Boosting Gallery Representativeness by Co-updating Face and Fingerprint Verification Systems

Abstract. The representativeness of a template gallery to the novel data is a well-known issue in a personal verification system based on biometrics. This problem has been recently faced by proposing “template update” algorithms that updates the enrolled templates in order to capture and represent better, the subject’s intra-class variations. Whilst the majority of the proposed approaches adopted “self” update technique, in which the system updates itself using its own knowledge. An approach named template co-update, using two complementary biometrics to “co-update” each other, has shown promising, but still preliminary, results. In this paper, we investigate the performance of the template co-update in comparison to self update algorithms in an uncontrolled environment. Reported results show that template co-update can outperform template “self” update technique, when initial enrolled templates are poor representative of the novel data