一种基于人眼结构特征的新颖虹膜定位算法  

Novel iris location algorithm based on the structure of human eyes

作　　者：苑玮琦[1] 林忠华[1] 徐露[1] 

机构地区：[1]沈阳工业大学信息科学与工程学院,辽宁沈阳110023

出　　处：《光电工程》2007年第1期112-116,125,共6页Opto-Electronic Engineering

基　　金：国家自然科学基金资助项目(60472088);辽宁省自然科学基金资助项目(20042028)

摘　　要：虹膜定位是虹膜识别系统的重要组成部分,定位的速度以及定位算法的健壮性是需要特别注意的因素。根据对人眼的结构特征进行分析,提出一种新颖虹膜定位算法。利用一种灰度累加算子找到瞳孔内一点,通过边缘检测模板找到虹膜内圆上三个不共线的点,从而快速而准确地定位出虹膜内圆,通过边缘检测模板找到虹膜外圆上四个不共线的点,从而快速定位出虹膜外圆。对已有的756幅虹膜图像组成的虹膜数据库进行定位,本算法只需要1.59min,平均定位时间达到0.126s,准确率达到99.07%。比Hough变换以及Daugman的圆检测算子等经典方法速度上有了很大的提高。而且保证了定位的准确率和算法的鲁棒性。Iris location is an important part of iris recognition, and speed of location and robustness of the location algorithm are factors which we should pay especial attention to. A novel iris location algorithm was proposed based on the analysis of the structure characteristics of human eyes. A point in the pupil was gotten by gray summing operator. Three inner boundary points which were not on the same line were found by the boundary detection template to locate the iris inner circle rapidly, and four outer boundary points that were not on the same line were found by the boundary detection template to locate the outer iris circle. 756 iris images comprising an iris image database were located. Our algorithm only costs 1.59min, average time of location reaches 0.126see and accuracy of location reaches 99.07%. This algorithm is much faster than any other classical methods such as Hough transform and the circle detection operator at speed. Furthermore, it ensures the accuracy of location and the robustness.

关 键 词：虹膜定位 边缘检测模板 灰度累加算子 模式识别 

分 类 号：O484[理学—固体物理] TP391[理学—物理]