A Sparse Reconstruction Algorithm for Multi-Frequency Radio Images

In radio interferometry, every pair of antennas in an array defines one sampling point in the Fourier domain of the sky image. By combining information from different wavelengths, sample coverage - and therefore reconstruction quality - can be increased. However, the images at different wavelengths can be dramatically dissimilar; this fact must be taken into account when reconstructing multi-frequency images. In this paper, we present a novel reconstruction algorithm based on the assumption that the spectrum is continuous. In contrast to prior work, we allow for sparse deviations from this assumption: this allows, for example, for accurate reconstruction of line spectra superimposed on a continuum. Using simulated measurements on synthetic multi-frequency images, we show that the proposed approach provides significant improvements over a comparable method based solely on a continuity assumption.