Interference Mitigation for Synthetic Aperture Radar Data using Tensor Representation and Low-Rank Approximation

Radio frequency interference (RFI) is a critical issue to synthetic aperture radar (SAR), which would cause great distortions to amplitude and phase information of the received echoes. Most of the existing literatures deal with the interference separation problem in time domain, frequency domain, or time-frequency domain using the matrix representation and matrix optimization tools, without further exploiting the correlation among multiple dimensional measurements. This paper proposes an interference separation for SAR data using tensor representation by formulating a novel time-frequency azimuth tensor. Then, the low-rank property of the interference is utilized and the interference contribution is estimated using low rank tensor approximation. Experimental results demonstrate that the interference components is effectively extracted, and well imaging results could be recovered.