Robust Principal Component Analysis for Sparse Face Recognition

In principal component analysis (PCA), l(2)/l(1)-norm is widely used to measure coding residual. In this case, it assume that the residual follows Gaussian/Laplacian distribution. However, it may fail to describe the coding errors in practice when there are outliers. Toward this end, this paper propose a robust PCA (RPCA) approach to solve the outlier problem, by modeling the cost function as a weighted regression problem. In face recognition progress, the observation samples and testing sample be projected on the principal space firstly. After that, in the new projection space, the face be classified based on the sparse representation. Simulation results illustrated the effectiveness of this approach.