Saeed Karimi

In this thesis, a new combination of two face recognition algorithms is proposed. The singular value decomposition method based on left and right singular vectors (SVD/LR) is introduced. Also, to improve the linear discriminant analysis method, the weighted linear discriminant analysis method with QR analysis (RW­LDA/QR) is presented. The hybrid method, which is expressed by the combination of two methods SVD/LR and RW­LDA/QR, is used for face recognition rate and time reduction. At the end, numerical tests on two databases ORL and GTFD, with methods SVD/LR, RW­LDA/QR and the new combined method SVD/ LR+RWLDA/QR is compared with applying DWT versus without applying DWT for face recognition rate and time.