An Improved Src Algorithm For Face Recognition Via Gradientfaces And Sqi

Zipeng Yu,Zhendong Wu,Jianwu Zhang
2013-01-01
Abstract:As the importance of information security being concerned, the technology of face recognition was carried on research widely. As one of the mainstream recognition algorithms, SRC cast the recognition problem as the problem of finding a sparse representation of the test image in terms of the training set as a whole and SRC had a good recognition. However, variable illumination and facial expression had a large effect on recognition in the real environment. In this paper, we proposed an improved SRC algorithm for face recognition via Gradientfaces and SQI, named GQ_SRC. First, GQ_SRC applied Gradientfaces and SQI to extract illumination invariant and emphasize the face contour and facial details. Then, GQ_SRC used PCA to emphasize the face feature and reduce the image dimensions and the time complexity. At last, by calculating the sparse representation of Gradientfaces and SQI and comparing the image residuals synthetically, we could identify the class of the test image. Experimental results on the ORL database and the actual face database showed that GQ_SRC had better generalization ability than SRC.
What problem does this paper attempt to address?