基于二维不可分小波相关性分析的虹膜识别  

Iris recognition based on correlation of 2-D nonseparable wavelet

作　　者：苑玮琦[1] 王浩[1] 

机构地区：[1]沈阳工业大学视觉检测技术研究所,辽宁沈阳110870

出　　处：《光电子．激光》2010年第4期593-597,共5页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金资助项目(60672078)

摘　　要：为了更有效地提取虹膜纹理特征区域和进一步减小虹膜特征的存储空间,提出了一种基于分块相关性分析的二维不可分B-样条小波的虹膜识别方法,通过对虹膜归一化图像进行二维不可分B-样条小波变换并提取小波系数特征,把这些特征等分成正方形的特征块并按照相关性由大到小排序,保留相关性大的特征块进行匹配。实验表明,本文算法比经典的虹膜识别方法能更准确地捕捉识别效果好的特征区域。In order to reduce storage space for saving iris feature and extract features efficiently for the iris texture, a correlation-based 2-D nonseparable B-spline wavelet transform approach is proposed for iris recognition. The proposed iris recognition algorithm extracts wavelet coefficients features from the normalized iris images by 2-D nonseparable Prspline wavelet transform, divides these features into square feature blocks equally, sorts the feature blocks in descending order, and preserves the feature blocks with large correlation value to match. Experiments show that the proposed algorithm can capture feature areas of good recognition performance more accurately than the classical iris recognition algorithms.

关 键 词：虹膜识别 小波 不可分 B-样条 相关性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]