【摘要】 "土地利用/土地覆被"(简称LUCC)是全球环境变化研究的热点问题之一,遥感技术是LUCC研究的重要手段,遥感影像LUCC信息的高精度自动提取成为众多专家学者关注的重要研究领域.Landsat系列卫星影像是最早且最为广泛使用的中分辨率陆地遥感影像.文章以2013年2月发射的Landsat8卫星的OLI影像为研究对象,利用ENVI软件,对太原市区的土地利用/土地覆被信息分别用最大似然法和决策树分类法实现自动提取,并进行精度比较.结果表明:最大似然法提取的总体精度为69.33%,Kappa系数为0.605 5,主要是耕地与林地、草地、建设用地以及建设用地与裸地的混分、错分现象较为严重,造成分类精度较低;而决策树分类法的自动提取通过确定适当的判别规则,使地类间的混分、错分现象明显改善,分类总体精度提高到91.33%,Kappa系数达到0.892 3,各土地利用类型的分类精度均有一定提高.更多还原

【Abstract】 " The land use and land cover(LUCC)" is one of the hotspot in global environmental change research,and RS technology is the important research means to LUCC study,so,the automatic extraction with high accuracy of LUCC information on RS image has become an important research field for many experts and scholars.Landsat Series Satellites image with medium resolution is the earliest RS data and has been used most widely.OLI image of Landsat8 satellite launched in February 2013 as the research object,based on ENVI,the paper studied the automatic extraction of land use/cover information using methods of decision tree and maximum likelihood in Taiyuan city center.The results showed that the accuracy of final classification was low,overall accuracy is 69.33%,Kappa coefficient is 0.605 5,because there were some error classification between cultivated land and forest land,grassland and construction land,so was it between construction land and bare land.However,the automatic extraction method of decision can obviously improve classification accuracy by appropriate discrimination rules,the overall accuracy could be 91.33%,Kappa coefficient reached 0.892 3,and higher precision can be seen in each land use type.更多还原