一种具有姿态鲁棒性的三维人耳识别方法  

A 3D ear recognition method with pose robustness

作　　者：王凯[1] 穆志纯[1] 

机构地区：[1]北京科技大学自动化学院,北京100083

出　　处：《中国科技论文》2013年第10期1029-1034,共6页China Sciencepaper

基　　金：国家自然科学基金资助项目(60973064);高等学校博士学科点专项科研基金资助项目(20100006110014);北京市自然科学基金资助项目(4102039);北京市教育委员会重点学科共建项目(XK100080537);中央高校基本科研业务费专项资金资助项目(FRF-SD-12-017A)

摘　　要：针对现有三维人耳识别方法在姿态变化情况下性能下降明显的问题,提出了三维人耳沟回结构特征以及相应的三维人耳识别方法。通过曲面变化量对曲面的凸凹特性进行度量,进而提取人耳关键生理部件组成的沟回结构信息。利用提取的沟回结构特征进行三维人耳的迭代最近点(iterative closest point,ICP)粗对准,并进一步进行三维人耳ICP精对准。在UND生物识别图像库集合F和集合G上的实验显示,在姿态变化情况下该方法具有较好的鲁棒性,同时也取得了较现有基于ICP的三维人耳识别更高的识别率,用时更短。The performance of existing 3D ear recognition methods is degraded sharply due to pose variation. A novel 3D ear rep- resentation called 3D auricle structural feature （3DASF） and the corresponding 3D ear recognition method with pose robustness are proposed. 3DASF, which is extracted by measuring ear surface characteristics with surface variation, contains key ear forma- tion, and then it is used to coarsely align gallary-probe ear pairs by iterative closest point （ICP） algorithm. A fine alignment is followed to get the alignment errors for identity recognition. Experimental results conducted on University of Notre Dame （UND） biometric datasets collection F and collection G outperform the state-of-the-art 3D ear recognitions based on ICP. The results also demonstrate that the proposed method is more robust to pose variation than the state-of-the-art methods.

关 键 词：生物特征识别 人耳识别 三维特征提取 迭代方法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]