Performance Comparison of Kekre's Transform with PCA and Other Conventional Orthogonal Transforms for Face Recognition

In this paper we propose fast face recognition system based on the Kekre's transform. This algorithm can be easily implemented and number of coefficients required for recognition reduces drastically compared to benchmark algorithm PCA. Thus computational burden decreases. We have compared the performance of this transform with conventional transforms like DCT, DST, Slant transform and WHT. The algorithm is tested for occlusion as well as different percentage of image energy levels. This transform needs minimum energy as compared to other transforms for reliable recognition. Here we have tested proposed scheme on standard ORL database and locally generated unconstrained database. ORL database gave the accurate results and local database gave accuracy of 80% which is higher than conventional transforms and PCA.