The Research Of Iris Location And Iris Matching

Iris recognition is the leading pattern of the recognition research, it is a novel authentication method based on biometric, it is also an authentication technology with widely prospect. It has important pertinent value in information and security field, it is concerned by the government and academia and industry.This thesis researched on the key technology of iris location and iris matching after summarizing previous work, then an improved sobel algorithm is proposed for the iris location algorithm, under the matlab6.0i did simulate and compare. Validity and accuracy of this algorithm is proved. At the same time a iris matching algorithm is proposed based on3-d curved surface with the C++, i did simulate and compare, experimental results verified this algorithm is feasible and practical for iris matching.Firstly, an improved sobel algorithm is proposed for the iris location algorithm, solving the problem of large amount of computation in standard iris location algorithm as well as location impreciseness. According to the geometric character of human eyes for iris location, select purposely the special region of sub-image as the target region, to which then applying the edge detection, projection for rough location, then use the improved sobel algorithm to fine position. This improved sobel algorithm not only use the advantages of the original algorithm interferrence, but also change the value of using gray value circle the maximum principle of superposition, make more accurate positioning, experimental results show that the algorithm with less computational cost can be accurately positioned inside an outside the boundary of the iris.Secondly, in order to solve the problems that the computation of the classic iris matching algorithm is large and the time-consuming is much,meanwhile, in order to overcome the complicated feature extraction. This article based on the principle of the average of the gray level difference of3-d curved surface,matching through the iris gray image directly, no feature extraction. The involved matching is not a set of data of the characteristic value, but the image of the entire iris region, based on the average of the gray level difference of3d curved surface, i built simulation platform of iris matching by c++, and gave the threshold at the same time and simulated. The results of the simulation shows that this algorithm not only has higher speed of the operation,but also has higher accuracy.Finally, the summary and the prospect on this research are given.