Face recognition and verification based on 2D circle technique

Abstract

This paper presents, how to recognize a face of portrait photographs by applying the principal components analysis (PCA) to get the feature vector and verification algorithm based on 2D circle. PCA was applied to modify the mean which is the cut of, 5% minimum and 5% maximum of data.  For verification process, each image was reduced into 2 dimension (2D). Those images of each individual in 2D are encircled with the diameter calculated form the maximum Euclidean distance between any two vectors in each pair of those images. In this paper, two experiment were conducted. We chose test image in training set, verified them by using circle and shows the results of the proposed algorithms and others in 5 distance measures for  the first experiment. Then, we also chose test image, which is not in training set and verify in the same way in the second experiment. We choose some images in Essex Faces94 (Female and Male Staff) to form a training set. The proposed algorithms show the better performance than other algorithms and they have recognition rate over 89%.