基于数据预分析的虹膜识别方法  

An Iris Recognition Method Based on Data Pre-analysis

作　　者：宋辉[1] 刘奉华 SONG Hui;LIU Feng-hua(School of Information Science and Engineering,Shenyang University of Technology,Shenyang 110870,China)

机构地区：[1]沈阳工业大学信息科学与工程学院,辽宁沈阳110870

出　　处：《计算机技术与发展》2018年第8期58-61,85,共5页Computer Technology and Development

基　　金：国家自然科学基金(61271365)

摘　　要：随着虹膜识别技术应用的加深,出现了诸多影响虹膜系统正确识别率的问题;其中,采集到的虹膜图像本身不满足识别条件,是导致识别率不理想的一种原因。将正确匹配和未正确匹配的虹膜图像分别作为两类不同的图像样本,采用提取虹膜图像的灰度共生矩阵和LBP(local binary pattern)特征作为虹膜图像特征数据,将特征数据作为训练集,对采用的支持向量机分类器的预分类器进行训练,并用不同的虹膜图像进行测试;最后将训练好的支持向量机分类器放到虹膜识别匹配过程中,进行图像预分类,筛选出不满足匹配识别条件的虹膜图像,然后采用相同的识别算法对虹膜图像进行匹配识别。实验结果表明,采用相同的识别匹配算法,能有效提高虹膜识别系统的正确匹配率。With the application of iris recognition technology deeply,there are many problems in the system of iris recognition rate,one of which is that iris image collected by itself does not meet the conditions of the recognition,so the recognition rate is not ideal.The iris image with correct matching and incorrect matching is respectively adopted as two different types of image samples,and the gray level co-occurrence matrix and LBP feature from extracted iris image as iris image feature data.The feature data,as training set,is applied to train the pre-classifier of support vector machine based classifier,and the test is carried on with diverse iris image.Finally the trained support vector machine classifier is in iris recognition and matching for image pre-classification,and the iris image without meeting the conditions of matching and recognition is selected.Then the same matching recognition algorithm is used for iris image recognition.The experiment shows that this method can effectively improve the correct matching rate of iris recognition system.

关 键 词：虹膜系统识别正确率 灰度共生矩阵 局部二值模式 支持向量机 预分类器 

分 类 号：TN919.81[电子电信—通信与信息系统]