第18卷第10期 2010年10月
文章编号
1004-924X(2010)10-2306-08
光学精密工程
Optics and Precision Engineering
Vol.18
No.10 Oct.2010
应用迭代圆环像素率法实现快速虹膜定位
程宇奇1,2，朱明"，李桂菊1，葛微1-2，陈彦平3
（1.中国科学院长春光学精密机械与物理研究所，吉林长春130033；
2.中国科学院研究生院，北京130039；3.长春工业大学电气与电子工程学院，吉林长春130012）摘要：针对虹膜定位易受噪声影响、速度慢、自适应性差等问题，提出了基于选代圆环像索率法的快速虹膜定位算法。提出的定位算法包括切割虹膜图像、Hough圆检测粗定位、微积分算子精定位3个部分，从切割虹膜图像、图像抽样、选代圆环像素率法、快速Hough圆检测以及分层定位思想5方面提高算法的速度。提出了缩小半径范围的送代圆环像素率法以及消除瞳孔光班的形态学方法，腹孔分割阔值以及小范围的圆心、半径候选集等参数都是通过计算得到，自适应性好，使用4个虹膜数据库进行实验，并与相近算法进行了对比。实验结果表明，该算法的定位准确率为97.75%～
99.07%，定位时间为52.847～158.502ms，是一种鲁棒、快速、自适应的虹膜定位算法。关键词：虹膜定住；微积分算于；Hough变换；选代围环像素率法；瞳孔光斑消除
中图分类号：TP391.4
文献标识码：A
doi;10. 3788/OPE. 20101810. 2306
Rapidirislocalizationbasedonmethodof
iterativepixelratiotocirquearea
CHENG Yu-qi'-, ZHU Ming', LI Gui-ju', GEI Wei'-*, CHEN Yan-ping
(l.Changchun Institute of Optics,FineMechanics and Physics,ChineseAcademy of Sciences,Changchun130033,China;2.GraduateUniversity of ChineseAcademy of Sciences,Beijing 100039,China; 3.School of Electric &Electronic Engineering,
ChangchunUniversity of Technology,Changchun 130012,China)
Abstract: Aimed at the problems of noise influence, high time consumption and bad adaptive perform ance for iris localization, a rapid iris localization algorithm based on the method of iterative pixel ratio to cirque area is proposed. The proposed method is composed of 3 parts: cutting iris image, coarse lo calization by circle detection of the Hough transformation and the precise localization by an integrodif-ferential operator. The methods of cutting the iris image, image sampling, iterative pixel ratio to cirque area, rapid circle detection of the Hough transformation and the layered localization theory are used to improve the algorithm speed. The method of iterative pixel ratio to cirque area is put forward to reduce the radius range, and the morphological method is used to eliminate the pupil faculae, The threshold of pupil segmentation, small-range candidates of circle centre and radius, etc. are all ob tained by the computation with good adaptive performance. Finally,four iris databases are applied in
收稿日期：2009-11-10；修订日期：2010-03-01.
基金项目：国家863高技术研究发展计划资助项目（No,2005AA778032）