Facial Biometric Templates and Aging: Problems and Challenges for Artificial Intelligence

Abstract

The performance of face recognition and/or authentication systems is greatly affected by within-person variations encountered in human faces. Within person facial variations distort the appearance of faces leading to inconsistencies between facial features stored in templates and features derived from a face, of the corresponding subject, captured at a different instance. For this reason a considerable amount of effort has been devoted to the development of methods for eliminating within person-variations during face recognition/authentication. Among all types of within-person variations encountered, aging-related variations display unique characteristics that make the process of dealing with this type of variation a challenging task. In this paper we describe experiments that enable the quantification of the effects of aging on the performance of face recognition systems. We also review typical approaches that aim to eliminate the effects of aging in face recognition and outline future research directions for this area.