基于指数图像的城镇建设用地信息识别与提取——以岳阳市为例  

Identification and extraction of urban construction land information by index image——a case study of Yueyang City

作　　者：杨晏立[1,2] 唐尧[1] 何政伟[1,2,3] 冯淦[2] 王乐[2] 

机构地区：[1]成都理工大学地球科学学院,成都610059 [2]成都理工大学地质灾害防治与地质环境保护国家重点实验室,成都610059 [3]首都师范大学资源环境与地理信息系统北京市重点实验室,北京100037

出　　处：《测绘科学》2011年第4期208-210,共3页Science of Surveying and Mapping

基　　金：国家自然科学基金(40972225);国家科技支撑计划项目(2008BAK49B02)

摘　　要：以岳阳市Landsat ETM+影像为信息源,分析了典型地物的光谱特征及可分性,将地物种类归并为建设用地、植被、水体三大类,分别选用归一化裸露指数(NDBI)、重归一植被指数(RDVI)和改进的归一化水体指数(修改后的NDWI)作为三种地类的指示因子,通过阈值分割、掩膜处理去除了非建设用地区域的噪音信息,得到了比较准确的城镇建设用地专题信息。综合运用三种指数提取建设用地比直接对各波段作复杂的逻辑判断思路更清晰,目的更明确,也弥补了单一指数识别建设用地的不足,精度检验结果表明其有较高的提取精度,是一种兼具自动与半自动、行之有效的城镇建设用地识别与提取方法。The Landsat ETM ＋ image which covers Yueyang was selected as data source in the paper. The spectral features and separability of typical objects were analyzed, and then the typical objects were grouped in three categories which are construction land, vegetation and water body. Normalized difference barren index （NDBI） , re-normalized difference vegetation index （ RDVI ） and im- proved normalized difference water index （revised NDWI） were considered as direct reflection of the three categories. The interference of non-construction land was removed by threshold segmentation and mask processing. The accurate subject information of urban con- struction land was obtained. The comprehensive use of three kinds of index was better than the complex logic judge for each band. It also made up a defect of single index to identify construction land. The result of accuracy test indicated that the method has higher extraction accuracy so that this both automatic and semi-automatic way would be an effective one to identify and extract urban construction land.

关 键 词：建设用地 指数图像 NDBI RDVI NDWI 岳阳市 

分 类 号：TP75[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]