Sparse target scene reconstruction for SAR using range space rotation

This paper considers target estimation in Spotlight Mode Synthetic Aperture Radar. Assuming that the number of targets inside the scene of interest is small, compressive sensing (CS) theory can be applied for target estimation. CS estimation relies on a grid applied on the target space and assumes that the targets fall on grid points. However, as the grid size decreases in order to capture closely spaced targets, the coherence of the sensing matrix increases, thus resulting in CS performance degradation. In this paper, we propose to use Range Space Property (RSP) Rotation of the dictionary matrix as means of solving the underlying ℓ1-problem, even in cases in which the sensing matrix coherence is high. The proposed approach can resolve closely spaced targets.