Research On Finger Vein Image Retrieval Method

With the rapid popularity of biometric,the biometric traits,such as iris,fingerprint and face,have been widely used for identity recognition in many applications due to their uniqueness and discriminative validity.In many biometric identification technologies,finger vein has some unique advantages compared with other biometrics,which mainly focus on four points:Firstly there is no need to worry about the duplication of finger vein by touching somewhere,as finger vein is under the skin;Secondly vein pattern can only be captured in a living finger,so vein pattern based recognition has high security;Thirdly the size of image capturing device is quite small and portable;and finally for most of us.everyone can get 10 finger vein patterns at most and all of them are discriminative,hence it is alternative and flexible to perform identity authentication.With these advantages,finger vein recognition attracts lots of research interests and it is regarded as one promising biometric technique.Many related studies have reported promising results in finger vein recognition,but it is still challenging to perform robust image recognition,especially in the application scenarios with large scale populations.With the purpose in consideration,this paper presents a binary search path of hierarchical vocabulary tree based finger vein image retrieval method.In detail,a vocabulary tree is built based on the local finger vein textons by the hierarchical k-means method.Each image patch is represented by the binary path in the search of its most similar leaf node,and the value of each bit in the path is labeled as 1 or 0 according to whether the corresponding node is passed or skipped in search.The similarity of two images is defined as the number of overlapped bits in all involved path pairs.And,the enrolled images with top t scores in the sorted score vector will be selected as candidates to narrow the search space.Experimental results on five finger vein databases confirm that the proposed method can improve the retrieval performance on both accuracy and efficiency.With ever-increasing data volume and access demand,the scale of biometric database is growing large and finger vein image retrieval is one significant technique for performing fast identification.However,most existing retrieval methods were based on fixed-scale feature of non-overlapped rectangular image block,in which the representation ability of feature and the local consistency of vein pattern were both overlooked.And the weak encoding(e.g.,predefined threshold based binarization)was also limited the retrieval performance.Focusing on these problems,this paper proposes a novel finger vein image retrieval framework based on similarity-preserving encoding of scale-varied superpixel feature.In the framework,locally consistent pixels in one superpixel are used as a unit of feature representation,and the feature length is varied with the category of the superpixel classified by the variance of lowest dimensional feature.Additionally,the feature compaction and feature rotation based encoding can minimize the quantization loss and preserve the similarity between the scale-varied feature and the encoded binary codes.Experimental results on six public finger vein databases demonstrate that the superiority of the proposed coding scale-varied superpixel feature based retrieval approach over the state-of-the-arts.