Bosser, Luc; Fiani-Nouvel, Myriam (THALES AIRBORNE SYSTEMS, France)
The paper deals with sparse bandwidth in SAR (Synthetic Aperture Radar) imagery. The problem is considered as an ill-posed inverse problem and its resolution is achieved by Bayesian Regularization method. A reconstruction method of the discontinuous spectrum is proposed and applied on simulated and real SAR data. The reconstructed image reaches similar quality than “standard” images processed from continuous data. The work’s originality is that the result in terms of images quality does not depend on the distribution of the holes in the spectrum. The only limitation comes from the ratio between available spectrum bandwidth versus reconstructed bandwidth.