Constrained Mutual Subspace Method (CMSM) is a powerful face recognition method which is able to achieve high recognition accuracy. However, CMSM requires a large number of images from each subject to perform training and testing, and it suffers serious p