Watchlist Risk Assessment using Multiparametric Cost and Relative Entropy

TL;DR

This paper introduces a novel risk assessment method for facial biometric watchlist screening using multiparametric cost and relative entropy measures, demonstrating effectiveness in detecting threats and reducing false identifications.

Contribution

It proposes new techniques for designing and analyzing biometric watchlist systems and evaluates impersonation impacts on border security performance.

Findings

Effective detection of mis-identification and impersonation scenarios

Quantitative analysis of impersonation impact on border security

Demonstrated robustness of proposed risk measures

Abstract

This paper addresses the facial biometric-enabled watchlist technology in which risk detectors are mandatory mechanisms for early detection of threats, as well as for avoiding offense to innocent travelers. We propose a multiparametric cost assessment and relative entropy measures as risk detectors. We experimentally demonstrate the effects of mis-identification and impersonation under various watchlist screening scenarios and constraints. The key contributions of this paper are the novel techniques for design and analysis of the biometric-enabled watchlist and the supporting infrastructure, as well as measuring the impersonation impact on e-border performance.