Research Of The Finger Vein Recognition Algorithm Based On Subspace Analysis And LBP

With the continuous improvement of science and technology,the traditional identity authentication technology is being replaced by the new biometric identification technology gradually,such as fingerprints,finger vein,face and iris.In addition,as the finger vein has advantages of good specificity and high security,it has attracted a wide attention and become a hot research topic in the field of biometrics recognition.This paper analyzes the research status of the finger vein recognition technology particularly,and makes an in-depth study of the finger vein recognition algorithm based on Subspace Analysis and the finger vein recognition algorithm based on Local Binary Pattern.The specific work in this paper is as follows:(1)The main contents of the finger vein recognition algorithm are reviewed.Firstly,we introduce the image-forming principle and the acquisition process of the finger vein image.Then we introduce the preprocessing steps,including ROI extraction,normalization and filtering.After that,the classification and the working principle of the finger vein feature extraction are briefly introduced.Finally,we elaborate the matching method and the performance analysis method of the finger vein recognition technology in detail.(2)We study the finger vein recognition algorithm based on Subspace Analysis,and then apply the Kernel Maximum Between-Class Margin Criterion(KMMC)on finger vein recognition.After that,we put forward the finger vein recognition algorithm based on two-directional KMMC further.Firstly,the finger vein recognition algorithms based on linear analysis methods,the Two-Dimensional Principal Component Analysis(2DPCA),the Two-Dimensional Linear Discriminant Analysis(2DLDA)and The Two-Dimensional Maximum Between-Class Margin Criterion(2DMMC),are applied on the pretreated finger vein image,and then the KMMC algorithm is used on the feature data further.Finally,we achieve the seven finger vein algorithms by MATLAB simulation experiments,2DPCA,2DLDA,2DMMC,KMMC,2DPCA+KMMC,2DLDA+KMMC,2DMMC+KMMC,and calculate the recognition rates of them for comparison.Experimental results show that the finger vein recognition algorithm based on two-directional KMMC has good recognition effect,and shortens the average matching time of the finger vein recognition algorithm based on KMMC effectively,and it has high recognition rate at the same time.(3)We study the finger vein recognition algorithm based on Local Binary Pattern,and then apply the Multi-scale Block Gabor Local Binary Pattern(MB-GLBP)algorithm on finger vein recognition.After that,we propose the finger vein recognition algorithm based on MB-GLBP and Logistic Mapping.Firstly,we extract the raw finger vein image feature by the MB-GLBP algorithm,and then scramble and encrypt the vein feature template by the Logistic mapping,which improve the security of the finger vein recognition algorithm effectively.Finally,with a lot of MATLAB simulation experiments,we get the optimal parameters of the MB-GLBP algorithm and the Logistic mapping,and achieve the finger vein recognition algorithm based on MB-GLBP and Logistic mapping.Experimental results show that the finger vein recognition algorithm based on MB-GLBP and Logistic mapping has high recognition rate and strong security.