http://arxiv.org/abs/1802.02013
Astronomical imaging based on photon count data is a highly non-trivial task. In this context we show how to denoise, deconvolve, and decompose multidimensional photon observations. The primary objective is to incorporate accurate and well motivated likelihood and prior models in order to give reliable estimates about morphologically different but superimposed photon flux components present in the data set. Thereby we denoise and deconvolve single photon counts, while simultaneously decomposing them into diffuse, point- like and uninteresting background radiation fluxes. The decomposition is based on a probabilistic hierarchical Bayesian parameter model within the framework of Information field theory (IFT). In contrast to its predecessor D3PO, its successor D4PO can reconstruct components which depend on multiple parameters, each defined over its own independent manifold, for example location and energy. Beyond that, D4PO may reconstruct correlation structures over each of the component manifolds separately. Thereby the inferred correlations implicitly define the morphologically different source components, except for the spatial correlations of the point-like flux. These are by definition spatially uncorrelated. The capabilities of the algorithm are demonstrated alongside a realistic high energy photon count data set. D4PO successfully denoised, deconvolved, and decomposed a single photon count image into diffuse, point-like and background flux. Further the algorithm provided a posteriori uncertainty estimates of the reconstructions and the correlation structure of the fields with respect to the manifolds they reside on.
D. Pumpe, M. Reinecke and T. Ensslin
Wed, 7 Feb 18
29/86
Comments: N/A
You must be logged in to post a comment.