Enhanced face recognition using 8-Connectivity-of-Skin-Region and Standard-Deviation-based-Pose-Detection as preprocessing techniques

Face appearance drastically changes under varying background, pose and illumination conditions. Face Recognition (FR) in such varying conditions becomes a difficult and challenging task. In this paper, we propose three novel techniques, viz., Face Detection based on 8-Connectivity-of-Skin-Region (FDCSR), Standard Deviation based Pose Detection (SDPD) and Gamma Ray Burst Rhombus Star (GRBRS) feature mask to improve the performance of FR systems. FDCSR is used as a preprocessing step to remove cluttered background from the image. SDPD is also a preprocessing step where pose neutralization technique is employed. GRBRS feature mask on the Fast Fourier Transform (FFT) of the preprocessed image is used to extract the salient features of the face. Binary Particle Swarm Optimization (BPSO) feature selection algorithm is used to search the feature vector space for the optimal feature subset. Experimental results show promising performance of the proposed techniques for FR on four benchmark face databases, namely, Color FERET, CMUPIE, HP and FEI.