Research On Dorsal Hand Vein Recognition Algorithm Based On Feature Fusion

With the rapid development of Internet technology,more and more attention has been paid to the security of personal information.For this reason,it is very important to identify the safe and reliable technology.The emergence of biometric technology allows the information protection industry to see new hope.Dorsal hand vein recognition as one of the biometric identification of the most popular research,with living body recognition,internal features,non-contact acquisition mode and higher security,attractting the attention of a large number of scholars.Therefore,it is of great significance to conduct in-depth study.Using the dorsal hand vein recognition technology to determine the identity.First,we must obtain a clear image of the dorsal hand vein.Simple processing of the acquired image to reduce the impact of redundant information.Then,extracting which can represent the information of the hand vein image,and finally use the classifier to identify the information of each hand vein image,and complete the personal identification.The main contents of this paper are as follows:This paper presents a dorsal hand vein image acquisition device,a database of hand vein image,including 200 kinds of dorsal hand vein images collected at different time with different age and sex of the volunteer,each kind of the left and right dorsal hand vein images is 5,a total of 2000 pieces of hand vein image.This paper presents a method for extracting interest maximum inscribed rectangle maximum inscribed circle sense based on region,to solve the traditional centroid extraction effective area error problem,to obtain more stable region of interest.This paper modifies the dorsal hand vein recognition method based on bilateral two-dimensional linear discriminant analysis algorithm.For bilateral two-dimensional linear discriminant analysis,when the class mean ang the global mean are close,it is hard to identify.A modified method is proposed.It integrates the cluster information in each class by redefining the between-class scatter matrix.This paper presents a fusion method based on D-S evidence theory.Two kinds of methods,which are the local binary pattern and the modified bilateraltwo-dimensional linear discriminant analysis,are fused.The distance matching values of the two methods are converted into the basic probability assignment function of each sample,and the final reliability value is obtained by the D-S evidence combination formulation.According to the probability of conflict in evidence,this paper puts forward an improved evidence combination formula,which is based on the weighted distribution of the probability of conflict to each sample according to the source of evidence.This paper designs a set of software platform of hand vein recognition based on MATLAB GUI interface is designed.