分类器和时相结合的流域下垫面分类方法  

Basin underlying surface classification based on classifiers and phase combined

作　　者：陈贝贝[1] 颜梅春[1] 李致家[2] 阮仁宗[1] 王元超[3] 

机构地区：[1]河海大学地球科学与工程学院,南京210098 [2]河海大学水文水资源工程学院,南京210098 [3]中铁第一勘察设计院集团有限公司航测遥感处,西安710043

出　　处：《测绘科学》2014年第9期141-144,共4页Science of Surveying and Mapping

基　　金：国家自然科学基金(41130639);中央高校基本科研业务专项(2009B11714)

摘　　要：文章探讨提高卫星遥感大范围流域下垫面分类方法:通过集合贝叶斯分类器和决策树分类器的优势,充分利用TM影像覆被信息,结合影像时相动态信息以获得分类准确的土地覆盖/利用类型;再结合DEM生成的坡度信息得到下垫面类型,最终确定水文地理单元。以河南省的东湾流域为例进行验证,结果表明:贝叶斯法和决策树法分类各有优势,两者结合可以获取更准确的土地覆盖/利用分类结果,辅以时相类信息可以进一步修正类别间的混淆。This article combined Bayes classifier and decision tree classifier, made full use of the cover information provided by the TM images, with the help of phase information to obtain the accurate land cover/use types firstly, then generated the underlying surfaee type with DEM slope information, finally determined the hydrographic units. Taking Dong-wan Basin in Henan Province as example, the experimen- tal results showed that combination of Bayes classifier and decision tree classifier would help to get higher accuracy of land cover/use classification with their respective superiority, moreover, the phase informa- tion could help to correct the confusion between the individual categories.

关 键 词：下垫面 贝叶斯分类 决策树分类 分类器结合 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]