A modification to the ASM filter for improving SAR interferograms

Abstract

SAR Interferograms illustrate an ambiguous (modulo 2π) and noisy phase. In this paper, we focus on the step of interferogram denoising using the Adaptive Switching Median Filter (ASMF) in the wavelet domain. Thus, we propose to filter the coefficients of the relative Discrete Packet Wavelet Transform (DPWT). Our main contribution in this paper concerns firstly, the methodology for computing the mask of noise corresponding to the InSAR phase. Secondly, the size of the median filter is computed considering the noise mask within a given neighborhood and taking into account the corresponding InSAR coherence values. This scheme is tested on simulated noisy interferograms as well as on a given pairs of single look complex (SLC) data from Envisat satellite. Validation was made by computing the Digital Elevation Model after unwrapping the filtered interferogram.