ABSTRACT
Figure 1.2 shows a schematic of how the biometric authentication process works in practice, using the sensor (ngerprint reader) in a mobile phone.
Because of the decision-theoretic framework involved, there is scope for error in any biometric authentication scheme. In particular, there are two types of errors in the context of a verication task, namely (1) false acceptance rate (FAR for short) and (2) false rejection rate (FRR for short). e rst one arises when an impostor is declared genuine by the system (i.e., the system nds a match when it should not have) and the second one arises when a genuine person is declared an impostor (when the system does not nd a match when it should). ese are typically determined with respect to thresholds that are set on the match scores, so that a set of FARs and FRRs are generated for a system with varying thresholds. e most commonly used metric for performance evaluation of a biometric system is called the equal error rate (or ERR for short) which is that value where the FAR and FRR approximately coincide.
