In this paper, we present an automatic face recognition system for personal identification in such applications as the access control. The face, the eyes, and the mouth are detected and located by using a context-free attention operator based on modified GST, resectively, from each multiresolution representation using facial symmetries. Geometrical features for the size of the facial components and their distance from the eyes and the mouthare extracted to form horizontal and vertical features. Face recognition is conducted suing backpropagation neural networks trained alternately for horizontal and vertical features. The rejetion threshold is selected experimentally for out-of-plane facial rotations. Experimental results show that the proposed system has high correct recognition rate
展开▼
