Face recognition in reduced Eigen-plane

This paper proposes a 2D face recognition technique in Eigen-Space with reduced feature vector size. In this paper image of each person are mapped to covariance Eigen plane to generate the feature vector. The technique is demonstrated to be a satisfactory real time face recognition approach as it can work on multiple types of face emotion instances and can handle varying degrees of occlusion. Moreover the feature representations of the faces are in scalar form and hence require low computational overheads. Absolute scalar difference is used for classification. The technique has achieved high recognition accuracy.