多特征融合的手掌静脉匹配算法研究  

Palm Vein Matching Algorithm Based on Multiple Feature Fusion

作　　者：黎蔚[1] 王剑[1] 陈家新[1] 李继杰[1] 庄玉册[1] 

机构地区：[1]河南科技大学电子信息工程学院,河南洛阳471003

出　　处：《通信技术》2010年第4期155-157,共3页Communications Technology

基　　金：河南省科技厅自然科学基金资助项目(NO:0411010200)

摘　　要：针对相关系数法在手掌静脉图像匹配中无法解决平移与旋转的问题,提出了一种基于多特征融合的改进算法。首先,在匹配前,用文中提出的改进的图像校正算法校正图像,它可以改进相关系数法对于仅发生平移图像的匹配通过率;再利用基于细节的点的特征点匹配与基于全局的不变矩匹配的融合决策来弥补相关系数法对旋转图像匹配通过率偏低的不足。实验结果证明,本算法能提高系统的通过率,降低误识率。For the correlation coefficient method could not solve the problem of translation and rotation,in the palm vein image matching,a new modified algorithm based on multi-feature fusion is proposed in the paper.First of all,the translation,the rotation and the scaling of the image are corrected by the modified correction algorithm before the image match,thus improving the matching rate to the translated image.The fusion of feature points and geometric invariants could improve the deficiency of correlation coefficient method.Experiment indicates that the modified method could increase adoption rate of the system and reduce the error rate.

关 键 词：手掌静脉图像匹配 相关系数 特征点 几何不变矩 多特征融合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]