基于QuickBird影像耕地信息提取方法研究——以绵阳市游仙区为例

面对耕地资源日益紧张的形势，快速准确地提取耕地信息已成为耕地动态监测的基础。该文以绵阳市游仙区QuickBird遥感影像为基础图件，借助于RS和GIS技术，分析影像本身特点与耕地光谱特征，通过非监督分类法、监督分类法、基于光谱知识的光谱阈值法分别提取耕地信息。试验结果表明非监督分类提取精度为88．4％，监督分类提取精度为91．53％，光谱阈值法提取精度为94．65％。光谱阈值法的提取效果最好。

A Study of the Extraction of Arable Land Information Based on QuickBird Image-Taking Youxian District of Mianyang City as an Example

With the increasingly tense situation of cultivated land resources,a quick and accurate extraction of arable land information becomes the basis for dynamic monitoring.In this paper,based on the QuiekBird image of the Youxian district of Mianyang,by RS and GIS technology and the analysis of the characteristics of the image and the spectral feature of the arable land,the author has extracted arable land information through unsupervised classification and supervised classification,as well as the spectrum threshold.The results show that the accuracy of the extraction through unsupervised classification is 88.4%,while that through supervised classification is 91. 53%,and that through the spectrum threshold is 94.65%.the last method is proved to be the best.