Face analysis and pose approximation

Abstract

The facial analysis is a very active research field because it concerns numerous fields of application such as: biometrics, robotics, video games and telecommunications. Many high-level vision applications such as facial feature tracking, facial modeling and animation, facial expression analysis, and face recognition, require reliable feature extraction. In this paper, we propose a novel approach for invariant facial feature detection (eyes, nose and mouth) followed by a matching and for facial expression recognition with respect to the identity. To represent all regions of a given face, we proceed by the extraction of landmarks, which amounts to the extraction of a set of contours.