A Faceted Prior for Scalable Wideband Imaging: Application to Radio Astronomy

Wideband radio-interferometric (RI) imaging consists in estimating images of the sky across a whole frequency band from incomplete Fourier data. Powerful prior information is needed to regularize the inverse imaging problem. At the extreme resolution and dynamic range of interest to modern telescopes, image cubes will far exceed Terabyte sizes, with data volumes orders of magnitude larger, making image estimation a very challenging task. The computational cost and memory requirements of corresponding iterative image recovery algorithms are extreme and call for high parallelism. A data-splitting strategy was recently introduced to parallelize computations over data blocks within an advanced primal-dual convex optimization algorithm. Building on the same algorithm, we propose an image faceting approach that consists in splitting the image cube into 3D overlapping facets with their own prior, reducing the computational bottleneck from full image to facet size. Simulation results suggest our prior provides similar if not superior reconstruction quality to the corresponding state-of-the-art non-faceted approach, with facet parallelization offering acceleration and therefore increased potential of scalability to large data and image sizes.