Fusion of color, local spatial and global frequency information for face recognition
Zhiming Liu,Chengjun Liu
DOI: https://doi.org/10.1016/j.patcog.2010.03.003
IF: 8
2010-01-01
Pattern Recognition
Abstract:This paper presents a novel face recognition method by means of fusing color, local spatial and global frequency information. Specifically, the proposed method fuses the multiple features derived from a hybrid color space, the Gabor image representation, the local binary patterns (LBP), and the discrete cosine transform (DCT) of the input image. The novelty of this paper is threefold. First, a hybrid color space, the RC"rQ color space, is constructed by combining the R component image of the RGB color space and the chromatic component images, C"r and Q, of the YC"bC"r and YIQ color spaces, respectively. The RC"rQ hybrid color space, whose component images possess complementary characteristics, enhances the discriminating power for face recognition. Second, three effective image encoding methods are proposed for the component images in the RC"rQ hybrid color space to extract features: (i) a patch-based Gabor image representation for the R component image, (ii) a multi-resolution LBP feature fusion scheme for the C"r component image, and (iii) a component-based DCT multiple face encoding for the Q component image. Finally, at the decision level, the similarity matrices generated using the three component images in the RC"rQ hybrid color space are fused using a weighted sum rule. Experiments on the Face Recognition Grand Challenge (FRGC) version 2 Experiment 4 show that the proposed method improves face recognition performance significantly. In particular, the proposed method achieves the face verification rate (ROC III curve) of 92.43%, at the false accept rate of 0.1%, compared to the FRGC baseline performance of 11.86% face verification rate at the same false accept rate.