干涉SAR相位条纹的鲁棒加权圆周期滤波

对相位条纹的降噪滤波，是干涉SAR成像中必须的步骤之一，由于相位条纹自身的特点，通常的滤波方法不能用来实现这一滤波。本文首先阐述了相位条纹的主要特点，圆周期性，各点可信度不同和噪声分布拖尾；然后详细介绍了已有的周期均值滤波法，包括对其原理和性能的讨论；在此基础上，给出了一种新的鲁棒加权圆周期滤波方法，包拓该方法的指导思想和具体实现，最后给出了将新老方法用于由SIR－C／X－SAR原始数据产生的相位条纹的降噪滤波的实验结果，并对比了滤波效果。

Robust Weighted Periodic Pivoting Filtering of Phase Fringe for InSAR

Noise reducing of interferometric phase fringe is a nec essary step for the image formation of InSAR, but general low-pass filtering ca n not apply it due to its specific features. Firstly, characte ristics of the phase fringe, such as circular period, uneven credibility and he avy-tailed distribution are explained. Secondly, a usual periodic pivoting filt ering is discussed in detail, including its principle and effect. Then, a new ro bust weighted periodic pivoting filtering is designed and its principle is disc ussed; Finally, experimental results of the two filters, using the phase fringe produced from the raw data of SIR-C/X-SAR, are compared.