一种新的INSAR干涉条纹图滤波方法

A New Filter on INSAR Interferogram Fringe

干涉条纹图滤波是合成孔径雷达干涉测量处理中不可缺少的重要环节。对干涉条纹图滤波的一个重要要求是在有效抑制噪声的同时尽可能地保持条纹的跳变纹理信息。在分析干涉条纹图一阶差分的分布性质和三种当前最常用干涉条纹图滤波的算法的优缺点的基础上,提出基于椭圆方程的干涉条纹方向检测方法,并基于此推导了基于误差方程的用于表述不同方向的像素与滤波窗口中心像素的相关关系的加权模型,以此加权模型修正窗口均值滤波和圆周期均值滤波,得到新的保边缘的滤波器。接着分析了新滤波器的计算复杂度和不同权函数对滤波结果的影响。以此新的椭圆加权的保边缘滤波器处理了河北尚义地区的ERS 1/2合成孔径雷达数据,证明了本文提出的新滤波算法具有噪声抑制效果好,边缘信息保持能力强,计算速度快和可灵活配置等优点。

Filtering on interferometry fringe is one of the key processing procedures involved in Synthetic aperture radar interferometry. The principal requirement to the filtering is to make the best maintenance to the fringe's texture as filtering out noises. Based on the discussion on the distribution of interferometry fringe's first order difference and three most often used filters' demerit, this paper proposed one elliptical function based means for getting the fringe's direction, the elliptical error equation for measuring the correlation-ships between pixels on different directions and the center pixel of the filter window was deduced too. Moreover, the paper designed two elliptical equation weighted fringe-preserving filters for interferogram fringe filtering, the filters' computation complexity as well as different weight model's effects were also analyzed. In processing on the Shangyi China area's SAR data collected by ERS-1/2, the new designed filters were proved to be well noise depressing, good fringe preserving, high computing speed, free for configuration and with other merits.