Study on Sparse High Resolution SAR Imaging Method Based on Compound Regularization

‍ ‍With the continuous improvement of high-resolution earth observation requirements， the application of synthetic aperture radar （SAR） will become more extensive. Considering the issues with high-resolution SAR imaging， such as the need for large amounts of data， high storage difficulty， and long calculation time， the commonly used solution involves introducing compressed sensing （CS） into the SAR imaging model to reduce the sampling rate and reduce the amount of data. Using a single regularization as a constraint condition can suppress the side lobes of point targets and achieve point target feature enhancement. However， since multiple target types may be present in the observation scene， adhering to the multiple feature enhancement requirements using a single regularization constraint is challenging. Therefore， this study proposes a sparse， high-resolution SAR imaging method based on composite regularization. When using compressed sensing to reduce the number of calculations， the linear combination of multiple regularizations is used as a constraint condition to enhance the characteristics of different types of targets in the observation scene， and the requirements of high-resolution earth observation in complex scenes are achieved. The sparse SAR imaging method based on composite regularization introduces non-convex regularization and total variation （TV） regularization as constraints in the sparse SAR imaging model to reduce the sparse reconstruction error， enhance the characteristics of regional targets， reduce the influence of noise on the imaging results， and improve the imaging quality. The improved alternating direction method of multipliers （ADMM） is used to reduce the composite regularization constraint， which reduces the calculation time and reconstructs the image quickly. The azimuth-range decoupling operator is used to replace the observation matrix and its conjugate transpose to further reduce the computational complexity. The simulation and measured data experiments show that the proposed algorithm can enhance the features of point targets and regional targets， reduce computational complexity， improve convergence performance， and achieve fast and high-resolution image reconstruction.