Super-resolution in SAR imaging: Analysis with the atomic norm

In this paper, we investigate the synthetic aperture radar (SAR) imaging problem via sparse atomic norm reconstruction. A stepped-frequency radar operation and a side looking radar is assumed. However, the analysis presented is general, and extensions to other SAR types, scenes and other radar waveforms are straightforward. The atomic norm is formulated and employed as a penalizer for SAR denoising and reconstruction, as a convex problem. The target positions are readily estimated through the peaks of the dual polynomial. Furthermore, due to strong duality in our convex formulation, we show that the target positions are obtained from either the primal or dual problem solution. Through simulations, we demonstrate the advantages of our approach when compared with traditional back-projection imaging and recent sparse reconstruction techniques.