Deep Parametric Imaging for Bistatic SAR: Model, Property, and Approach

Bistatic synthetic aperture radar (BiSAR) parametric imaging can reconstruct the structural information of the target, which is one of the research hotspots for BiSAR imaging. However, compared with monostatic synthetic aperture radar (SAR), there are several challenging problems to be faced in terms of BiSAR parametric imaging, such as complex echo models, compute and storage burden, and azimuth-dependent phase (ADP). To this end, we first propose the BiSAR parametric echo model, followed by the analysis of the characteristics of BiSAR parametric imaging. Based on the above-mentioned analysis, we propose a deep adaptive BiSAR parametric imaging network (DAPI-net), which consists of three parts, namely, the adaptive dimensionality reduction module (ADRM), the parameter estimation module (PEM), and the image reconstruction module (IRM). The main idea of DAPI-net is to employ coarse imaging to determine the approximate range of target parameters and construct a local observation matrix to reduce computational complexity. On this basis, a variant observation matrix learned solver and ADP compensation unit are used to achieve high-precision BiSAR target parameter estimation. Among them, the ADP compensation unit mainly mitigates echo model mismatch caused by ADP and improves the parameter estimation performance. Finally, BiSAR parametric image output is achieved through image dilation. The efficacy of DAPI-net is validated through simulation experiments and microwave anechoic chamber data. The proposed algorithm not only addresses the identified challenges but also advances the state-of-the-art in BiSAR parametric imaging.