【摘要】 以东方市为研究区,对OLI影像分别采取基于像元的五种分类方法实施土地利用分类,自评和对比验证分析影像在不同分类方法、不同土地利用类型上的分类精度。研究结果表明:在相同样本量和没有辅助数据的情况下,监督分类精度高于非监督分类,支持向量机的分类方法能够较好地提取土地利用信息,各地类的用户精度、生产者精度较高,总体精度达88.13%,Kappa系数为0.86。不同地类的分类精度差异明显,建设用地和水域的分类精度较高,未利用地和耕地的分类精度较低,林地、草地、耕地等绿色植被之间存在混淆。该研究成果为合理选择分类方法应用于土地利用信息提取提供科学依据。更多还原

【Abstract】 This paper selected Dongfeng City as the study area,classified the land use information by 5 classification methods based on pixel from the OLI images,analyzed the classification precision of different classification methods and land use types by self rating and contrast precision verification.The results show that:The classification precision of supervised classification is higher than that of unsupervised classification in the same sample and without auxiliary data.The Support Vector Machine(SVM)can effectively extract the information of land use.All uses precision and producer precision is high.The overall precision reaches 88.13% and Kappa index is 0.86.The classification precision of different kinds of classification is obvious.The classification precision of waters and construction land is high,but that of unused and cultivated land is low.There is confusion between woodland,grassland and cultivated land.The results provide scientific basis for rational choice classification method using for land use information extraction.更多还原