摘要
阐述了决策树分类CART算法原理,将纹理信息、NDVI指数引入决策树方法对影像进行分类,并将分类结果与最大似然分类结果进行比较,研究表明决策树分类方法相对传统分类方法总体精度提高了8.9148%,Kappa系数提高了0.1074。
The paper discussed decision-tree classification principle and CART algorithm, introduced the texture information and NDVI index into decision-tree method to complete classification, and compared with the maximum likelihood classification results. Resuhs show that the Decision-tree classification method improves the overall ac- curacy of 8.9048%, Kappa coefficient of 0.1074 inceased by comparision with traditional classification methods relatively.
出处
《地理空间信息》
2009年第6期15-17,共3页
Geospatial Information
基金
国家863计划资助项目(2007AA12Z154)
