期刊名称：International Journal of Advanced Computer Science and Applications(IJACSA)
印刷版ISSN：2158-107X
电子版ISSN：2156-5570
出版年度：2016
卷号：7
期号：3
DOI：10.14569/IJACSA.2016.070314
出版社：Science and Information Society (SAI)
摘要：Linear representation based face recognition is hotly studied in recent years. Competitive representation classification is a linear representation based method which uses the most competitive training samples to sparsely represent a probe. However, possible noises on a test face image can bias the representation results. In this paper we propose a facial noise detection method to remove noises in the test image during the competitive representation. We compare the proposed method with others on AR, Extended Yale B, ORL, FERET, and LFW databases and the experimental results show the good performance of our method.
