摘要
在二维主成分分析的基础上,考虑样本的流形分布特点,引入样本相似系数,重新定义了样本拉普拉斯散布矩阵,进而给出了基于拉普拉斯二维主成分分析的特征提取方法.在ORL,FERET人脸库上的试验证明了基于拉普拉斯二维主成分分析方法的有效性.
Based on two-dimensional principal component analysis, this paper investigates the features of manifold distrihution. Sample similarity coefficient is introduced to redefine Laplaeian scattering matrix, and thus a feature extraction method of this kind is worked out. The results of the experimenls conducted on ORL and FERET face database indieate that the method is effective.
出处
《南京工程学院学报(自然科学版)》
2009年第4期55-59,共5页
Journal of Nanjing Institute of Technology(Natural Science Edition)
基金
江苏省自然科学基金(BK2009352)
关键词
二维主成分分析
拉普拉斯
特征抽取
人脸识别
two-dimensional principal component analysis
Laplacian
feature extraction
face recognition