基于二维双向FLD的掌纹识别方法  

Palmprint recognition based on two-directional two-dimensional Fisher's linear discriminant analysis

作　　者：秦娜[1] 金炜东[1] 

机构地区：[1]西南交通大学电气工程学院,成都610031

出　　处：《计算机应用》2008年第8期2043-2045,共3页journal of Computer Applications

基　　金：国家自然科学基金资助项目(60572143)

摘　　要：采用二维双向Fisher线性判别分析对掌纹图像进行特征提取,即通过在水平和垂直2个方向上各执行1次二维Fisher线性判别分析,能消除掌纹图像行和列的相关性。运用Fisher准则选取更适合于分类的矩阵分量,将特征信息压缩到图像矩阵的左上角,缩小了特征的维数。测试结果表明,该方法具有更高的识别率和更低的计算复杂度。A new method named two-directional two-dimensional Fisher＇s Linear Discriminant analysis was proposed for palmprint image representation and recognition. The method perform 2DFLD twice, one was in horizontal direction and the other was in vertical direction, so pertinency of rows and columns in palmprint image was avoided. The Fisher criterion was used to select tl^e most discriminative features, and the discriminant information was compacted into the up-left comer of the image. Experiment results show that this method has higher recognition rate and is efficiency than the traditional LDA algorithms for palmprint extraction.

关 键 词：生物特征识别 FISHER线性判别分析 特征提取 掌纹识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]