Face recognition: A template based approach

Abstract

In this paper, we proposed a template based face recognition approach. Here we compared our approach with the holistic feature based approach Principal Component Analysis (PCA). PCA is a statistical feature based approach works on Eigen space. PCA is a simple approach for face recognition of only frontal faces and proposed system is based on grey level template matching. To know the greatness of proposed system we have done experiment and compared with existing systems in a systematic way to check the performance of the systems. We observed that the correctness or efficiency of recognition rate using PCA is only about 70-75%, PCA was not able to recognize the faces if there is change in illumination, pose, in-plane rotation, noise etc.,. in the query input image. Where as for template matching we observed that got better results (more than or equal to 20%) than PCA, template matching recognition process can recognize the faces efficiently and invariant to all above factors.