基于类别可分离性的遥感图象特征提取方法
A Feature Extraction Algorithm for Remote Sensing Image Classification Based on Class Separability
-
摘要: 本文从类别可分离性出发推导出一种用于遥感图象特征提取的线性变换方法.该方法并 不着眼于变换子空间的整体信息保持,而在于使当前待分类别具有最好的分离度.文中给出 了变换核的理论推导及计算方法.试验结果表明该方法具有良好处理效果.Abstract: This paper develops a linear transformation algorithm for feature extraction based on class separability. The algorithm takes the separability of the current classes as objective function instead of maintaining perfect information of the original image. The derivation of the transformation kernel function and the computational method are given. Experimental results show the effectiveness of the algorithm presented.
-
Key words:
- Feature extraction /
- linear transformation /
- separability /
- eigenvalue and eigenvector
-
计量
- 文章访问数: 2444
- HTML全文浏览量: 158
- PDF下载量: 1127
- 被引次数: 0
下载:
