基于TM影像的城市绿地信息提取方法研究  

Extraction method of urban vegetation information based on TM image

作　　者：姚静[1,2] 武文波[1] 康停军[3] 

机构地区：[1]辽宁工程技术大学测绘与地理科学学院,辽宁阜新123000 [2]广东省国土资源信息中心,广州510075 [3]中山大学遥感与地理信息工程系,广州510275

出　　处：《测绘科学》2010年第1期113-115,共3页Science of Surveying and Mapping

基　　金：教育部博士点基金(20050147002);辽宁省教育厅重点实验室项目(20060370);辽宁省教育厅创新团队项目(2007T073)

摘　　要：基于TM遥感影像,运用ERDAS,对某地城市绿地专题信息进行了提取。实验过程中首先对图像进行预处理,然后通过四种绿地信息提取方案进行比较分析,这四种方案分别为:原始波段合成法、主成份分析法、归一化植被指数(NDVI)法和实验波段组合法。将以上几种方案的图像进行反复比较,根据研究对象的实际情况,植被景观目视效果最好的是NDVI植被指数法。对以上四种方法的彩色合成图像进行监督分类,利用目视判读的方法对TM影像的分类结果进行精度检验,由此可以看出实验波段组合法的精度最高,该方法是一种有效的绿地提取的方法。Based on remote sensing image, the special information of urban vegetation in one area was extracted using ERDAS. In the process of the experiment, image pretreatment was finished as the first step. Four methods for urban vegetation extraction were compared. They are original bands synthesis method, principal component analysis method, the normalized difference vegetation index （NDVI） method, experimental bands combination method. Comparing the images produced by the four projects, according to the actual situation of the study object, we find that NDVI vegetation index method has the best visual effect. Supervised classification was done with the four images synthesized in pseudo-color, and accurate assessment was finished with ERDAS. Test results show that experimental bands combination method has the highest accuracy in the four methods, and it is an effective method for the extraction of urban vegetation.

关 键 词：绿地提取 主成分分析 NDVI 波段组合 监督分类 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]