|
|||||
|
|
| 基于图像抽样重组的2维核鉴别分析 | |
| 引用本文： | 程正东,樊祥,章毓晋.基于图像抽样重组的2维核鉴别分析[J].电子与信息学报,2009,31(12):2958-2962. |
| 作者姓名： | 程正东 樊祥 章毓晋 |
| 作者单位： | 1. 脉冲功率激光技术国家重点实验室(电子工程学院),合肥,230037;电子工程学院,合肥,230037;清华大学电子工程系,北京,100084 2. 脉冲功率激光技术国家重点实验室(电子工程学院),合肥,230037;电子工程学院,合肥,230037;中国科技大学六系,合肥,230027 3. 清华大学电子工程系,北京,100084 |
| 摘 要： | 2维核鉴别分析(2DKDA)存在离散度量矩阵过大而无法计算的问题。该文通过将图像抽样重组与2DKDA的结合，提出了3种基于图像抽样重组的2DKDA(SR2DKDA)，它们不仅克服了2DKDA在计算上的困难，识别性能也优于2维线性鉴别分析(2DLDA)。在ORL人脸库和UMIST人脸库的实验验证了SR2DKDA的有效性。 |
| 关 键 词： | 2维线性鉴别分析 2维核鉴别分析 图像抽样重组 抽样重组2维核鉴别分析 |
| 收稿时间： | 2008-12-8 |
| 修稿时间： | 2009-9-25 |
2-Dimensional Kernel Discriminant Analysis Based on Image Sampling and Regrouping |
|
| Cheng Zheng-dong,Fan Xiang,Zhang Yu-jin.2-Dimensional Kernel Discriminant Analysis Based on Image Sampling and Regrouping[J].Journal of Electronics & Information Technology,2009,31(12):2958-2962. | |
| Authors: | Cheng Zheng-dong Fan Xiang Zhang Yu-jin |
| Affiliation: | State Key Laboratory of Pulsed Power Laser Technology (Electronic Engineering Institute), Hefei 230037, China; Electronic Engineering Institute, Hefei 230037, China; Electronic Engineering Department of Tsinghua University, Beijing 100084, China; Science and Technology University of China, Hefei 230027, China |
| Abstract: | 2-Dimensional Kernel Discriminant Analysis (2DKDA) can not be performed since its scatter metric matrices are too large. This paper combines the sampling and regrouping images with 2DKDA and gives three kinds of Sampling and Regrouping 2-Dimensional Kernel Discriminant Analysis (SR2DKDA). These algorithms not only overcome the drawback of 2DKDA but also have superior recognition accuracy to 2-Dimensional Linear Discriminant Analysis (2DLDA). The experiments on ORL database and UMIST database verify the efficiency of the SR2DKDA. |
| Keywords: | 2-dimensional linear discriminant analysis 2-dimensional kernel discriminant analysis Image sampling and regrouping Sampling and regrouping 2-dimensional kernel discriminant analysis |
| 本文献已被 万方数据 等数据库收录！ | |
| 点击此处可从《电子与信息学报》浏览原始摘要信息 | |
| 点击此处可从《电子与信息学报》下载全文 | |