基于模型的单幅高分辨SAR图像建筑物高度反演方法

Model-Based Building Height Retrieval from Single HR SAR Image

高分辨合成孔径雷达(SAR)图像的丰富使得建筑物三维信息获取成为城区遥感的一项重要研究课题.针对单幅SAR图像建筑物的高度估计问题,提出了一种基于模型的高度反演方法.该方法遵循假设检验的思路,通过正交投影模型实现模型假设的快速生成,并设计了基于分割的似然函数,利用模拟退火算法在高维模型假设空间中搜索最优解.基于模拟和实测机载SAR图像的实验验证了算法的有效性.

With the development of high resolution （HR） synthetic aperture radar （SAR） imaging, the technology of building 3D information extraction from SAR images is becoming an important issue in radar remote sensing. In order to retrieve building height information from single FIR SAR image, a model-based height estimation approach is proposed. An orthogralahic projection model is established to achieve the quick generation of the model hypothesis. A segmentation-based likelihood function is derived to examine the consistency between the model hypothesis and the actual SAR image. The hypothesis test procedure, which is iterated with the simulated annealing algorithm,is designed to optimize the hypothesis from the high dimensional model space. The experimental results using both of the simulated and ac0aal airborne SAR images show the effectiveness of the proposed method.