Interference Mitigation for Synthetic Aperture Radar Based on Deep Residual Network

TLDR

Radio Frequency Interference degrades Synthetic Aperture Radar imaging quality, leading to misinterpretation of target scattering characteristics and hindering subsequent image analysis. The study proposes a narrow‑band and wide‑band interference mitigation algorithm based on a deep residual network to address this issue. The algorithm first applies short‑time Fourier transform to characterize interference‑corrupted echoes, uses a deep convolutional neural network to detect interference, then reconstructs time‑frequency features with a ResNet and converts the recovered spectrum back to the time domain via inverse STFT, and is validated on simulated and measured SAR data in strip and TOPS modes. Compared with notch filtering and eigensubspace filtering, the proposed method improves interference mitigation performance while reducing computational complexity.

Abstract

Radio Frequency Interference (RFI) is a key issue for Synthetic Aperture Radar (SAR) because it can seriously degrade the imaging quality, leading to the misinterpretation of the target scattering characteristics and hindering the subsequent image analysis. To address this issue, we present a narrow-band interference (NBI) and wide-band interference (WBI) mitigation algorithm based on deep residual network (ResNet). First, the short-time Fourier transform (STFT) is used to characterize the interference-corrupted echo in the time–frequency domain. Then, the interference detection model is built by a classical deep convolutional neural network (DCNN) framework to identify whether there is an interference component in the echo. Furthermore, the time–frequency feature of the target signal is extracted and reconstructed by utilizing the ResNet. Finally, the inverse time–frequency Fourier transform (ISTFT) is utilized to transform the time–frequency spectrum of the recovered signal into the time domain. The effectiveness of the interference mitigation algorithm is verified on the simulated and measured SAR data with strip mode and terrain observation by progressive scans (TOPS) mode. Moreover, in comparison with the notch filtering and the eigensubspace filtering, the proposed interference mitigation algorithm can improve the interference mitigation performance, while reducing the computation complexity.