THE LAYERED FACE REPRESENTATION MODEL AND FACE RECOGNITION

Automatic human face recognition is a difficult but significant problem. A novel method for face recognition based on layered face representation is put forward in this paper. The method first locates the face and key facial features including eyes, nose, and mouth in image quickly, and then normalizes that face depending on the center location of eyes. Then PCA neural network is used to extract characteristics of the located faces and facial features. This can be viewed as a layered representation of faces: where a coarse and low\|resolution description of the whole head is augmented by additional high\|resolution details in terms of salient facial features. This method has been tested on 1300 facial images and shows better performance than traditional PCA for face recognition, especially when head poses and expressions are changed.