Misalignment-robust Face Recognition via Efficient Locality-constrained Representation

Misaligned face recognition has been studied in the past decades, but still remains an open challenge. To address this problem, we propose a highly efficient misalignment-robust locality-constrained representation (MRLR) algorithm. Specifically, MRLR first aligns the query face via the L2-norm locality-constrained representation, and then recognizes it by a standard L2-norm collaborative representation algorithm. It achieves a high degree of robustness even with a small training set. Moreover, we take advantage of the block matrix inverse to develop an efficient solving algorithm whose efficiency and scalability are verified by computational complexity analysis. Experimental results on public data sets show that MRLR beats several state-of-the-art approaches in terms of efficiency and scalability with comparable performance.