Illumination Invariant Novel Approaches for Face Recognition

Most of current identification and verification systems like access control and surveillance require reliable person identification technique. The primary goal of such systems is that resources are accessed only by legitimate users. Inefficient person identification systems may become vulnerable thereby allowing impostors. Under these circumstances biometric systems such as automated face recognition finds its immense application. In this paper, we investigate the use of five novel approaches for learning lowdimensional representation of a face image using the concept of transmutation and its variants. Experiments were conducted on two standard datasets namely ORL and Grimace and comparisons among the proposed algorithms have been carried out. The results revealed that transmutation method outperforms its variants. 