An improved SAR interferogram denoising method based on principal component analysis and the Goldstein filter

作者:Wang, Bao-Hang; Zhao, Chao-Ying*; Liu, Yuan-Yuan
来源:Remote Sensing Letters, 2018, 9(1): 81-90.
DOI:10.1080/2150704X.2017.1392633

摘要

Interferogram filtering is an important data processing step in Interferometric synthetic aperture radar (InSAR) applications, which has a direct impact on the accuracy of the phase unwrapping and digital elevation model (DEM) or deformation results retrieval. An improved synthetic aperture radar (SAR) interferogram denoising method based on principal component analysis and the Goldstein filter is proposed, which can improve the coherence of interferogram remarkably and get more coherent targets. First, homogeneous pixels are identified with stacks of SAR amplitude data, which can obtain the unbiased coherence estimation. Then, the noise phase of one resolution unit is suppressed based on the principal component analysis of multi-baseline InSAR coherence stacks by considering the relationship between pixel size and scattering mechanism. Finally, the remaining noise is smoothed with the iterative Goldstein filter over spatial domain. The proposed method is tested over one deformed and low-coherence region to verify the better performance in the terms of noise reduction and coherence increase.