SAR Image Reconstruction Method of Incomplete Raw Data Based on Matrix Completion

This paper discusses a new approach of SAR image reconstruction based on matrix completion from afew number of samples of the received SAR signal required by the Nyquist
theorem in both the azimuthand range. The reconstruction method has goal to fill the missing information in the raw data caused byunder sampling based on matrix completion of Fourier coefficients. Experiments were performed on the simulated point target and the real RADARSAT1data in noise freeand noisy condition. Compared with the conventional method as range doppler algorithm, the proposed method presented slightly better results in suppressing of side lobe and canmaintain resolution SAR imagery, but with the advantage with a few number of measurements. Asexpected, the larger the number of measurements, a target SAR image can be reconstructed rarelybetter. The experiments on the RADARSAT1data showed the almost same performance as in thesimulation on the target points.