methodology

• Examines current research results with this framework

• Consolidates results from publicly available databases, using well defined protocols

This unique text/reference combines state-of-the-artresearch with practical tools for researchers, practitioners and graduates in the field of Biometrics and Pattern Recognition, and with its comprehensive coverage, it will prove an indispensable resource and frequently used tool.

 

Key topics

BioSecure Benchmarking Methodology

Biometric Performance Evaluation

Iris Recognition

Fingerprint Recognition

Hand Recognition

Online Handwritten Signature Verification

Text-independent Speaker Verification

2D and 3D Face Recognition

Talking Face Verification