Face And Iris Fusion Research, And Identify A Number Of Issues

In the booming information society, digital and recessive identity in the network which results that effective identity authentication techniques have been paid more and more attention to around the world.As single biometric identity authentication has its own shortage, which is hard to be conquered. Up to date, there is no single biometric system that could satisfy the requirement perfectly. Fusion several biometric features information, multimodal biometric system can offer a feasible method to solve the problems coming from single biometric system, and improve the performance of the whole biometric authentication system. Therefore, it's very significant to study multimodal biometric feature fusion and recognition methods.In this paper, combined with face recognition and iris recognition, fusion techniques of multimodal biometric system are studied. The main content includes:①Face feature extraction method is studied, the method of Fisher is employed, shortage of this method is distinct as it could not extract nonlinear features from face image. Kerner Fisher Discriminant Analysis(KFDA) is used to extract nonlinear features which is caused by noise effectively.②Wavelet is used widely as it can decompose images into high frequency signal and low frequency signal, then an approach of face recognition based on wavelet transform and KFDA is presented by which the dimension is reduced and the efficiency is improved.③In iris recognition, the method of iris feature extraction is deeply studied, an approach of iris feature extraction on energy-weighted is offered. The information of feature which is insensitive to noise is extracted.④The information fusion techniques of multimodal biometric system are studied, and wavelet transform and KFDA is discussed deeply. Meanwhile, combined with the characteristics of KFDA, fusion of face feature and iris feature at the feature extraction level is done, and a model of face and iris feature fusion recognition system is carried out. Experimental results show that, face and iris feature fusion is implemented efficaciously on the basis of the fusion strategy presented, and a higher correct recognition rate is gained. A new approach is supplied for multimodal biometric feature fusion and recognition.