FACE RECOGNITION BASED ON IMAGE INTEGRATION AND DCT ANALYSIS

Abstract

This paper aims at enabling computers to recognize faces without human intervention. This isaccomplished by capturing five images for the face under test, the information of five images areconcentrated in one integral image by using KL transforms. The wavelet transform was used todiminution the integrated image, DCT was applied on the first band of the wavelet transformedimage to extract the recognition features. The recognition features assumed to be driven from thefirst six DCT coefficients. The recognition was established by searching a match between thefeatures extracted from the test image and library of the face image models. The performance of thedesigned recognition system gave a promising verification percentage of about 83% in which thecontribution weights of the adopted features were unequally