基于C5.0决策树分类算法的ETM+影像信息提取

利用C5.0决策树算法对ETM 影像进行信息提取,通过与其他分类方法提取结果的对比,得出C5.0决策树分类算法精度较高。大气校正与数据融合可明显提高分类精度,利用经过NDVI、NDBI、缨帽变换处理后的影像组合数据进行信息提取可进一步提高分类精度。研究发现,C5.0决策树算法用未处理的资料生成决策树的效果较差,而经大气校正和数据融合后计算出NDVI、NDBI及缨帽变换的前3个分量的组合数据生成的决策树深度最小,并且分类精度最高。

Extracting Information from ETM+ Image Using C5.0 Decision Tree Algorithm

Decision tree algorithms have significant potential for classification of remotely sensed data.In this paper,C5.0 decision tree algorithm is used to extract information from ETM image.Comparing with the overall accuracy computed by different classification methods using different proceeded remote sensing data,it concluded that data processing is the important step in extracting information from remote sensing data,which can influence the classification overall accuracy greatly.Atmospheric correction and data fusion can improve the classification accuracy if operated correctly.Using the combining data of NDVI,NDBI and the first three factors of Tasseled Cap can further improve the overall accuracy.It also concludes that the classification result is obviously better using C5.0 decision tree algorithm based on the combining data processed by a sequence of atmospheric correction,data fusion and transformed by NDVI,NDBI and Tasseled Cap than the raw data.