Research On Key Technologies Of Identity Recognition Based On Hand Vein Images

In recent years,hand vein recognition technology has been widely concerned in the field of biometrics recognition on account of its advantages of vitality,uniqueness,stability,universality,hard to forge,and moderate cost;at the same time,because of its suitability for high-level safety requiring applications,so it is valuable to take a research.Hand vein recognition can be divided into three forms including palm vein recognition,back hand vein recognition and finger vein recognition technology.These have the same collection and recognition principles,but there are differences in collection equipment and application scenarios.Starting from finger vein recognition with small acquisition equipment,wide range of applications and low cost to make,this paper studies the image preprocessing,quality judgment in finger vein recognition,improvement in the identifiability of rotating offset images and the effective use of image features.The specific research contents and results are as follows:(1)In order to solve the problem of unclear or even missing texture in image acquisition,this paper proposes a method to judge image quality based on the cosine value of image gray level.The method adopts the form of grading strategy,which calculates the local gray value of the peak value to judge whether the image is exposed,and constructs a mean triangle based on the peak value to detect the continuity of vein texture information later.This method can effectively identify the high local brightness and lack of texture in the image.(2)Aiming at the problems of rotation,offset,and uneven illumination of some images in finger vein recognition technology,this paper proposes a method based on selected SURF feature points to increase the recognition accuracy.Firstly,this paper proposes a method that combines Gabor filtering and contrast-limited adaptive histogram equalization algorithm to enhance image.Secondly,the reflectivity of finger vein images is used for texture segmentation,and the Speeded Up Robust Features(SURF)in the vein images are detected at the same time;then the feature points are constrained according to the spatial distribution characteristics of SURF feature points in the segmented image.Finally,the similarity of feature points in the image is used for image recognition.The experimental results on the open finger vein database show that the proposed method not only has a high recognition rate,but also has excellent robustness against interferences such as finger rotation,finger offset and uneven image illumination during the acquisition of finger vein images.(3)Aiming at the problem of low utilization of finger vein image features,this paper proposes a finger vein recognition method based on feature progression.This method combines the characteristics of fast recognition speed of uniform local binary pattern(ULBP)and high accuracy of SURF feature recognition.First,the ULBP feature is used to retrieve a group of images that are most similar to the image to be tested,and then this group of similar images are compared to complete the recognition.Experiments on public databases have shown that the use of ULBP features as primary features can quickly and effectively perform pre-detection of the similarity of finger vein images;at the same time,then use SURF and ULBP features to perform weighted summation score recognition methods,which can achieve 99.16% and 99.59% recognition rates in the public databases SDUMLA and Ploy U.The false accept rate is 5.97% and 0.21%,and the rejection rate is respectively 1.15%,0.13%.(4)Design a finger vein recognition software on the Visual Studio platform based on above algorithms,which contains quality evaluation,information management and identification of the collected images,it can easily identify and judge the input finger vein images.