Wideband Interference Mitigation for Synthetic Aperture Radar Data Based on Variational Bayesian Inference

The existence of wideband interference (WBI) would seriously reduce the SAR imaging quality and the following image interpretation accuracy. However, it is difficult to mitigate WBI owing to its large bandwidth and severe overlapping with useful signal. This paper proposes a WBI mitigation algorithm based on variational Bayesian inference. Firstly, a low-rank matrix factorization model for WBI is established according to the low rank characteristics of WBI in time-frequency domain. Then, we build the Bayesian posterior probability model for the low rank matrix factorization. Finally, the variational Bayesian inference is utilized to estimate the model parameters and reconstruct the WBI. The experimental results of WBI mitigation using measured WBI data acquired by the sentinel-1 satellite have verified the effectiveness of the proposed algorithm.