摘要
针对人脸识别问题,提出一种基于多特征联合稀疏表示的方法。首先,分别采用主成分分析(PCA)、核主成分分析(KPCA)和非负矩阵分解(NMF)提取人脸图像的特征矢量。三种特征从线性、非线性以及非负表示三种层面描述了人脸图像的特征。在分类阶段,采用联合稀疏表示对三种特征进行综合决策,考察它们的内在关联。最后,基于三种特征的整体重构误差判定测试样本的类别。实验中,基于AR和Yale-B人脸数据库对提出方法进行性能测试。结果表明文中方法的有效性。
For the face recognition problem,a method based on joint sparse representation of multiple features is proposed.Firstly,PCA,KPCA and NMF are used to extract feature vectors from face images.The three features describe the characteristics of face images from linearity,nonlinearity and non-negativity.In the classification stage,joint sparse representation is used to classify the three features thus considering their inner correlations.Finally,the total reconstruction error of the three features are calculated to determine the target label.In the experiments,performance tests are conducted on AR and Yale-B face databases.The results show the effectiveness of the proposed method.
作者
申杨
SHEN Yang(Information and Communication Branch of State Grid Liaoning Electric Power Co.,Ltd.,Shenyang 110006,China)
出处
《信息技术》
2019年第9期154-157,162,共5页
Information Technology
