决策树方法在遥感信息提取中的应用  

Application of Decision Tree Method in Remote Sensing Information Extraction

作　　者：陈晔[1] 何政伟[2,1,3] 龚兵[4] 

机构地区：[1]成都理工大学地球科学学院,四川成都610059 [2]地质灾害防治与地质环境保护国家重点实验室,四川成都610059 [3]首都师范大学资源环境与地理信息系统北京市重点实验室,北京100037 [4]西北核技术研究所,陕西西安710024

出　　处：《地理空间信息》2012年第1期67-69,72,共4页Geospatial Information

基　　金：国家863计划资助项目(2007AA120306)

摘　　要：介绍了决策树分类器的结构和理论基础,并结合了遥感影像各波段的特征对地物进行分类。在此基础上,以重庆渝北区为例进行了分类实验。实验说明了实现分类的具体过程,结果表明其方法在依据感兴趣区训练样本进行影像分类是切实可行的。Firstly,classifier structure of decision tree and its theoretical basis was introduced.Secondly,classification was carried out for landuse types with the characteristics of remote sensing image in each band.Then,as a case study,the classification experiment was conducted over the area of Yubei in Chongqing.The specific process of implementing classification was demonstrated,and the results showed that the method was practical and feasible to make the classification of remote sensing image based on some training samples of interested area.

关 键 词：决策树 多叉树与二叉树 CART算法原理 分层信息提取 

分 类 号：P237.3[天文地球—摄影测量与遥感]