中文摘要：

以南京市1988年TM影像为信息源，组合影像4、5波段得到建筑用地指数（NDBI），采用决策树分类方法对NDBI进行修正，分类提取建筑用地；对照TM影像修改误分的地块后转化为栅格数据，生成50×50个像元大小的格网，统计每个格网单元内的建筑用地率，将该值赋给各个格网单元的几何中心点；采用距离倒数方法由点数据插值得出整个影像范围内的建筑用地区域分异情况，进而反映出不同建筑用地率的辐射影响范围。该方法可定量提取不同建筑用地率的辐射边界，有助于根据单张影像数据分析城市的扩张趋势，可在一定程度上预测城市的发展；对于需要跨江发展的城市（如南京），该方法可以辅助桥梁等过江通道的选址。

英文摘要：

Take Nanjing for an example,this research is carried out with TM images covering Nanjing City in 1988. Using the decision tree classification algorithm to modify the NDBI, to extract the structure land use area, then to convert them into vector polygons, comparing with the TM image and to delete the vector polygons, which are wrong classified. Subsequently, the 50 ;〈 50 cells grid to cover the image was generated,and the structure land use area ratio of each unit was gotten. These data were given to geometric center points of all grid units. Finally, the point data were used for interpolation to get regional differentiation data for the structure land use area in the whole images, so as to reflect various radiation range of different structure land use ratio. By using this method the researchers can extract the radiation range of the different structure land use ratio quantitatively in effect and use only one image to analyze the urban modality and forecast the urbanization development.