A Bayesian Reconstruction algorithm for Synthesis Aperture Imaging Radiometer

Regularization methods are very efficient in reconstructing the radiometric brightness temperature maps form interferometric measurements. The performance of these methods depends on selecting an optimal regularization parameter, which is typically defined by using the cross-validation or the L-curve. In this paper, we present a novel brightness temperature reconstruction method, which can automatically learn the optimal parameter by Bayesian learning without reducing the performance of the brightness temperature image. To support the theory, numerical simulations are presented and analyzed with emphasis on stability and error analysis.