基于ENVI ZOOM面向对象的高海拔灌丛植被提取——以定日县为例  

Extraction of Shrub Vegetation by Object-Oriented Classification Method Based on ENVI ZOOM in High-Altitude Area:A Case of Dingri County

作　　者：张学儒[1,2] 刘林山[1] 张镱锂[1] 张继平[1,2] 聂勇[1,2] 张华[1,2] 

机构地区：[1]中国科学院地理科学与资源研究所,北京100101 [2]中国科学院研究生院,北京100049

出　　处：《地理与地理信息科学》2010年第4期104-108,共5页Geography and Geo-Information Science

基　　金：国家科技支撑计划项目(2007BAC06B08);中国科学院对外合作重点项目(GJHZ0954);国家重点基础研究发展计划项目(2005CB422006)

摘　　要：利用ASTER数据,基于The Environment for Visualizing Images(ENVI)ZOOM软件平台,采用面向对象的多尺度影像分割技术与规则创建相结合方法,提取了珠峰保护区核心区的灌丛植被类型。在对影像进行去阴影处理和各种辅助信息融合的基础上,使用Feature Extraction模块对影像进行分割,基于分割对象的高程、NDVI、纹理和光谱信息创建了适合研究区的灌丛提取规则。研究表明:该分类方法不仅能够克服传统基于像元分类方法中的"椒盐效应"问题,而且能够综合利用辅助信息(DEM、NDVI等)和地物本身的信息(光谱特征、纹理特征等),有效提高解译精度。以已有数据对分类结果进行了检验,分类精度达84.7%,分类结果较理想。the alpine shrub is one of the major vegetation types in Tibetan Plateau,and extracting vegetation types by remote sensing has been a hot area of research.In this paper,based on the Environment for Visualizing Images（ENVI） ZOOM software platform and ASTER data,shrub vegetation types in the core area of Qomolangma Nature Reserve was extracted by using object-oriented multi-scale image segmentation techniques and creating rules.The shadow of the image was removed and a variety of auxiliary information to image was fused.Then the image was segmented by the Feature Extraction module and rules which can extract the alpine shrub vegetation types in the study area was created based on elevation,NDVI,texture and spectral information of segmented image.The research showed that：this classification method is not only able to overcome the problem of the ＂salt and pepper effect＂ by the traditional pixel-based classification,but also to use the auxiliary information（DEM,NDVI,etc.） and images information（spectral characteristics,texture features,etc.） comprehensively.The results of the classification were tested by existing data.The accuracy of classification is 84.7%,the results are satisfactory.It is found that it is very important to select a reasonable parameter of division scale for the object-oriented image classification,and Feature Extraction module of ENVI ZOOM can inquiry the accuracy of segmentation scale which is a good method of image scale segmentation.Therefore,the result shows this vegetation extraction method can improve the interpretation accuracy effectively,and the precision is high.And this vegetation extraction method provides effective method and theoretical foundation for extracting vegetation automatically.

关 键 词：面向对象 高寒灌丛 ENVIZOOM 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]