Learning Low Rank Approximation IAA for Forward-Looking Scanning Radar Super-Resolution Imaging

In forward-looking scanning radar imaging, spectral estimation method can improve the azimuth resolution. However, when dealing with large-scale echo data, the computational complexity is significantly elevated due to the inversion of high-dimensional matrices. To solve these problems, this paper proposes a low rank approximate iterative adaptive angular (IAA) super-resolution imaging method based on deep networks. By combining the advantages of data driven and model driven, this method avoids the problems of manual parameter selection and slow iteration update in traditional algorithm. At the same time, due to the learning and fitting ability of deep networks, the proposed method can maintain good angular super-resolution performance. Simulation results demonstrate that the algorithm exhibits reduced computational complexity and improved resolution compared to conventional methods.