Signal and Image Processing for Biometrics

Chapter 9Multibiometrics ¹

9.1. Introduction

Biometric systems are increasingly used in various domains: passport, identity card, access control, etc. In theory, they represent an increasingly reliable and efficient method for identifying a person. Indeed, biometrics defines itself as what we are, as opposed to what we have (e.g. badges, cards, and USB drive), or what we know (password and personal identification number (PIN) code). Thus, it is intrinsically linked to the individual. But in practice, biometric systems have certain limitations. Thus, we define false acceptance rate (FAR), false rejection rate (FRR), and equal error rate (EER), which allow comparison between different systems. Multibiometrics represents one of the proposed solutions for overcoming these limitations. This chapter offers a summary of the different aspects and techniques of multibiometrics. Until now, the reference book has been the Handbook of Biometrics by Ross et al. [ROS 06b].

Generally, all uses of a biometric system include two phases: enrollment and verification.

— Enrollment phase

This phase aims at capturing biometric data (fingerprint, face, iris, etc.) using advanced sensors, by extracting features that are stored as a form of biometric reference of the individual in the system’s database.

— Verification phase

This phase aims at comparing the new data capture, called a query, with the reference data of the considered individual.

In principle, a biometric system uses only one modality (physical ...