Face recognition via non-negative sparse low-rank representation classification

Rong Wang, Caikou Chen, Jingshan Li, Tianchen Dai

Abstract: This paper presents a novel method for robust face recognition, termed non-negative sparse low-rank representation classification (NSLRRC). NSLRRC seeks a sparse, low-rank and non-negative matrix over all training samples. Sparse constraint makes representation vector discriminative, while low-rank matrix will expose the global structures of data. Meanwhile, non-negative representation vectors guarantee that the coefficients are significant and better reflect the dependence among the data. NSLRRC can approximate the test sample and classify it to the correct class on account of the minimal reconstruction residual. Extensive experiments on several public face datasets prove robustness and effectiveness of our method.