Research On Deformation Of Finger Vein Recognition Method

With the fast development of information era,the biometric identification technologies have gradually become the mainstream of development due to high reliability and uniqueness.Among the many biometric techniques,finger vein and other biometrics(such as face,fingerprint,palmprint,iris,gait,etc.)have unique advantages.At present,many research work has achieved good results in the field of finger vein recognition.However,the images can be captured without contact,which means the fingers can be placed arbitrarily,causing homologous images have a great difference,which undoubtedly finger vein image identification poses a great challenge.All in all,this non-contact,unlimited imaging and the large gap between the light source and the camera cause finger placement to vary widely,further creating the problem of finger vein image deformation.If the problem of finger vein image deformation cannot be solved well,the similarity between the homologous images will be reduced,which lead to the false rejection of recognition and eventually affect the recognition performance of the whole system.Therefore,it is very significant to solve it effectively.In view of the above-mentioned finger vein image deformation,we analyze the causes of finger vein image deformation and provide a deformation-based finger vein recognition framework.Specifically,the vein is first segmented in preprocessing and vein PCA-SIFT feature is extracted for each vein point.Because the SIFT feature is a 128-dimensional feature,it is necessary to reduce dimension of SIFT feature by PCA,thus saving the running time of the program and improving the execution efficiency of the algorithm.Finally,the extracted vein PCA-SIFT feature is matched by bidirectional deformable spatial pyramid to obtain the final matching score.In addition,since the robustness of methods to image deformation cannot be exactly evaluated on the existing databases,we build a finger vein deformation database SDUD to imitate all kinds of image deformations in a real application.The experimental results,on the self-built deformation database and one public database,prove the effectiveness of the proposed framework for dealing with the image deformation problem.Based on the overview of the deformation-relevant work,we can see that most methods are related to feature extraction and matching,and only a few methods are performed in preprocessing.Therefore,we presents a geometric analysis based detection and correction(GADC)method for deformable finger vein recognition.Firstly,the geometric shape of the finger is analyzed based on the statistics.Then,the deformed finger vein image is detected based on the variation of the finger geometric shape.Lastly,the deformed image is corrected by a linear transformation and a nonlinear transformation.The experimental results,on two opened finger vein databases and one self-built deformation database,show that our approach can improve the recognition performance of the deformed finger vein images.