Profile-Based 3D Face Registration and Recognition

With the rapid development of 3D imaging technology, face recognition using 3D range data has become another alternative in the field of biometrics. Unlike face recognition using 2D intensity images, which has been studied intensively by many researchers since the 1960’s, 3D range data records the exact geometry of a person and it is invariant with respect to illumination changes of the environment and orientation changes of the person. This paper proposes a new algorithm to register and identify 3D range faces. Profiles and contours are extracted for the matching of a probe face with available gallery faces. Different combinations of profiles are tried for the purpose of face recognition using a set of 27 subjects. Our results show that the central vertical profile is one of the most powerful profiles to characterize individual faces and that the contour is also a potentially useful feature for face recognition.