Robust Recognition Of Hand Vein Images In The Cross Database

With the rapid development of Internet plus economic model,the network brings convenience to modern life,accompanied by the identification of increasingly urgent demand,challenges the new technologies of identity recognition.Especially in the distributed vein recognition system,there are some differences in the quality of the acquired hand vein image will inevitably reduce the accuracy of identification,so that the identification system can not meet the needs of the times.Although the excellent biological properties of hand vein has created a good opportunity,but how to recognise the low constraint images smoothly and accurately is particularly important.In this paper,based on the different equipment collected under the cross database of hand vein image,the main innovative work of this paper is summarized as follows:(1)According to the main influence factors of image quantitative analysis,a morphological closing computation operator is applied to remove the handle and background noise,use the maximum inscribed circle to normalize the size of hands,use the position of the knuckle to normalize the angle,and the method of linear function is used to normalize the gray scale.It can unify the quality standard of image,reduce the difference of image,and ensure the robustness of recognition.(2)An optimization method based on SIFT operator is proposed on the three aspects:scale parameter,neighborhood structure and matching threshold,the extraction quantity and matching number of key points are increased,and the recognition rate of back vein images is improved,and the robustness of recognition is improved.Based on the above work,the average recognition rate of the hand vein image of the cross database was increased from 47.83%to 89.83%,which proved the validity of the work.