Research On The Identification Of The Hand Vein Recognition Based On The Brain Computing

In recent years,as the rising star of biometric identification technology,hand vein recognition technology has attracted more and more attention from the academic and commercial circles for its characteristics of living,universality,uniqueness and invariance.The traditional recognition algorithm has a high recognition rate for normal dorsal vein images.But compared with the normal hand vein image,it is difficult for a scale,contrast,translation,rotation,and revealing the fingers,wrist and handle of the heterogeneous hand vein image to guarantee the accuracy of recognition.However,the human brain can still recognize objects quickly and accurately under the influence of uncontrollable factors,such as light,position and angle.Therefore,through the research of the working mechanism of the brain visual perception system,and by the latest research results of Neurology and anatomy,the heterogeneous hand vein recognition research based on brain computing is proposed.The visual information processing mechanism of the visual perception system of the brain is simulated by constructing the model.The innovation of this paper is as follows:(1)The images of heterogeneous dorsal vein images were defined,and then the database of heterogeneous dorsal vein images was established.The factors of scale,rotation and translocation were analyzed qualitatively.(2)The multi-hierarchy model based on brain-like computing is constructed to extract the characteristics of heterogeneous dorsal vein images.The brain visual perception system of the brain visual cortex of the V1 and V2 regions of the nerve cells process the visual information layer by layer to obtain the position,direction and other information of the visual information.The multi-level structure model is to simulate the visual information processing mechanism of the V1 and V2 regions of the visual cortex of the brain,and to extract the characteristics of invariantity such as translation,rotation and scale,and to ensure the characteristics of the extracted dorsal vein Sticky.(3)A deep belief network(Deep,Belief,Network,DBN)is established to classify and decide the robust heterogeneous hand vein features extracted from the hierarchical structure model.The visual cortex V4 and IT regions of nerve cells play an important role in the classification and identification of objects,and DBN simulates the visual information processing mechanism of the V4 and IT regions of the visual cortex of the brain to ensure that the heterogeneous hands Accuracy of vein image recognition.Aiming at the above proposed innovative methods,a large number of experiments show that the distance between the eigenvector in the first case of the multi-hierarchy model parameter setting is the smallest and the distance between the classes is the largest.At the same time,the eigenvector is used as the input of the deep belief network,and the recognition accuracy of the dorsal vein image is higher than that of other traditional algorithms,which verifies the validity of the method.