Font Size: a A A

Research On Feature Extraction Algorithm Of Finger Vein Image

Posted on:2021-01-26Degree:MasterType:Thesis
Country:ChinaCandidate:M L DuFull Text:PDF
GTID:2428330629980259Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
As a kind of biometric recognition technology,finger vein recognition has been paid much attention.Compared with other recognition technologies with biological characteristics,finger vein recognition has become a research hotspot,which is detected in vivo,difficult to copy and forge,easy to be accepted,and more secure.Finger vein recognition technology mainly includes four steps: acquisition of digital vein image,preprocessing,feature extraction,matching and recognition.Among them,the extraction of digital vein features is a key step,which significantly affects the performance of the digital vein recognition system.This study mainly explores the extraction method of finger vein features,and proposes the extraction method of finger vein features based on the variable window width Gabor transformation in the space frequency domain and the time domain.The main work of the study is summarized as follows:(1)Finger vein recognition based on Gabor transform of variable window width.Aiming at the characteristic of vein image with abundant texture feature and frequency variation,a method of vein texture feature extraction based on air frequency analysis was proposed.Specifically,when setting the width of the window function,the width of the window function is dynamically adjusted according to the local difference excitation of the image,so that the window function can adjust the air frequency resolution adaptively.Secondly,according to the obtained real value,the energy sum of Gabor transformation coefficient is discretized,and the information entropy of each block is calculated to form the characteristics of finger vein.Compared with the existing methods of feature extraction with finger vein,the recognition rate of the method based on Gabor transformation of variable window width in FV-TJ and FV-USM databases reaches 100% and 99.39%,and the equal error rate is 0.2019% and 0.4743% respectively,the results indicate that the proposed method has better performance of recognition.(2)Weber local descriptors with variable curvature Gabor filter for finger vein recognition.To better recognize the finger vein image,according to the characteristics of the finger vein image line with different degrees of curvature,a Webber local descriptor(DCGWLD)based on variable curvature Gabor filter is proposed.First,according to the direction difference between the central pixel and the neighboring pixels,the differential excitation is improved.When calculating differential excitation,add direction information to Expanded the difference between different categories of finger vein images.Secondly,according to the different bending degrees of the line features in the finger vein image,the direction operator is improved.The improved Gabor filter is used to extract the double line features of the image with different curvatures and directions to enhance the robustness of translation and rotation.Finally,to better measure the similarity between features,the normalized correlation coefficient(NCC)scale matching score algorithm is used to further improve the recognition rate and reducing the error rate.To verify the recognition performance of the proposed method,the proposed variable curvature Gabor filter Weber local descriptors are compared in PolyU and SDUMLA-FV databases.Compared with the method proposed in this paper,the recognition rate reaches 99.89% and 99.42%,while the error rate reaches 0.6410%,0.7862%,which has more advantages and better recognition performance.
Keywords/Search Tags:Biometric recognition, finger vein recognition, feature extraction, local descriptors, Gabor transform, variable curvature
PDF Full Text Request
Related items