Significance of dictionary for sparse coding based pose invariant face recognition

This paper deals with the dictionary for sparse coding based pose invariant face recognition. A two stage face recognition system is proposed in sparse coding frame work. The first stage is designed to recognize the pose of an incoming test image. It is proposed to perform pose classification as a three class classification problem in sparse coding frame work, each class representing images for frontal, left side or right side pose. This is done to choose the appropriate Weighted Decomposition Face (WD Face) dictionary, which is found to be suitable for sparse coding based face recognition. The second stage obtains the identity of the person with the help of pose specific WD Face dictionary chosen in the first stage. Experimental results have shown that the proposed approach is a simple yet robust sparse coding based face recognition system which is invariant of pose and illumination.