Radar Automatic Target Recognition Based on Disjoint Multi-Static K-Space Coverage

This paper considers SAR image reconstruction from incomplete K-space coverage. In particular, the problem of image reconstruction from the type of disjoint K-space coverage that may arise from multi-static SAR systems is addressed. In previous work the effectiveness of reconstruction schemes were assessed in terms of their impact on automatic target recognition performance. It was found that a simple Fourier reconstruction, despite the significantly distorted point spread function that results, has advantages over a compressive sensing approach. In this paper, the idea of re-applying a SAR point spread function to the compressive sensing reconstruction is considered. It is found that this approach provides comparable performance to the use of the Fourier reconstruction. An interpretion of this behavior is provided in terms of a feature-based analysis.