Face Recognition across Poses Utilizing Feature Transformation

Face recognition is an advanced identification solution which can meet the crying needs in security areas. Pose effect is a big challenge for robust applications of this technology. We proposed a feature transformation approach to cope with the head rotation roughly within half profile view. Comparing with algorithms based on computer vision technology, the proposed feature transformation method is not dependent on heavy computation and easily to apply in live conditions. Popular feature extractions, least square (LS) and total least square (TLS) solution in calculating as well as some properties of transformation were explored on the FERET database