Characterization and Removal of RFI Artifacts in Radar Data via Model-Constrained Deep Learning Approach

Microwave remote sensing instruments such as synthetic aperture radar (SAR) play an important role in scientific research applications, while they suffer great measurement distortion with the presence of radio frequency interference (RFI). Existing methods either adopt model?based optimization or follow a data?driven black?box learning scheme, and both have specific limitations terms efficiency, accuracy, interpretability. In this paper, we propose hybrid model?constrained deep approach for RFI extraction mitigation by fusing classical model-based advanced data-driven method. Considering temporal-spatial correlation target response, well random sparsity property time?varying interference, joint low?rank sparse framework is established. Instead applying iterative process uncertain convergency, proposed scheme approximates stacked recurrent neural network. By adopting strategy, original unsupervised decomposition problem converted to supervised problem. Experimental results show validity method under diverse scenarios, which could avoid manual tuning model hyperparameters speed up efficiency.