A new hybrid face recognition system

Abstract

This paper presents a method of implementing the proposed hybrid intelligent face recognition system to a standard compression system to reduce the computation complexity of DWT. The proposed system is combining DWT and RBF neural network to achieve the advantage of both of the methods. The image dimension is first being reduced and represented in the low frequency domain by DWT and the face features are being classified by RBF neural network. The simulation results on the proposed system indicate that a higher recognition rate is achieved as compared to PCA and PCA+LDA. Besides, the results also show that the computation complexity of the DWT face recognition system can be reduced when the proposed system is implemented in the standard compression system. Experiments are carried out on two aging probe sets of FERET database.