A Simple Approach for Adaptive Noise Filtering of InSAR Interferometric Phase Images

摘要

Interferometric Synthetic Aperture Radar (InSAR) can provide three-dimensional digital topographic maps globally with very high spatial resolution and vertical accuracy by extracting topographic information from the interferometric phase image. Unfiltered phase image is very noisy and appears a fringe pattern whose fringe density depends on the actual surface topography. The more drastically the topographic change, the finer the fringe spacing. How to remove the phase noise as much as possible and preserve the fringe edges well simultaneously is very crucial. According to the noise distribution of interferometric phase image, averaging filtering is an efficient means to filter out the phase noise, but at the cost of spatial-resolution-loss if the window-length of the filter is improper. In this paper, we have developed an algorithm to calculate the number of fringes along the range direction and modify the length of filtering window adaptively. Thus, the sparser the fringes, the longer the length of the filtering window. Otherwise, the window-length becomes shorter or even zero if the fringes are dense. Interferometric phase images got from SIR-C/X SAR data has been used to illustrate the effectiveness of this means.