Taking Tibet grassland as the survey region, the grassland types was classified by remote sensing image of Spot and TM using eCognition software platform and applying the object oriented multiscale segmentation method.This method synthetically utilized the characteristic of spectrum, shape, texture, etc.and extracted rapidly and effectively the information classification of grassland types.The overall accuracy reached 93.69% and the accuracy of grassland type classification reached 90.71%, which obtained good effect.%以西藏全区草原作为研究区域,利用eCognition软件平台,运用面向对象的多尺度分割方法,对研究区Spot、TM遥感影像进行草原类型分类.该方法综合利用了光谱、形状与纹理等特征,快速、有效地进行草原类型的信息分类提取,取得了良好的效果,总体精度达到93.69%,草原类型分类精度达到90.71%.
展开▼