基于加权Gabor滤波器的虹膜识别  

Iris recognition based on weighted Gabor filter

作　　者：李欢利[1,2] 郭立红[1] 王心醉[3] 李小明[4] 董月芳[3] 方艳超[1] 

机构地区：[1]中国科学院长春光学精密机械与物理研究所,长春130033 [2]中国科学院大学,北京100039 [3]中国科学院苏州生物医学工程技术研究所,江苏苏州215163 [4]中国航空工业集团公司洛阳电光设备研究所,河南洛阳471023

出　　处：《吉林大学学报（工学版）》2014年第1期196-202,共7页Journal of Jilin University:Engineering and Technology Edition

基　　金：国家自然科学基金项目(51205268)

摘　　要：为了有效地提高虹膜识别系统的识别率,对Gabor滤波器进行了研究,并提出了基于层次分析法的加权Hamming距离的虹膜识别方法。首先,用不同尺度和方向的奇对称Gabor滤波器对归一化后的虹膜滤波生成特征图像;然后,对特征图像进行Hamming距离匹配得到不同滤波器的正确识别率和等错率,根据鉴别能力,选取合适的尺度和方向进行加权Hamming距离计算,其权值用层次分析法计算;最后,利用加权后的距离进行识别。其中CASIA2.0和MMU1图像库的等错率分别降低了0.64%和0.56%。实验结果表明:本文方法具有很好的识别性能和鲁棒性。In order to improve recognition rate of the iris recognition system, the Gabor filter is studied, and the weighted Hamming distance based on analytic hierarchy process is proposed. First, the feature image is generated through different scales and directions of the odd symmetric Gabor filter. Then, the recognition correct rate and error rate are acquired by Hamming distance matching. The appropriate scale and direction are selected to calculate the weighted Hamming distance. The weight is calculated by analytic hierarchy process. Finally, the weighted Hamming distance is used to iris recognition. The equal error rate of CASIA2. 0 and MMU1 are decreased by 0. 64% and 0. 56%, respectively. The experimental results show that this method has better recognition performance and robustness.

关 键 词：计算机应用 虹膜识别 GABOR滤波器 HAMMING距离 层次分析法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]