小波重构与局部DCT的二维主元分析掌纹识别  

A Novel Palmprint Recognition Algorithm Based on the Noise Characteristic of the Palmprint

作　　者：陈晓华[1] 李春芝[1] 蒋云良[1,2] 

机构地区：[1]湖州师范学院信息工程学院,浙江湖州313000 [2]浙江大学计算机科学与技术学院,杭州310027

出　　处：《光子学报》2009年第6期1566-1571,共6页Acta Photonica Sinica

基　　金：国家自然科学基金(60573056);浙江省自然科学基金(Z106335,Y107759,Y107293);浙江省科技计划项目(2008C21G2200007);广东省自然科学基金(07010869,032356);湖州市自然科学资金(2006YZ03)资助

摘　　要：为了解决掌纹特征提取过程中,手掌的非平面问题导致的伪主线噪音信息并简化运算,对重构的掌纹图像进行了局部离散余弦变换.解决了重构后图像噪音敏感性问题,有效地区分了掌纹主线、掌纹褶纹和乳突纹.由二维主元分析算法获得较稳健的识别特征.通过香港理工大学公布的PolyU掌纹数据库的实验,同二维主元分析算法相比,小波重构与局部离散余弦变换的2DPCA掌纹识别算法正确识别率较高,识别效率较高.In order to solve the problem that the palm-print image quality is influenced by the arch of the palm-print,which caused platn-print recognition tare to descend, the noise from reconstructed image was local DCT eliminated. The palm-print principle-line, plicate-line and papilla-line were distinguished efficiently. And 2DPCA obtained robust recognition features. Experimental results on HongKong PolyU palm-print experiments show that by the method reeogrintion tate and recognition efficiency is higher than that by 2DPCA.

关 键 词：掌纹识别 提升小波重构 局部离散余弦变换 二维主元分析 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]