3D face authentication by mutual coupled 3D and 2D feature extraction
Ting-Yu Lin, Wen‐Pin Shih, W. C. Chen, Wen-Jen Ho · 2006
In this paper, we present a novel method for automatic 3D face authentication. We introduce a coupled 2D and 3D feature-extraction method to determine the positions of eye sockets. The nose tip is considered as the extreme vertex along the normal directions of eye sockets. Once the nose tip and eye sockets are found, the bilateral symmetrical plane will be determined. The central profile which is on the bilateral symmetrical plane is the foundation for recognizing human face in our method. We use a weighting function for ICP according to the bilateral symmetrical behavior. We take 2.5D range image and its corresponding texture as the input data and compare the scanned model with the specified database model. The value of the weighted distance between two compared models is used for authentication. We have successively implemented this method for the authentication of the human faces. The result illustrates that this method work well in self authentication.