Bayesian compressive sensing in synthetic aperture radar imaging

To achieve high-resolution two dimension images, synthetic aperture radar (SAR) with ultra wide-band faces considerably technical challenges such as long data collection time, huge amount of data storage and high hardware complexity. In these years, several imaging modalities based on compressive sensing (CS) have been proposed which can provide high-resolution images using significantly reduced number of samples. However, the CS-based methods are sensitive to noise and clutter. In this study, a new imaging modality based on Bayesian compressive sensing (BCS) is proposed along with a novel compressed sampling scheme. Clutter, which the previous CS-based methods not considered, is also included in this study. This new imaging scheme requires minor change to traditional system and allows both range and azimuth compressed sampling. Also, the Bayesian formalism accounts for additive noise encountered in the compressed measurement process. Experiments are carried out with noisy and cluttered imaging scenes to verify the new imaging scheme. The results indicate that the Bayesian formalism can provide a sharp and sparse image absence of side-lobes, which is the common problem in conventional imaging methods and has fewer artifacts compared with the previous version of CS-based methods.