SparseRI: A Compressed Sensing Framework for Aperture Synthesis Imaging in Radio Astronomy

In radio interferometry, information about a small region of the sky is obtained in the form of samples in the Fourier transform domain of the desired image. Since this sampling is usually incomplete, the missing information has to be reconstructed using additional assumptions about the image. The emerging field of compressed sensing provides a promising new approach to this type of problem that is based on the supposed sparsity of natural images in some transform domain. We present a versatile CS-based image reconstruction framework called SparseRI, an interesting alternative to the CLEAN algorithm, which permits a wide choice of different regularizers for interferometric image reconstruction. The performance of our method is evaluated on simulated data as well as on actual radio interferometry measurements from the VLA, showing that our algorithm is able to reproduce the main features of the test sources. The proposed method is a first step toward an alternative reconstruction approach that may be able to avoid typical artifacts like negative flux regions, to work with large fields of view and noncoplanar baselines, to avoid the gridding process, and, in particular, to produce results not far from those achievable by human-assisted processing in CLEAN through an entirely automatic algorithm, making it especially well suited for automated processing pipelines.