摘要
论文针对二维主成分分析法(2DPCA)表征信息不全面且系数多的不足,提出分块加权处理的双向2DPCA((2D)2PCA)方法提取人脸特征。该方法利用(2D)2PCA方法对人脸的各个分块提取特征,并对各分块的特征进行加权处理,然后应用支持向量机(SVM)实现分类识别。经过在ORL人脸库的实验研究表明,该方法压缩了人脸识别系数,缩短了识别时间,提高了识别准确率。
On the insufficient of incomplete face characterization information and many coeficient of 2DPCA （Two - dimensional Principal Component Analysis） ,the algorith of feature extraction based on modular weighted （2D）2PCA （Two -direction 2DPCA） is put forward .in this paper .First ,each sub block features were extracted in using （2D）2PCA method ,then the block features were weighted to realize feature fusion ,and last classification recognition was achieved by using SVM（Support Vector Machine） .The results of experiments on ORL face database show that the modular weighted （2D）2 PCA feature extraction method is compressed the amount of feature coefficient ,shortened the time of recognition ,improved the recognition accuracy rate .
出处
《激光杂志》
CAS
CSCD
北大核心
2013年第5期25-26,共2页
Laser Journal
关键词
二维主成分分析法
加权
特征融合
支持向量机
特征提取
Two - dimensional principal component analysis
weighted
Feature fusion
support vector machine
feature extraction
