Compressive High Range Resolution Radar Imaging Based on Continuous Dictionary

In this paper, we consider high range resolution radar imaging via compressed sensing. The conventional compressive imaging method assumes that the target to be recovered lies on a prior known grid. However, this condition usually cannot be satisfied in reality. To address this issue, the paper adopts a continuous sparse representation model also known as continuous dictionary which can take continuous value in parameter space and has no gridding induced error. We choose the atomic norm minimization to promote sparsity for sparse recovery and present an efficient algorithm using alternating direction method of multipliers to solve the equivalent semidefinite programming problem. Experimental results based on both synthetic and high frequency electromagnetic prediction data validate its higher reconstruction accuracy compared with the conventional methods.