MODIFIED LOCAL BINARY PATTERN FOR HUMAN FACE RECOGNITION BASED ON SPARSE REPRESENTATION

摘要

Over the last decades, research on facial analysis has witnessed a growing interest and became a very active topic in computer vision. Broadly, it can be addressed in either of the two ways, namely facial representation and classification. Considering the former category, many representations could be found in the literature. One of the most popular representations is the well-known local binary pattern (LBP). In this respect, we propose in this paper a novel alternative to the basic LBP for face representation, termed a modified local binary pattern (MLBP), which we prove its outperformance over other popular techniques. On the other hand, we exploit the sparsity of the representative set of MLBPs for recognizing different face classes. Therefore, compressive sensing theory was employed to construct a so-called sparse representation classifier. Experimental results conducted on three popular face databases pointed out the superiority of our proposed strategy over other state-of-the-art techniques.