基于序列图像提取稳定特征点的虹膜识别算法  

Iris recognition algorithm based on extracting stable feature points of sequence images

作　　者：苑玮琦[1] 张开营[1] 杨冉冉[1] 刘博[1] 

机构地区：[1]沈阳工业大学视觉检测技术研究所,沈阳110870

出　　处：《仪器仪表学报》2011年第5期1069-1076,共8页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(No.60672078);沈阳市科学技术计划项目(No.F10-213-1-00)资助项目

摘　　要：为了尽可能降低不稳定特征点对识别率的影响,研究提出了基于序列图像提取稳定特征点的虹膜识别算法。该算法首先用二维Gabor滤波器对序列虹膜图像提取特征编码,然后对该序列特征编码求交集以提取稳定特征点,并利用这些稳定特征点建立虹膜的特征模板库,最后通过计算相似度,获得识别结果。在实验室采集的序列虹膜图库上,当等错率为0.3017%,分类阈值为0.6402时,正确识别率可以达到99.73%。实验证明该算法是有效、可行的,并更好地提高了虹膜的分类精度和改善了虹膜的识别性能。In order to reduce the influence of unstable feature points on iris recognition,an iris recognition algorithm based on extracting stable feature points of sequence images is proposed.Firstly,this algorithm extracts feature codes from sequence iris images with two-dimensional Gabor filters.Secondly,the stable feature points can be extracted by calculating the intersection of the feature codes of the sequence images.At the same time,iris feature templates can be built based on these stable feature points;and the iris feature templates can be used for iris classification.At last,the identification results can be obtained with pattern recognition.When the equal error rate is 0.3017% and the distance threshold equals to 0.6402,the identification accuracy is 99.73% for the iris database acquired in laboratory by oneself.Experimental results demonstrate that the iris recognition algorithm based on extracting stable feature points of sequence images is effective and feasible,and the proposed algorithm can achieve high classification accuracy and improve iris recognition performance effectively.

关 键 词：生物特征识别 虹膜识别 序列图像 稳定特征点 特征模板 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]