基于组稀疏的极化差分SAR层析成像方法  

Polarization differential SAR tomography imaging method based on group sparse

作　　者：杨牡丹 张冰尘[1,2] 魏中浩[1,2,3] 徐志林 洪文 YANG Mudan;ZHANG Bingchen;WEI Zhonghao;XU Zhilin;HONG Wen(Key Laboratory of Technology in Geospatial Information Processing and Application System, Institute of Electronics, Chinese Academy of Sciences, Beijing 100190, China;Institute of Electronics, Chinese Academy ofSciences, Beijing 100190, China;University of Chinese Academy of Sciences, Beijing 100049, China)

机构地区：[1]中国科学院电子学研究所空间信息处理与应用系统技术重点实验室,北京100190 [2]中国科学院电子学研究所,北京100190 [3]中国科学院大学,北京100049

出　　处：《系统工程与电子技术》2019年第5期1007-1012,共6页Systems Engineering and Electronics

基　　金：国家自然科学基金(61331017)资助课题

摘　　要：差分合成孔径雷达(synthetic aperture radar,SAR)层析成像通过多航过数据重建观测目标的后向散射系数和视线方向形变速率。将全极化与差分SAR层析相结合,针对城市建筑成像高程向稀疏的特点以及全极化数据反演中,信号稀疏支撑集相同的特点,提出将稀疏约束与组稀疏约束相结合的求解模型,用基于层次稀疏的阈值迭代方法进行求解。通过仿真实验和基于BioSAR 2007实测数据的半仿真实验验证,实验表明全极化差分SAR层析成像方法,重构结果相较于单极化,提高了高程向和形变速率精度,且有更好的鲁棒性,在信噪比为10dB时,相较于单极化差分SAR层析成像方法,能更好恢复高程向信息和形变速率。Differential synthetic aperture radar (SAR) tomography reconstructs the backscattering coefficient and line-of-sight deformation rate of the target by multi-pass data. We combine the full-polarization and differential SAR tomography. Aiming at urban building imaging elevation sparseness and the same signal sparse support set in full-polarization data inversion, a solution model combined with sparse constraints and group sparse constraints is solved by threshold iterative method based on hierarchical sparseness. The simulation and semi-simulation experiment based on BioSAR data are done to verify the results. The experiment shows compared with the monopole the reconstruction results the full-polarization differential SAR tomography method improve the accuracy of elevation and has better robustness. When the signal-to-noise ratio is 10 dB, it can recover the elevation information and deformation rate better than the single-polarization differential SAR tomography method.

关 键 词：全极化 差分合成孔径雷达层析成像 组稀疏 阈值迭代 

分 类 号：TN958[电子电信—信号与信息处理]