With the growing scarcity of spectrum resources, frequency sharing among different systems is becoming quite common, which causes radio frequency interference (RFI) to normal synthetic aperture radar (SAR) applications. Recently, in the RFI suppression field, algorithms based on the low-rank property (LRP) assumption have become a popular research topic. However, this assumption is not valid in complicated cases, especially when there are a wide variety of RFIs. To address this problem, in this article, factors that affect the LRP are studied through theoretical analysis and simulation verification, a general RFI model is established to fit complicated cases, and an advanced RFI suppression scheme based on LRP restoration (LRPR) is proposed. The LRPR scheme has three steps. The first step is RFI localization, which captures the rough profile of the RFI in the time–frequency domain for RFI spectrum separation and extraction. The second step is RFI clustering, in which a novel similarity evaluation metric and corresponding two-step clustering algorithm are designed. The final step is RFI suppression. Operations for LRP recovery in each RFI group are proposed, and consequently, classical low-rank approximation methods are applied for RFI suppression. The simulation results for various types of RFI show the robustness and performance improvement of the proposed scheme. In addition, in LuTan-1 data processing, the proposed scheme outperforms classical algorithms in both intensity image recovery and phase preservation.

中文翻译：

---

复杂低级别违规案件的RFI抑制方案

随着频谱资源的日益稀缺，不同系统之间的频率共享变得相当普遍，这对正常的合成孔径雷达（SAR）应用造成射频干扰（RFI）。近年来，在RFI抑制领域，基于低秩属性（LRP）假设的算法已成为热门研究课题。然而，这种假设在复杂的情况下并不成立，特别是当存在各种各样的 RFI 时。针对这一问题，本文通过理论分析和仿真验证，研究了影响LRP的因素，建立了适合复杂情况的通用RFI模型，并提出了一种基于LRP恢复的高级RFI抑制方案（LRPR）。 LRPR 方案分为三个步骤。第一步是 RFI 定位，它捕获时频域中 RFI 的粗略轮廓，以进行 RFI 频谱分离和提取。第二步是RFI聚类，设计了一种新颖的相似性评估指标和相应的两步聚类算法。最后一步是 RFI 抑制。提出了每个 RFI 组中 LRP 恢复的操作，因此，经典的低秩近似方法被应用于 RFI 抑制。各种类型 RFI 的仿真结果表明了所提出方案的鲁棒性和性能改进。此外，在LuTan-1数据处理中，该方案在强度图像恢复和相位保持方面均优于经典算法。