BGRem: A background noise remover for astronomical images based on a diffusion model

¹ Institute for Mathematics, Astrophysics and Particle Physics (IMAPP), Radboud University Nijmegen, Heyendaalseweg 135, 6525 AJ Nijmegen, The Netherlands
² Nikhef, Science Park 105, 1098 XG Amsterdam, The Netherlands
³ Department of Physics, University of Oxford, Denys Wilkinson Building, Keble Road, Oxford OX1 3RH, UK
⁴ University of Nova Gorica, Centre for Astrophysics and Cosmology, Vipavska 11c, Ajdovščina, Slovenia
⁵ Instituto de Física Corpuscular (IFIC), CSIC-UV, Spain
⁶ Department of Astronomy and Inter-University Institute for Data Intensive Astronomy, University of Cape Town, Private Bag X3, Rondebosch 7701, South Africa
⁷ South African Astronomical Observatory, PO Box 9, Observatory, Cape Town 7935, South Africa

^★ Corresponding authors: This email address is being protected from spambots. You need JavaScript enabled to view it. ; This email address is being protected from spambots. You need JavaScript enabled to view it.

Received: 16 October 2025
Accepted: 28 March 2026

Abstract

Context. Astronomical imaging aims to maximize signal capture while minimizing noise. It is difficult and expensive to enhance the signal-to-noise ratio directly on detectors, which has led to extensive research into advanced post-processing techniques.

Aims. Removing background noise from images is a valuable preprocessing step for catalog-building tasks. We introduce BGRem, a machine-learning (ML)-based tool to remove background noise from astronomical images. Our aim is to improve image quality and enhance the performance of the subsequent analysis pipeline, from detecting faint sources to performing source characterization tasks.

Methods. The BGRem tool uses a diffusion-based model with an attention U-Net as backbone, trained on simulated images for optical and gamma (γ)-ray data from the MeerLICHT and Fermi-LAT telescopes. The tool learns to denoise astronomical images in a supervised manner over several diffusion steps. We performed preprocessing and postprocessing techniques, including normalization and median subtraction, on these images to make them suitable for the analysis pipeline.

Results. We compared the performance of BGRem with SourceExtractor (SExtractor), a widely used tool for cataloging astronomical sources. The number of true positive sources using SExtractor increased by about 7% for MeerLICHT data when we used BGRem as a preprocessing step. We also show the generalizability of BGRem by testing it with optical images from different telescopes and on simulated γ-ray data representative of the Fermi-LAT telescope. In both cases, BGRem improves the source detection efficiency.

Conclusions. The BGRem tool improves the source detection accuracy of traditional pixel-based methods by removing complex background noise. Using zero-shot approach, BGRem generalizes well to a wide range of optical images. The successful application of BGRem to simulated γ-ray images, alongside optical data, demonstrates its adaptability to distinct noise characteristics and observational domains. This cross-wavelength performance highlights its potential as a general-purpose background removal framework for multiwavelength astronomical surveys.