Part-based 3D face recognition under pose and expression variations

Abstract

The advances in sensor technologies and the several years of research in recognition of biometric modalities increased the expectations from 3D face recognition systems. An important reason of scientific interest on 3D face recognition is the ability of acquisition of the facial data nonintrusively. This makes 3D face recognition applicable to real life tasks in terms of security and human computer interaction. In this study, a fully automatic part-based 3D face recognition system has been proposed. The proposed system is based on pose-correction and curvature-based facial segmentation for recognition tasks. Utilization of facial parts in the recognition step provides robustness to the system even in facial expression variations. Since the nose is anatomically the most stable part of the face, it is largely invariant under expressions. For this reason, we have concentrated on locating the nose tip and segmenting the nose. Furthermore, the nose tip and other nose landmarks enable pose correction. Pose correction feature of the proposed recognition system, allows the identification of people under significant amount of pose variations. For the face recognition task, we try both one-to-all and Average Nose Model (ANM) based methodologies. Our results show that the utilization of anatomically-cropped nose region in 3D face recognition increases the rank-one recognition success rates up to 94.1 per cent for frontal facial expressions and 79.41 per cent for pose variations in the Bosphorus database.