3D Face Recognition

This paper describes the developments of the RMA/SIC department in 3D face recognition and situates them in the research activities in this field. 3D face recognition appears as a promising approach for biometric person identification, bringing robust and specific features, with easy face detection from depth and quite difficult faking possibilities. The successful development of a prototype for real-time automatic profile identification by the SIC led us to the more complete geometrical analysis of facial surfaces. Several versions of cost effective and fast 3D acquisition systems have been realized to capture facial surfaces. Two databases were created and used to make recognition experiments by matching facial surfaces through planar profile comparison. Our developments and other works confirm that a face recognition system should integrate 3D capabilities to exploit the complementarity of robust geometrical features and normalized grey-level cues.