Removing contaminated data for illumination-robust face recognition

Recently low-rank matrix decomposition (LR) and sparse representation classification (SRC) have been successfully applied to address the problem of face recognition. Low-rank matrix decomposition is employed as the first step of robust principal component analysis (RPCA), it is robust to illumination-contaminated image data. In this paper, we propose a novel method based on low-rank decomposition and sparse representation classification which is more robust to illumination-contaminated data. This method is a kind of test-data-drive illumination-robust face recognition. Our experimental results demonstrate the effectiveness of our proposed method.