Improved OMP selecting sparse representation used with face recognition

With the worldwide strengthening of anti-terrorism and other identity verification, the products based on face recognition are used in real life more and more. The recognition as an important ways has become the focus of academic research in the world. Face recognition accuracy can be improved by increasing the number of training samples, but increasing number will result in a large computing complexity. In recent years, the sparse representation becomes hot in face recognition. In this paper, we propose an energy constraint orthogonal matching pursuit (ECOMP) algorithm for sparse representation in face recognition. It selects a few training samples and hierarchical structure for face recognition. In this method, we re-select training samples by ECOMP, calculate the weight of all the selected training samples and find the sparse training samples which can recover the test sample. While the AR and the ORL database experimental results show that this method has better performance than other identification methods.