摘要
为实现分布式的遥感图像分类,达到计算的高性能,提出一种利用决策树SVM和网格技术实现遥感图像分类的解决方案。通过共享计算资源,将经预处理后的遥感数据切割成块并分别分配到网格计算节点进行并行计算。针对比例尺为1∶50 000的TM遥感图像的实验表明,此方案提高了分类效率,并为海量遥感数据的分类开辟了一条新的途径。
This paper proposed a new distributed image classification scheme of remote sensing in order to get a high performance computing, which was based on the decision tree SVM algorithms and grid technology. By sharing the computing resources in a distributed network environment, datas of the pretreatmented remote sending images would be cut into blocks and be assigned to the grid nodes doing parallel computing, thus realized distributed classification processing. The experimental data of a 1 ： 50 000 TM remote sensing image proves the validity of this scheme, and then provides a new method for the classification of mass remote sensing images.
出处
《计算机应用研究》
CSCD
北大核心
2009年第5期1944-1946,1954,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(40473029)
关键词
支持向量机
网格计算
遥感
图像分类
SVM（ support vector machine）
grid computing
remote sensing
image classification
