双频InSAR高程重建噪斑去除方法  

The Noise Removal of Dual-Frequency InSAR DEM Rebuild Result

作　　者：蔺祥楠 禹卫东[1,2] IN Xiangnan;YU Weidong(Institute of Electronics, Chinese Academy of Sciences, 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,北京100190 [2]中国科学院大学,北京100049

出　　处：《雷达科学与技术》2019年第2期173-183,共11页Radar Science and Technology

基　　金：国家重点研发计划资助项目(No.2017YFB0502700);航天"十三五"技术预研项目

摘　　要：双频InSAR不依赖于相邻相位差小于π的限制而进行解缠,可以实现陡变地形的正确解缠绕,从而可以实现复杂地形的高程反演。基于最大似然估计(Maximum-Likelihood Estimation,MLE)的双频干涉相位解缠算法耗时少,但其噪声鲁棒性差,解缠结果中含有大量噪声斑块。针对此弊端,提出了一种对ML解缠结果进行噪声斑去除的方法。实际操作中,首先利用聚类分析(Cluster Analysis,CA)的思想,去掉解缠结果中较大量出错区域对应的模糊类。再利用可变长窗口判定错点,并利用非均匀加权窗去掉单噪点和噪声斑,得到正确的解缠相位。实验章节首先列出了仿真模型双频干涉处理结果,将改进结果与最大后验概率估计法(MAP)得到的结果作了比较。最后列出了对中国科学院电子学研究所航天微波遥感系统部的双频干涉数据处理结果。实验结果表明,解缠结果存在的问题得到解决,在保证较好解缠效率的同时,去除了噪声斑块,提高了相位解缠的精度。Dual-frequency InSAR phase unwrapping breaks the limit of the assumption that the adjacent phase difference is less than π. It can realize the correct unwrapping of the steep terrain and achieve the height inversion of the complex terrain. Maximum likelihood estimation (MLE) based dual-frequency phase unwrapping algorithm takes less time, but its noise robustness is poor and there are a lot of noise patches in the unwrapping results. In order to solve this problem, a method for removing noise spots from ML unwrapping results is proposed. In practice, first of all, cluster analysis (CA) is used to remove the fuzzy classes corresponding to the error areas in unwrapping results. Then the variable length window is applied to determine the error points, and the non-uniform weighted window is used to remove the noise spots and patches, so that the correct unwrapping phase is obtained. The results of dual-frequency interference processing of the simulation model are listed, and the improved results are compared with the maximum a posteriori (MAP) based unwrapping results. Finally, the results of dual-frequency interference data processing are given, the data was provided by the Department of Aerospace Microwave and Remote Sensing System, Institute of Electronics, Chinese Academy of Sciences. The results show that the problem existing in the unwrapping results is solved, the noise patches are removed, and the precision of phase unwrapping is improved while better unwrapping efficiency is ensured.

关 键 词：干涉合成孔径雷达 双频InSAR 最大似然估计 噪声斑块去除 

分 类 号：TN957.52[电子电信—信号与信息处理]