唐山市区土地利用的Landsat 8影像分层分类  

Land use hierarchical classification based on Landsat 8 image in Tangshan city

作　　者：郭力娜 张梦华 张永彬 牛振国 李孟倩 

机构地区：[1]华北理工大学矿业工程学院,河北唐山063009 [2]中国科学院遥感与数字地球研究所遥感科学国家重点实验室,北京100101

出　　处：《测绘科学》2017年第10期88-94,125,共8页Science of Surveying and Mapping

基　　金：河北省教育厅优秀青年基金项目(YQ2014016)

摘　　要：针对土地利用遥感分类方法多样、分类精度高低不一等问题,该文以土地利用变化明显的唐山市路南区、路北区为研究区域,并以中分遥感影像Landsat 8OLI为信息源,在对地类样本进行可分离性分析的基础上,建立研究区土地利用分层分类体系。通过监督分类实验,选择分类效果最好、分类精度最高的最大似然分类器进行地类初分;通过绘制归一化植被指数(NDVI)、归一化建筑指数(NDBI)、两指数差值(NDVI-NDBI)的曲线及地类光谱特征曲线,建立决策树分类规则,进行地类再分。该方法可以较好地完成多种土地利用二级地类的划分,有助于提高中分影像土地利用分类效率。According to the problems that the methods in land use remote sensing classification are various and the classification results are different, this paper took the Lunan and Lubei districts of Tangshan city as an example and set up the land use hierarchical classification system using Landsat 8 remote sensing image based on the analysis of the separability of land type sample. Through the supervised classification, the maximum likelihood method was chosen for initial classification; then, the curve of normalized differ ence vegetation index （NDVI）, normalized difference built-up index （NDB1）, NDVI-NDBI were drew and the classification rules of decision tree was built for secondary classification. Finally, the land use hierar- chical classification was completed. The proposed method could well classify the second land use type, which improved the efficiency of land use classification in mid and sub regional images.

关 键 词：LANDSAT 8 土地利用 分层分类 监督分类 决策树分类 

分 类 号：P237[天文地球—摄影测量与遥感]