Study of Occluded Robust Face Recognition Approach Based On Homotopy Algorithm and Color Information Fusion

Robust face recognition (FR) under occluded condition is considered more and more important gradually in FR field, and it is one of the difficult problems. While sparse representation theory hit the spot of human visual characteristic and neural information effective expression, and it’s consistent with the human face inner feature, the effective fusion strategy of color face information is studied in this paper, and the homotopy algorithm is used to solve the l1 norm problem in occluded sparse representation based FR. Experimental results in AR database show that compared with traditional color face image fusion FR method and the traditional gray-scale conversion FR method, the homotopy algorithm and color information fusion based FR method can achieve high recognition performance in both unoccluded and occluded face image. It is also show that the effective integration of color information on feature fusion, also contributes to improve the efficiency and performance of face recognition system.