多方位高分辨率SAR的三维目标自动重建(二)多方位重建  

Automatic reconstruction of 3D objects from multi-aspect part II: Multi-aspect reconstruction

作　　者：徐丰[1] 金亚秋[1] 

机构地区：[1]复旦大学波散射与遥感信息国家教育部重点实验室,上海200433

出　　处：《电波科学学报》2008年第1期23-33,共11页Chinese Journal of Radio Science

基　　金：国家重点基础研究项目(2001CB309401);国家自然科学基金(605710520;40637033)

摘　　要：提出从多方位米级分辨率SAR图像自动重建三维立体建筑物目标的方法。第一部分给出了建筑物目标像的自动检测与提取。第二部分定义一组几何参数描述建筑物目标像,给出参数分布的概率描述。由此建立多方位目标像的参数相关模型,给出其极大似然估计。设计了一种有效的多方位自动重建算法,从多方位检测结果中以较高的可靠性重建建筑物目标。分别用4方位虚拟场景的模拟SAR图像和机载全极化X波段Pi-SAR图像做试验。比较实际场景,重建结果证明了本方法的有效性。也给出了发展多方位星载SAR应用的初步建议。Automatic reconstruction of three-dimensional （3D） building objects from multi-aspect metric-resolution SAR images is developed. In previous Part Ⅰ, the building object-images are extracted from SAR images. In part Ⅱ, the object- images are parameterized, and the probability density functions of the orientation, position and scale of the objects in multi-aspect SAR images are presented. Using the maximum-likelihood estimation and automatic multi-aspect reconstruction algorithm, the building objects are reconstructed using full information from multi-aspect SAR data. Both simulated image and 4-aspect Pi-SAR images are investigated for the stereo objects reconstruction. The reconstruction results well demonstrate the feasibility of our novel inversion approach. Some proposals for further develop- ment of multi-aspect SAR technology are also discussed.

关 键 词：高分辨率SAR 建筑目标重建 极大似然估计 多方位相关 多方位配准 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]