AstroAI Lunch Talks - March 18, 2024 - Axel Donath
18 Mar 2024 - Joshua Wing
The video can be found here: https://www.youtube.com/watch?v=m9IkPz5HrK4
Speaker: Axel Donath
Title: Joint Deconvolution of Astronomical Images in the Presence of Poisson Noise
Abstract: I’ll present a new method for Joint Likelihood Deconvolution (Jolideco) of a set of astronomical observations of the same sky region in the presence of Poisson noise. The method reconstructs a single flux image from a set of observations by optimizing the a posteriori joint Poisson likelihood of all observations under a patch based image prior. The patch prior is parameterised by a Gaussian Mixture model (GMM) which I trained on astronomical images with high signal to noise ratio, including data from the James Webb Telescope as well as the GLEAM radio survey. During the reconstruction process the patch prior adapts to the patch structures in the data by finding the most likely GMM component for each patch in the image. By applying the method to simulated data I show that both the combination of multiple observations as well as the patch based prior lead to a much improved reconstruction quality in many different source scenarios as well as signal to noise regimes. I also show that the method yields superior reconstruction quality to alternative standard methods such as the Richardson-Lucy method. I will conclude with a few illustrative results of the method applied to example data from the Chandra observatory as well as the Fermi -LAT.
Watch the talk below!