基于ETM＋的地物信息分层分类方法研究

该文基于青岛经济技术开发区及周边地区ETM＋数据,结合光谱特征阈值算法、遥感图像计算机分类法,人工目视解译对实验区土地利用信息进行分层提取。研究表明：正确地使用基于光谱特征阈值法和非监督分类法的分层分类法,能在一定程度上提高研究区的分类精度,分类总精度达80%,基本满足遥感分类与制图的要求。其中水体、滩涂、城镇居住区的分类精度较为理想,达83%以上;林地、裸地、农田以及农村居住区分类精度相对较低,还有待进一步研究。

A Study on Stratified Classification of Extracting Land Use Information Based on Landsat Etm+by Remote Sensing

Based on the ETM ＋ data of the Qingdao economic and technological development zone and its surrounding areas, combined with the threshold algorithm of spectral characteristics, remote sensing image classification methods with computer, we stratified extracting the land - use information of the study area with artificial visual interpretation. Research shows that the proper use of hierarchical classification based on the threshold algorithm of spectral characteristics and the unsupervised classification methods can improve the accuracy of classification of the study area, and the overall accuracy can reach to 80% which can meet the basic requirements with the classification of and mapping remote sensing. Among the results the accuracy of the body of water, tidal fiats, urban residential are better which are higher than 83 %. the classification accuracy of forest land, bare land, farmland and rural residential are relatively low, has yet to be studied further.