最大似然估计的改进多基线InSAR解缠算法  

An improved multi-baseline InSAR unwrapping algorithm based on maximum likelihood estimation

作　　者：马靓婷 卢小平[1] 周雨石 MA Liangting;LU Xiaoping;ZHOU Yushi(Key Laboratory of Spatio-temporal Information and Ecological Restoration of Mines,MNR,Henan Polytechnic University,Jiaozuo,Henan 454000,China)

机构地区：[1]河南理工大学自然资源部矿山时空信息与生态修复重点实验室,河南焦作454000

出　　处：《测绘科学》2020年第8期123-129,共7页Science of Surveying and Mapping

基　　金：2016年国家重点研发计划项目(2016YFC0803103);河南省高校创新团队支持计划项目(14IRTSTHN026);河南省创新型科技创新团队支持计划项目。

摘　　要：多基线InSAR相位解缠算法能够突破相位欠采样及相位连续性假设的问题,可获得比单基线更为精确的解缠结果,但现有的多基线相位解缠算法存在噪声鲁棒性差或运行时间长的缺点。为提升精度减少运行时间,该文提出了结合最大似然估计算法与扩展卡尔曼滤波算法的多基线相位解缠算法。该算法首先对基于最大似然估计算法重建的预估地形高程进行误差点判断,之后利用扩展卡尔曼滤波的方法对误差点高程进行重建,获得最终估计的地形高程。为验证算法的适用性,采用模拟数据和实测数据进行实验处理,以归一化均方根误差和算法运行时间作为评价指标,将此算法与最长单基线最小费用流解缠算法、最大似然估计多基线解缠算法和最大后验估计多基线解缠算法进行比较,实验结果表明该方法精度较高、运行时间较短。The multi-baseline phase unwrapping algorithm can overcome the problem of phase undersampling and phase continuity assumptions,and obtain more accurate unwrapping results than single baseline.However,the traditional multi-baseline phase unwrapping algorithm has poor noise robustness and long running time.In order to improve the accuracy and reduce the running time,this paper proposed a multi-baseline phase unwrapping algorithm combining maximum likelihood estimation(ML)algorithm and extended Kalman filter(EKF)algorithm.The algorithm firstly judged the error point of the estimated terrain elevation map reconstructed by ML algorithm,and then used the extended Kalman filter phase unwrapping(EKFPU)method to reconstruct the error point elevation to obtain the final estimated terrain elevation map.In order to verify the applicability of the algorithm,the experimental data was processed by using simulated data and measured data.The root mean square error and algorithm running time were used as evaluation indexes.The algorithm was compared with the longest baseline unwrapping result,ML algorithm and MAP algorithm.The results showed that the method was simple and effective.

关 键 词：相位解缠 最大似然估计 最大后验估计 误差点 扩展卡尔曼滤波 

分 类 号：P237[天文地球—摄影测量与遥感]