城区地表形变差分TomoSAR监测方法

发布时间：2019-06-28 15:39

【摘要】：压缩感知技术(CS)的差分TomoSAR技术解决了中高分辨率SAR数据在城区出现的叠掩问题,实现了城区地表形变信息的重构,但是该方法仅利用了目标的稀疏特性并没有考虑目标的结构特性,对具有这两种特性的目标进行重构时其性能较差。针对这一问题,本文采用联合Khatri-Rao子空间和块压缩感知(KRS-BCS),提出了一种差分SAR层析成像方法。该方法依据目标的结构特性和重构观测矩阵具有的Khatri-Rao积性质,将稀疏结构目标的差分TomoSAR问题转化为Khatri-Rao子空间下的BCS问题,然后对目标进行块稀疏的l1/l2范数最优化求解,最后通过理论分析和仿真试验对分辨能力和重构估计性能进行了定性和定量评价,仿真结果表明本文所采用的KRS-BCS方法不仅保持了高分辨率的优点,而且有效地降低了虚假目标出现的概率,大幅度提高了散射点准确重构概率,切实可行地解决了CS方法的不足。应用实例研究中,利用34景Envisat卫星ASAR时间序列影像对日本千叶县茂原市城区进行地表形变监测,并以一等水准点和实时测量的GPS站点观测数据作为参考形变结果进行验证,试验结果表明采用KRS-BCS方法反演的结果与参考形变结果保持了良好的一致性且形变速率整体偏差也较小,实现了较高精度的城区地表形变估计。
[Abstract]:The differential TomoSAR technology of compression sensing technology (CS) solves the overlap problem of medium and high resolution SAR data in urban area, and realizes the reconstruction of urban surface deformation information. However, this method only makes use of the sparse characteristics of the target and does not consider the structural characteristics of the target, and its performance is poor when reconstructing the target with these two characteristics. In order to solve this problem, a differential SAR tomography method is proposed by combining Khatri- Rao subspace and block compression sensing (KRS-BCS). According to the structural characteristics of the target and the Khatri-Rao product property of the reconstructed observation matrix, the differential TomoSAR problem of sparse structural target is transformed into the BCS problem in Khatri- Rao subspace, and then the block sparse l1/l2 norm is optimized. finally, the resolution and reconstruction estimation performance are evaluated qualitatively and quantitatively through theoretical analysis and simulation experiments. The simulation results show that the KRS-BCS method used in this paper not only maintains the advantages of high resolution, but also effectively reduces the probability of false targets, greatly improves the accurate reconstruction probability of scattering points, and solves the shortcomings of CS method. In the application case study, 34 Envisat satellite ASAR time series images are used to monitor the surface deformation in Maoyuan City, Chiba Prefecture, Japan, and the first level point and real-time measured GPS station observation data are used as reference deformation results to verify the experimental results. The experimental results show that the inversion results by KRS-BCS method are in good agreement with the reference deformation results, and the overall deviation of deformation rate is also small. The high precision surface deformation estimation in urban area is realized.
【作者单位】：中国科学院电子学研究所微波成像技术国家级重点实验室;中国科学院大学;中国资源卫星应用中心;
【基金】：国家发改委卫星及应用产业发展专项([2012]2083)~~
【分类号】：P227

【相似文献】