Discriminative Common Tensorface for Face Recognition

There is a growing interest in subspace discriminative feature extraction techniques based on tensor (multilinear) representation, which encodes an image object as a general tensor of second or even higher order. However, on one hand the computational convergence of its iterative algorithms is not guaranteed, on the other these methods are impractical for real-time applications for large training sets because the test sample must be compared to all training samples. In this paper, we present a novel approach, named discriminative common tensorface, to solve such questions mentioned above. This new method presents an image as a tensor presentation and gives an iterative algorithm to extract the discriminative common tensorface each person in the training set of the face database. Experiments on test data show that the proposed algorithm has strong discriminant ability and is practical for real-time applications for large training sets.