基于LandsatTM数据的潍坊市土地利用分类方法研究.doc

山东科技大学本科毕业设计（论文） 山东科技大学本科毕业设计（论文） PAGE 2 - PAGE 1 - 摘 要 利用遥感技术对区域性土地利用分类，定量提取土地利用信息，是全球土地研究的重要领域之一。而遥感影像的自动信息提取和分类识别是实现这一目标的关键。本研究以潍坊市作为研究区，选取Landsat TM影像作为基础数据源，进行数据预处理，获得研究区影像，建立训练样本，将潍坊市土地利用类型分为林地、草地、耕地、水域、住宅工矿用地和其他土地6类。在遥感图像处理系统(ENVI)支持下，采用最大似然法、最小距离法、支持向量机法和IsoData法四种分类方法进行土地利用分类对比研究，运用不同的精度指数进行讨论，其中包括总体精度、分类精度以及决定最优算法的Kappa一致性系数。结果表明：最大似然分类方法具有较高的分类精度，以此获得了研究区的土地利用信息，建立土地利用柱状图，为潍坊市建设规划和发展提供参考。 关键词：土地利用变化；最大似然法；最小距离法；支持向量机法；IsoData法；Kappa系数 Abstract The use of remote sensing technology for dynamic monitoring of regional land use / cover, quantitative extraction of land use / cover change information is one of the important areas of global change research. And the automatic information extraction and classification recognition of remote sensing image is the key to achieve this goal. This study selects Weifang city as the research area, selects the TM images of December 6, 2009 and January 14, 2010 as the basic data source, establishes the training samples, divides the land use types in Weifang city into forest land, grassland, farmland, water area, residential area and other land. Under the support of remote sensing image processing system ( envi ), the comparison study of land use classification using the maximum likelihood method, minimum distance method, support vector machine method and IsoData method, using different precision index, including the overall accuracy, classification accuracy and the kappa consistency coefficient of the optimal algorithm. The results show that the maximum likelihood classification method has high classification accuracy, and obtains the land use information of the research area to provide reference for the construction planning and development of Weifang city. Key words: Land use changes; Maximum likelihood method; Minimum distance method ;Support?Vector machine method; IsoData method; Kappa coefficient PAGE 2 - PAGE 1 - 目 录 TOC \o 1-3 \h \u HYPERLINK \l _Toc16713 目 录 PAGEREF _Toc16713 I HYPERLINK \