Ultrafast invariant face recognition

In this paper, we investigate human face recognition for facial images involving a high degree of in-plane and out-of-plane 3D distortions such as rotation. In the proposed technique, we investigated the problem of invariant face recognition by considering four different classes of images. A small subset of images is used for every class to create a composite image, which represents each class. The proposed technique is implemented by using fringe-adjusted filter based joint transform correlator (JTC) technique due to its superior performance over alternate JTCs and the feasibility of its implementation in the optical domain. Furthermore, the utilization of synthetic discriminant function to generate the composite image for enhancing the correlation output is also investigated. Computer simulation results are presented to show the validity of the proposed technique