一种自适应迭代的非局部干涉相位滤波方法  

An Adaptive Iterated Nonlocal Interferometry Filtering Method

作　　者：林雪[1,2,3] 李曾玺 李芳芳[1,2,3] 胡东辉[1,2] 丁赤飚[1,2] 

机构地区：[1]中国科学院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,北京100190 [4]中国科学技术大学,合肥230027

出　　处：《雷达学报（中英文）》2014年第2期166-175,共10页Journal of Radars

基　　金：国家部委基金资助课题

摘　　要：干涉相位滤波是获取高精度DEM及正射图像过程中必不可少的关键步骤。传统相位滤波方法在滤除噪声和保持相位细节方面不能很好地兼顾,尤其在相干性较差或地形变化剧烈的情况下,难以得到高质量的干涉相位。该文针对上述问题提出了一种自适应迭代的非局部干涉相位滤波方法,其基于非局部的滤波思想,利用图像的冗余信息进行滤波。该方法根据干涉相位的自身特性,实现了衰减参数的自适应,之后通过自动迭代并在迭代过程中调节窗口大小使得滤波精度进一步提高。该文通过仿真和实测数据对现存算法与该文算法进行了对比分析,验证了所提算法的有效性。Interferometry filtering is one of the key steps in obtain high-precision Digital Elevation Model （DEM） and Digital Orthophoto Map （DOM）. In the case of low-correlation or complicated topography, traditional phase filtering methods fail in balancing noise elimination and phase preservation, which leads to inaccurate interferometric phase. This paper proposed an adaptive iterated nonlocal interferometry filtering method to deal with the problem. Based on the thought of nonlocal filtering, the proposed method filters the image with utilization of the image redundancy information. The smoothing parameter of the method is adaptive to the interferometry, and automatic iteration, in which the window size is adjusted, is applied to improve the filtering precision. Validity of the proposed method is verified by simulated and real data. Comparison with existed methods is given at the same time.

关 键 词：干涉相位滤波 非局部 自适应 迭代 低相干 

分 类 号：TN957.52[电子电信—信号与信息处理]