摘要
在GIS辅助下,基于ETM+遥感卫星数据,采用分辨率融合提取地物和基于NDVI计算的分区、分层分类以及结合有关地理数据(如DEM、坡度等)基于知识分类方法相结合,通过多步骤信息提取来完成实验区进行土地利用/土地覆盖分类研究,并对分类结果进行精度检验。研究结果表明:该方法不仅消除了单一步骤分类时多类间的相互影响,而且将遥感影像数据与地利信息系统数据有机的结合起来,具有技术操作简单,分类结果理想的特点;对城镇用地、机场用地、水库、湖泊及河流的分类识别精度可达100%,基于知识规则对农村居民地和耕地的提取精度达70%以上,较单一的信息提取方法精度有很大提高。
This paper presents the LUCC of Shangri-la county research by means of Multi-Stage Classification Approach with GIS, based on Land-sat ETM＋, and the results show that the extraction techniques are also verified. This method not only removed the influence mutually of several classes with Single-Stage Classification Approach, but also combined the information derived from RS with that derived from GIS, and the procedure of implement is effective; And, the airport, the cities, the reservoir, the lake and the rivers classified recognition precision may reach 100%, withdraws the precision based on the knowledge rule to the countryside inhabitant and the farming to reach above 70%, Comparatively uses the sole information extraction method precision to have a bigger enhancement.
出处
《云南地理环境研究》
2007年第2期98-102,共5页
Yunnan Geographic Environment Research
基金
国家自然基金项目(40361007)
973计划项目(2003CB415105)
云南自然科学基金项目(2002D0036M)资助.
