中文摘要：

针对目前遥感影像分类中面向对象和决策树相结合的研究较少的情况,该文提出基于C5.0决策树的面向对象分类方法,并以广州市从化区进行实证研究。基于Landsat-8OLI影像数据,采用面向对象分类对影像进行多尺度分割,提取出影像对象的光谱、纹理特征以及影像对象相对应的DEM信息;然后利用C5.0决策树根据特征信息来挖掘分类规则;最后根据规则对分割后影像进行分类。结果表明,基于C5.0决策树的面向对象影像分类精度高、效果好,总体精度和Kappa系数分别为89.75%和87.5%。该方法可准确、快速地提取土地利用/覆被信息。

英文摘要：

According to the present situation that the study combining decision tree with object-oriented to classify remote sensing image is less,a remote sensing image classification based on C5.0algorithm and object-oriented technique was proposed in this paper.Based on Landsat-8OLI data,the image was segmented with multi-scale segmentation method by eCognition software,and the features of objects like spectral,texture and DEM were extracted;classification rules were excavated using C5.0module of IBM SPSS Modeler software integration based on feature information;finally,the image was classified according to the rules.Results showed that the classification based on C5.0algorithm and object-oriented image achieved high classification accuracy with overall accuracy and Kappa coefficient of 89.75%and 87.5%respectively.This method could quickly and accurately extract the land use/land cover information.