Face recognition using entropy-augmented face isolation and image folding as pre-processing techniques

The appearance of the face varies drastically when background, pose and illumination change. Variations in these conditions make Face Recognition (FR) an even more challenging and difficult task. In this paper, we propose two novel preprocessing techniques, viz., Entropy Augmented Face Isolation (EAFI) and Image Folding, to improve the performance of the FR system. EAFI is used to localize the face region to minimize the effect of cluttered background, thereby enhancing face recognition. Image Folding uses the property of vertical symmetry in the human face to normalize pose variance. The resulting pre-processed image contains the salient details of the face and prepares the ground for feature extraction. Individual stages of the FR system are examined and an attempt is made to improve each stage. DWT and DCT are used for efficient feature extraction and a Binary Particle Swarm Optimization (BPSO) based feature selection algorithm is used to search the feature space for the optimal feature subset. Experimental results show the promising performance of the proposed techniques for face recognition on three benchmark face databases, namely, Color FERET, CMUPIE and Georgia Tech (GT) databases.