Bayesian Reconstraction with Adaptive Image Notion

Bayesian probability theory is employed to provide a stable and unique solution to the ill-posed inverse problem of image analysis applied to ALMA calibrated interferometric data sets. The novel algorithm, Bayesian Reconstruction with Adaptive Image Notion (BRAIN), is under development aiming at detecting the weak signal in the presence of varying noise also in extreme cases as sparse data and mosaic of images. The technique makes use of Gaussian statistics for a joint source detection and background estimation through a probabilistic mixture-model technique. Statistics is rigorously applied throughout the algorithm, so pixels with low intensity can be handled optimally and accurately, without binning and loss of resolution. BRAIN makes use of a 2D adaptive kernel deconvolution method to prevent spurious signal that may arise from the deconvolution process of the dirty image from the well-known dirty beam. Continuum, emission and absorption lines detection occurs without an explicit subtraction. An automated decision in the separation between different kind of signals is developed while minimizing spurious detections. BRAIN has the objective to become a new CASA task. Preliminary results are shown on the application of BRAIN to ALMA simulated data.