融合双向主成分分析的二维线性判别方法  

Two-dimensional linear discriminating method fused with two-way principal component analysis

作　　者：许爽[1,2] 索继东[1] 丁纪峰[2] 

机构地区：[1]大连海事大学信息科学技术学院,辽宁大连116026 [2]大连民族学院信息与通信工程学院,辽宁大连116605

出　　处：《大连海事大学学报》2011年第3期73-76,共4页Journal of Dalian Maritime University

基　　金：辽宁省教育厅科学技术研究项目(L2010094);中央高校基本科研业务费专项资金资助(DC10010103)

摘　　要：通过分析已有掌纹识别中特征提取的方法,提出一种融合双向主成分分析的二维线性判别方法.首先,对掌纹感兴趣区域的图像矩阵进行行和列双方向的二维主成分分析,消除图像中行和列的相关性,降低特征维数;然后,在其子空间内实现二维线性判别,得到最佳投影矩阵;最后,提取判别特征完成特征识别.实验结果表明,该方法提取速度快、识别率高、鲁棒性好.Two-dimensional Fisher linear discriminating（2DFLD） method fused with two-way principal component analysis（PCA） was proposed by analyzing feature extraction method of the existing palmprint recognition.Firstly,two-dimensional principal component analysis（2DPCA） was done to both rows and columns of the image matrix interested in the palmprint image region,which could eliminate the correlation between columns and rows,and feature dimensions were further reduced.Secondly,the 2DFLD was performed in new subspace,and the best projection matrix was obtained.Finally,the discriminating features were extracted to complete feature recognition.Experimental results show that the proposed method has faster extraction speed,higher recognition rate and better robustness.

关 键 词：二维线性判别(2DFLD) 二维主成分分析(2DPCA) 掌纹识别 特征提取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]