Bayesian Deconvolution of Astronomical Images with Diffusion Models: Quantifying Prior-Driven Features in Reconstructions

Alessio Spagnoletti, Alexandre Boucaud, Marc Huertas-Company, Wassim Kabalan, Biswajit Biswas
11/28/2024
astro-ph.IMastro-ph.GAcs.CV

No official code found. Best implementation: astrodeepnet/diffusion4astro (11 stars, MIT, last commit 2025-07-03).

Abstract

Deconvolution of astronomical images is a key aspect of recovering the intrinsic properties of celestial objects, especially when considering ground-based observations. This paper explores the use of diffusion models (DMs) and the Diffusion Posterior Sampling (DPS) algorithm to solve this inverse problem task. We apply score-based DMs trained on high-resolution cosmological simulations, through a Bayesian setting to compute a posterior distribution given the observations available. By considering the redshift and the pixel scale as parameters of our inverse problem, the tool can be easily adapted to any dataset. We test our model on Hyper Supreme Camera (HSC) data and show that we reach resolutions comparable to those obtained by Hubble Space Telescope (HST) images. Most importantly, we quantify the uncertainty of reconstructions and propose a metric to identify prior-driven features in the reconstructed images, which is key in view of applying these methods for scientific purposes.

View on arXivView PDF

Code Implementations(1)

astrodeepnet/diffusion4astroCommunity100%
MIT

Official repository for "Bayesian Deconvolution of Astronomical Images with Diffusion Models: Quantifying Prior-Driven Features in Reconstructions" (NeurIPS2024 - Machine Learning and the Physical Sciences workshop)

111HTML, PythonApr 11, 20241 years agoMIT

Similar Papers

An Expectation-Maximization Algorithm for Training Clean Diffusion Models from Corrupted Observations

83% match3 repos52026

Learning Image Priors through Patch-based Diffusion Models for Solving Inverse Problems

83% match2026

Local Patches Meet Global Context: Scalable 3D Diffusion Priors for Computed Tomography Reconstruction

83% match2025

Gaussian is All You Need: A Unified Framework for Solving Inverse Problems via Diffusion Posterior Sampling

83% match1 repo2024

Using Galaxy Evolution as Source of Physics-Based Ground Truth for Generative Models

83% match1 repo12024

Solving Inverse Problems with Latent Diffusion Models via Hard Data Consistency

83% match1 repo872026

Cite this paper

@article{spagnoletti2024bayesian,
  title  = {Bayesian Deconvolution of Astronomical Images with Diffusion Models: Quantifying Prior-Driven Features in Reconstructions},
  author = {Alessio Spagnoletti and Alexandre Boucaud and Marc Huertas-Company and Wassim Kabalan and Biswajit Biswas},
  year   = {2024},
  eprint = {2411.19158},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2411.19158v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2411.19158.svg)](https://codeofpaper.com/paper/2411.19158)

Discussion