Deconvolution with hybrid parameterizations for radio emission reconstruction

Abstract

Abstract Deconvolution is used to eliminate imperfections in the point spread function, such as sidelobes caused by incomplete sampling of radio telescopes, and is a key technology for radio synthesis imaging. Modern telescopes have such high sensitivities that the observed celestial images may contain both compact and diffuse emission. This essentially requires deconvolution technology to have the ability to model both. In this paper, a deconvolution algorithm based on hybrid parameterization is proposed for the rapid reconstruction of complex celestial structures. In this algorithm, scale-free parameterization is utilized to reconstruct compact emission, while multi-scale parameterization is employed to reconstruct diffuse emission. Simulated data representing Square Kilometre Array (SKA) observations with realistic celestial brightness distributions are applied to test the performance of the algorithm. Our experiments show that, compared with other state-of-the-art deconvolution algorithms, the algorithm proposed in this paper can reconstruct complex celestial structures well and provide competitive reconstruction results while greatly improving the reconstruction speed.