Databases for Biometric Identification

Abstract

In this lecture we consider databases that are used in biometric identification settings. We assume that the biometric sequences describing the individuals consist of independent and identically distributed variables. It is our objective to find the fundamental limits that characterize such systems. First we consider an unprotected database. We determine the so-called identification capacity, i.e., the maximum rate of individuals that makes reliable identification of the individual based on a noisy observation of the corresponding enrolled sequence possible [1],[2].