改进的算法在虹膜识别中的应用研究  

Research on Application of Iris Recognition Based on Improved Algorithm

作　　者：许威[1] 权芳芳[2] 姜媛媛[1] 李陈陈[1] 皇淼淼[1] 

机构地区：[1]安徽理工大学电气与信息工程学院,安徽淮南232001 [2]安徽理工大学理学院,安徽淮南232001

出　　处：《计算机技术与发展》2013年第3期34-37,共4页Computer Technology and Development

基　　金：安徽省教育自然科学基金(KJ2012B062)

摘　　要：文中通过虹膜图像的直方图选取合适的阈值,对虹膜图像进行二值化处理。采用基于几何灰度投影的方法粗定位虹膜内边界,使用改进的Daugman定位算子定位外边界,再用最小二乘圆拟合方法对内外边界进行圆拟合,实验结果表明此种算法使得虹膜定位精确度较高,定位速度快,能较好地满足实时性的要求。同时提出了一种基于线段提取的虹膜归一化方法。在特征提取方面,采用基于二维Log-Gabo滤波的方法通过多尺度多方向构成的多通道滤波器提取虹膜特征信息,减少了虹膜编码的长度及特征匹配的时间,匹配准确率较高,能较好地提高虹膜识别的准确率。The appropriate threshold is obtained through the histogram of the iris image, which is used in iris binary processing. The iris inner boundary is orientated roughly based on geometric graylevel projection method and the iris external boundary is orientated by the improved Daugman localization operator, both of which are fitted accurately by the method of the least squares round fitting. The experi mental results show that the localization algorithm has the advantages of high positioning accuracy and high positioning speed, and can satisfy the requirement of realtime. Meanwhile, put forward an iris normalization method which is based on line extraction. In the feature extraction, iris feature information is extracted by the method of 2D LogGabo filter which can generate multichannel filter though multi scale and multiple directions, the method also can reduce the length of the iris code and the time of feature matching,improve the feature matching accuracy, can effectively improve the accuracy of iris recognition.

关 键 词：虹膜定位 Daugman定位算子 虹膜归一化 二维Log—Gabo滤波 虹膜识别 

分 类 号：TP31[自动化与计算机技术—计算机软件与理论]