Mutual Terrain Scattered Interference Suppression for SAR Image via Multi-view Subspace Clustering

With the development of satellite constellations and fierce competition for limited spectrum resources, Mutual Terrain Scattered Interference (MTSI) has become an emerging issue for spaceborne SAR systems. Existing mitigation methods mainly focus on strong wideband MTSI that satisfies the low-rank property. However, in most scenarios, MTSI presents as weak wideband or ultrawideband interference occupying most of the spectrum, violating the low-rank assumption. This paper introduces two mitigation schemes employing the multi-view subspace representation to tackle these challenges. The first scheme divides the spectrum to construct a clean dictionary composed of subspaces with high correlation, which makes it possible to mitigate wideband MTSI by sparse constraint. Further, based on the differences in amplitude statistical characteristics between polluted and clean pulses, the second scheme utilizes histogram normalization to construct a clean dictionary from the polluted spectrum. Therefore, the wideband and ultrawideband MTSI could be mitigated by solving the subspace clustering problem. Experimental results in the simulated and real measured Sentinel-1 and GaoFen-3 data demonstrate superior image quality improvement by the proposed mitigation schemes.