| 基于特征选择的二维主分量分析 |
| |
| 引用本文： | 于建江,王琪,徐春明. 基于特征选择的二维主分量分析[J]. 计算机应用与软件, 2008, 25(2): 71-73 |
| |
| 作者姓名： | 于建江 王琪 徐春明 |
| |
| 作者单位： | 盐城师范学院计算机系,江苏,盐城,224002;盐城师范学院数学系,江苏,盐城,224002 |
| |
| 基金项目： | 江苏省教育厅自然科学基金 |
| |
| 摘 要： | 提出了一种基于分类性能的二维主分量特征选择方法.即将二维主分量分析中图像总体散布矩阵的特征向量在二维线性鉴别分析的目标函数上进行投影,并选择分类性更好的特征向量进行投影.另外,为了保持原有的二维主分量分析主特征的优点,对最后的投影特征向量进行组合,也就是最后的投影特征向量选取对图像重建和图像分类分别起着重要作用的特征进行组合.在XM2VTS标准人脸库上的试验结果表明,所提出的方法融合了两种具有互补性的图像并行特征,在识别性能上优于传统的二维主分量分析方法.
|
| 关 键 词： | 二维主分量分析 特征选择 特征融合 人脸识别 |
| 收稿时间： | 2006-01-23 |
| 修稿时间： | 2006-01-23 |
FEATURE SELECTION BASED TWO-DIMENSIONAL PRINCIPAL COMPONENT ANALYSIS |
| |
| Yu Jianjiang,Wang Qi,Xu Chunming. FEATURE SELECTION BASED TWO-DIMENSIONAL PRINCIPAL COMPONENT ANALYSIS[J]. Computer Applications and Software, 2008, 25(2): 71-73 |
| |
| Authors: | Yu Jianjiang Wang Qi Xu Chunming |
| |
| Affiliation: | Yu Jianjiang1 Wang Qi1 Xu Chunming2 1(Department of Computer,Yancheng Teachers' College,Yancheng 224002,Jiangsu,China) 2(Department of Mathematics,China) |
| |
| Abstract: | A new two-dimensional principal component analysis method based on feature selection is proposed. Firstly,the feature vectors of image total scatter matrix that have good classification performance are selected to project to the object function of two-dimensional linear discriminant analysis. In addition,to make the best use of the advantages of two-dimensional principal component analysis, the two kinds of vectors got from two-dimensional principal component analysis are combined into new feature vectors. The method is applied to XM2VTS face database,and the experimental result shows that the proposed method is more available than the traditional two-dimensional principal component analysis. |
| |
| Keywords: | Two-dimensional principal component analysis Feature selection Feature fusion Face recognition |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
