面向对象的喀斯特地区地表景观遥感信息提取  

Object-oriented Remote Sensing Information Extractionof Surface Landscape in Karst Areas

作　　者：林国敏 刘继庚 邓仕雄 周彬 LIN Guomin;LIU Jigeng;DENG Shixiong;ZHOU Bin(Guizhou Vocational and Technical College of Water Resources and Hydropower,Guiyang 551416,China)

机构地区：[1]贵州水利水电职业技术学院,贵州贵阳551416

出　　处：《科技创新与生产力》2023年第12期125-128,共4页Sci-tech Innovation and Productivity

基　　金：贵州水利水电职业技术学院校级课题(GZSK2022015)。

摘　　要：针对传统面向像素的地表景观提取方法对喀斯特地区提取精度低的问题,为提高喀斯特地区地表景观信息提取的精度,本文以覆盖典型喀斯特流域赤水河中上游地区的Landsat8 OLI影像为数据源,采用面向对象的分类方法,将研究区的空间特征、光谱特征、形状特征等参与多尺度分割,形成“同质影像”对象,再结合坡度、亮度、归一化建筑物指数(NDBI)及归一化植被指数(NDVI)和多波段等参数,建立了适应喀斯特地区的地表景观分类体系,从而实现了研究区地表景观信息的自动提取。结果表明,该方法对喀斯特地区地表景观信息提取精度高于85%,分类方法和参数可以为喀斯特地区大面积遥感影像地表信息提取提供参考。Aiming at the problem that traditional pixel oriented surface landscape extraction methods have low accuracy in extracting karst areas,in order to improve the accuracy of surface landscape information extraction in karst areas,this paper uses Landsat 8 OLI images covering the middle and upper reaches of the Chishui River in a typical karst basin as the data source,and uses object-oriented classification methods to participate in multi-scale segmentation of spatial features,spectral features,shape features,and other features of the study area,forming a"homogeneous image"object.A surface landscape classification system suitable for karst areas is established by combining parameters such as slope,brightness,normalized building index(NDBI),normalized vegetation index(NDVI),and multi band,in order to achieve automatic extraction of surface landscape information in the research area.The results show that the accuracy of this method in extracting surface landscape information in karst areas is higher than 85%,and the classification method and parameters can provide reference for extracting surface information from large-scale remote sensing images in karst areas.

关 键 词：遥感 地表景观 地表景观提取 面向对象 喀斯特地区 赤水河 

分 类 号：TP7[自动化与计算机技术—检测技术与自动化装置] P237[自动化与计算机技术—控制科学与工程] P642.25[天文地球—摄影测量与遥感]