High Performance Technique for Face Recognition Based on DCT

Abstract

Face recognition is an important and fundamental problem in computer vision, and there have been many attempts to address it. Correlation and high information redundancy in face images result in inefficiencies when such images are used directly for recognition. In this paper, an efficient hybrid approach to face recognition is presented, which combines image compression and neural network (NN) techniques together. The compression is achieved by applying fast discrete cosine transforms (DCTs) to reduce image information redundancy because only a subset of the transform coefficients are necessary to preserve the most important facial features such as hair outline, eyes and mouth. The compressed transform coefficients are used for back propagation NN classification. A high recognition rate can be achieved by using a very small proportion of transform coefficients. This makes DCT-based face recognition much faster than other approaches. The proposed system is implemented and tested using gray-scale images contains 40
distinct persons, each person having 10 different images and it gives good performance at high speed.