Binary Multiresolution Wavelet based Algorithm for Face Identification

A binary multiresolution wavelet based framework for face identification is presented in this paper. This paper proposes the feature extraction algorithm based on multiresolution information of face. Proposed feature extraction algorithm extract discrete wavelet transform based binary features to represent face images.DWT plays very important role in efficient feature extraction which results in the high recognition rates. Face images are decomposed upto fifth level using wavelet transform. After wavelet decomposition detail coefficients are coded in binary form. Hamming distance classifier is used for binary feature classification. Experimental results show the promising performance of the proposed technique on three face databases: ORL, JAFFE, IIT female database. Proposed wavelet based algorithm also causes reduction in feature vector size. The algorithm has successfully handled pose variances, expression variations, lightning condition variation. We have achieved recognition rate of 95% for ORL database, RR of 97.77% for JAFFE database and RR of 89.28% for IIT female database .The results of experiments with arbitrary variations in lightning, expression, poses and backgrounds show that the proposed methodology has proven to be promising technique