Face recognition in unconstrained environments

This paper proposes a novel face recognition system having good recognition performance through elaborate face region detection in unconstrained illumination environments. The proposed system consists of face and eyes detection, rotated-angle compensation, face region cropping, preprocessing, and classification modules as sequential steps. In particular, the elaborate face region is obtained from automatic face cropping procedure based on distance information between the eyes. The performance evaluation was carried out with various preprocessing images on the Yale B and the CMU-PIE databases. From the experimental results, we confirmed that the proposed method showed the best recognition accuracy compared to different approaches.