A Novel Group-Parametric Model for RFI Suppression on Spaceborne SAR

Yuan Mao,Yan Huang,Xutao Yu,Xuezhi Chen,Zaichen Zhang,Zhanye Chen,Wei Hong
DOI: https://doi.org/10.1109/tgrs.2024.3508254
IF: 8.2
2024-01-01
IEEE Transactions on Geoscience and Remote Sensing
Abstract:As an advanced remote sensing technology, synthetic aperture radar (SAR) generates high-resolution images by transmitting continuous electromagnetic waves towards the target area. SAR has played a pivotal role in both contemporary research and practical applications. This underscores the importance of maintaining imaging integrity. However, the performance of SAR systems is severely affected by the increasingly prevalent radio frequency interference (RFI). RFI not only degrades the quality of SAR images but also hinders the accurate interpretation of SAR data. The rapid development and widespread use of modern electromagnetic devices have led to a diversification of interference types, resulting in complex mixed-mode interference. Traditional interference mitigation techniques struggle to effectively alleviate these issues. Moreover, varying terrains add significant difficulty to mitigating interferences, often resulting in residual interference in processed images and the loss of substantial scene information. To tackle these challenges, this paper proposes a novel interference mitigation method called the group-parametric method. Unlike previous semi-parametric methods, the group-parametric method refines both the interference and target models, and achieves more effective interference mitigation and scene preservation by applying distinct regularizations to the refined models. Based on the new model, we have designed a structured tri-factorization (STF) algorithm across frequency and time domain, which achieves data recovery through regularizations of low-rank and sparsity applied to the interference. Experimental verification with Level-1 data from LuTan-1 (LT-1) and Sentinel-1 confirms the effectiveness and superiority of our proposed model and method.
What problem does this paper attempt to address?