RFI Extraction for Synthetic Aperture Radar in 2D Domain Based on Low Rank and Sparsity Property

Abstract

Radio frequency interference (RFI) is a core issue of synthetic aperture radar (SAR), which significantly reduces the signal-to-noise ratio of SAR echo and image interpretation accuracy. Therefore, RFI mitigation plays an important role in the SAR imaging. Based on the azimuth echo analysis, some methods based on low rank characteristic or sparsity are introduced to reconstruct RFI and recover target echo signal. However, the property of RFI described in these models is not accurate enough, and there is a large signal loss problem. In this paper, an interference extracted algorithm is introduced for SAR data based on low rank and sparsity property. Via the measured SAR data analysis, a separation optimization model is established joint the low rank characteristic for RFI and sparsity assumption for the target echo signal. And the optimization problem can be solved iteratively by bilateral random projection and soft threshold mapping. Meanwhile, the mask procession is used to constrain the location of RFI in 2-dimensional domain to improve the reconstruction accuracy. Finally, the RFI mitigation experiments of the measured SAR data verify the effectiveness of the proposed algorithm.