Super-resolution Imaging Method for Swarm Targets Based on Group Lasso

Drone swarms are increasingly prevalent in military and civilian use scenarios, thus their effective detection and resolution are highly desirable. Traditional super-resolution imaging methods rarely utilize the prior information of the swarm distribution, thus the structural information of the target is prone to loss. Also, they cannot simultaneously meet the requirements of high resolution, adaptation to different signal-to-noise ratios (SNR), etc. In this paper, we propose a super-resolution imaging method based on the clustering and individual sparsity properties of swarm targets. We introduce group Lasso, a mixed $l_{1}$/$l_{f}$ regularization method to preserve the swarm structure and contour. Firstly, group Lasso is used as a regularization constraint to establishing the objective function and the alternating direction method of multipliers is used to solve the optimization problem. Then the swarm contours and quantities are obtained by constant false alarm detection, clustering, and convex hull algorithms in turn. Finally, simulation results are given under different swarm shapes and SNRs to verify the effectiveness of the proposed algorithm.