Reconstructions of the 3-D model of buildings from single high-resolution SAR (Synthetic Aperture Radar) image were studied. After a detailed analysis on the electromagnetic backscattering model of buildings, the methods on how to calculate the backscattering coefficient of different reflectivity areas and on how to create a simulated SAR image from the CAD model of a building was proposed. The outlines and orientations of buildings were extracted using the double-bounce structures. The building height was estimated based on the iterative simulation and matching. The simulated SAR images are divided into different kinds of scattering areas. The Kullback-Leibler divergence between these areas was treated as a measurement of the matching degree. The tested height, at which the simulation SAR image obtains the best matching degree, is taken as the final estimated height. The above methods were evaluated on two real high-resolution SAR images. The reconstruction results are satisfying and prove the feasibility of 3-D reconstruction of building from single high resolution SAR image and the validity of the proposed approach.%提出了一种利用高分辨率SAR图像进行建筑物提取和三维重构的方法.首先,分析了高分辨率SAR图像建筑物产生的电磁散射的类型,给出了不同类型散射区域的后向散射计算方法,并在此基础上给出了一种利用建筑物三维CAD模型进行SAR建筑物特征区域图像仿真的方法；其次,给出了利用建筑物的二次散射结构确定建筑物底部轮廓位置和方向的方法,并提出了一种基于分布密度函数差异的仿真图像迭代匹配方法,进行建筑物高度的反演.仿真SAR图像后向散射系数用来划分建筑物不同的散射区域,通过计算特征区域之间的分布密度函数差异,以取得最大匹配度值的仿真图像对应的检验高度作为建筑物的反演高度；最后,选用了两幅不同屋顶类型的实际机载高分辨率SAR图像进行建筑物提取和三维重构实验,试验结果较为理想,验证了所提方法的可行性和有效性.
展开▼
