ABSTRACT

The Goldstein filter was a well-known approach of synthetic aperture radar interferogram filtering for its simplicity and high efficiency. The filtering performance of Goldstein and its improved filters is mostly affected by several parameters, such as the noise suppression parameter α, the sliding patch size, and the kernel window size. Yet, the fixed parameters will cause the filtering intensity to mismatch the noise levels of different regions. In this paper, we propose an adaptive patch-based Goldstein filter (AP-Goldstein filter), which takes the pseudo-variation coefficient as the measure of patch homogeneity and adaptively selects the patch sizes in different regions of interferogram. This filter controls the noise suppression parameter α by pseudo-coherence, which is a common and effective indicator applied to filter the interferogram phase. Meanwhile, the previously determined patch size of each pixel is used to confirm the kernel window size at the same region. The proposed and other advanced filters were tested on real and simulated interferograms. The qualitative and quantitative comparison of the experimental results shows that this new method can better restrain the phase noise of interferogram while maintaining the edge detailed information and can improve the automation of filtering.

摘要

Goldstein 滤波器因其简单和高效而成为众所周知的合成孔径雷达干涉图滤波方法。Goldstein 及其改进滤波器的滤波性能主要受噪声抑制参数α、滑动块大小和内核窗口大小等几个参数的影响。然而，固定的参数会导致过滤强度与不同区域的噪声水平不匹配。在本文中，我们提出了一种基于补丁的自适应 Goldstein 滤波器（AP-Goldstein 滤波器），它以伪变异系数作为补丁均匀性的度量，并自适应地选择干涉图不同区域的补丁大小。该滤波器控制噪声抑制参数α通过伪相干，这是一种常用的有效指标，用于过滤干涉图相位。同时，预先确定的每个像素的补丁大小用于确认同一区域的内核窗口大小。所提出的和其他先进的滤波器在真实和模拟干涉图上进行了测试。实验结果的定性和定量比较表明，该新方法在保留边缘细节信息的同时，能较好地抑制干涉图的相位噪声，提高滤波的自动化程度。