一种改进的最大间距准则人脸识别方法  

An Improved Face Recognition Method Based on Maximum Margin Criterion

作　　者：刘亚亚[1] 程国[1] 

机构地区：[1]商洛学院数学与计算科学系,商洛726000

出　　处：《科学技术与工程》2013年第9期2566-2571,共6页Science Technology and Engineering

基　　金：陕西省教育厅科研计划项目(11JK0512);陕西省教育科学"十二五"规划课题(SGH12443);商洛学院科研基金项目(11SKY003;12SKY010);商洛学院教育教学改革项目(10jyjx02006)资助

摘　　要：针对最大间距准则在人脸特征提取过程中的不足,提出一种统计不相关的加权最大间距准则人脸特征提取方法。首先对最大间距准则的类间散度矩阵和类内散度矩阵加乘权函数。然后在准则函数中利用双参数调节类间散度和类内散度对特征抽取的影响力。最后通过Schmidt正交化得到统计不相关的最佳鉴别矢量集。在ORL和Yale人脸图像库上的仿真实验结果表明,克服了最大间距准则的缺点,提高了人脸识别率。According to the shortage of maximum margin criterion in the face feature extraction process, a sta- tistical uncorrelated weighted maximum margin criterion face feature extraction method is presented. Firstly, weight function by class between divergence matrix and class inside divergence matrix is added. And then double parame- ters in the criterion function to control the influence of the feature extraction on class between divergence and class inside divergence are used. Finally statistical uncorrelated optimal identification vector set through the Schmidt orthogonalization got is. Experimental results on ORL face database and Yale face database show that this method overcomes the shortcomings of the maximum margin criterion, and improve the face recognition rate.

关 键 词：人脸识别 加权最大间距准则 统计不相关 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]