A New Face Recognition Method Based on the Energy Image of Facial Contour

In order to further improve the effectiveness and enhance practicability of face recognition under guaranteeing the recognition accuracy, this paper proposes a new method of face recognition based on the energy image of facial contour. First, image binaryzation and edge detection method are used to extract the facial contour, then the energy image of facial contour of training samples are obtained, second, matrix logic "and" operation is adopted to calculate the similarity between a test sample and the energy image of facial contour of each class, finally, the test sample is classified according to the similarity criteria. The method is based on a kind of similarity match, so it can make good use of two-dimensional structure information of face images, moreover, it adopts matrix logic "and" operation in classification instead of solving the Eigen value problem and calculating the Euclidean distance of traditional classification algorithm, which can reduces the computational complexity dramatically. Finally, the contrast experiment with principal component analysis and sparsity preserving projections shows that the proposed method is more effective and stable on the Yale face database.