Research On Iris Recognition Algorithm

Biometric identification method has better performance in such aspects as safety, real-time and accuracy than the traditional way of identity authentication.Among them iris recognition has the highest reliability, stability and is the most extensive application. But the iris identification equipment cost is higher, the iris recognition algorithm also has some defects, all thesemake the iris recognition is not implemented on a large scalein the actual production and life.In this paper, we study of the iris recognition algorithm, hope that through this paper work, promote the iris recognition in applications of requiring higher security.Improvementbased on the iris recognition algorithm has the following several aspects:(1) In the iris edge extraction, aiming at the problem of edge points is easy to be lost, more template edge detection algorithm is proposed. This algorithm improved the classical edge detection operatorsdefects in the direction of the edge;(2) In the outer boundary location, aiming at the problem of boundary parameter search range is too wide and the amount of calculation is toolarge, based on the scaling and cut out image of iris boundary localization algorithm is proposed. On the guarantation of the accuracy, greatly reduces the search range of Hough transform, improve the real-time performance of the system;(3) In the iris image normalization part, due to the complexity of the iris texture information closely related to radius, the larger the radius, the less likely the information contained in, therefore, puts forward the iris normalization algorithm based on partition of different sampling rate. Improved rubber paper model normalization algorithm and thealgorithm of single sampling rate to determine the deficiency of sampling point;(4) In the iris texture feature extraction part, on the direction the iris texture performs like edges of different directions, so each pixel point need to usethe binaryedge information in different directions to represent. In this paper, an algorithm based on local texture edge is advanced.This iris feature extraction algorithmsolves the problem of losing of characteristics, complicated operations, poor anti-jamming, high storage requirementsin the classicalgorithm;(5) In the feature matching part, adopted the feature matching algorithm which can eliminate rotation and noise. Solved theinaccurate problem cause to rotation changes and noise effects.The above algorithms are all proved by simulation, the results show that the presented algorithm is effective.