SAR imaging from randomly sampled phase history using compressive sensing

Reconstructing synthetic aperture Radar (SAR) images from gapped phase history or k space data, is a major problem for SAR engineers. In this work we use the newly proposed compressive sensing (CS) algorithms to form SAR images of randomly and sparsely sampled k space data. We also investigate the effect of adding phase noise of various degrees of severity in the sparse and random k space data. We show that CS based algorithms can intelligibly reconstruct SAR images from randomly sparse phase history data and can tolerate a good amount of phase noise corruption. Dantzig selector based CS algorithm was found to perform better than the usual l₁ norm based CS algorithm.