![[2601.23276] Denoising the Deep Sky: Physics-Based CCD Noise Formation for Astronomical Imaging](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2601.23276] Denoising the Deep Sky: Physics-Based CCD Noise Formation for Astronomical Imaging
Summary
This article presents a physics-based framework for synthesizing CCD noise in astronomical imaging, addressing noise limitations in current calibration methods and enhancing data for supervised learning.
Why It Matters
The research highlights a significant advancement in astronomical imaging by providing a method to create paired datasets for training machine learning models. This can improve the accuracy of astronomical observations and data analysis, which is crucial for scientific discoveries in astrophysics.
Key Takeaways
- Introduces a physics-based noise synthesis framework for CCD noise in telescopes.
- Addresses limitations of existing calibration methods that leave stochastic noise unresolved.
- Demonstrates effectiveness through extensive experiments on real-world multi-band datasets.
- Enables the construction of abundant paired datasets for supervised learning.
- Aims to enhance both photometric and scientific accuracy in astronomical imaging.
Astrophysics > Instrumentation and Methods for Astrophysics arXiv:2601.23276 (astro-ph) [Submitted on 30 Jan 2026 (v1), last revised 26 Feb 2026 (this version, v2)] Title:Denoising the Deep Sky: Physics-Based CCD Noise Formation for Astronomical Imaging Authors:Shuhong Liu, Xining Ge, Ziying Gu, Quanfeng Xu, Lin Gu, Ziteng Cui, Xuangeng Chu, Jun Liu, Dong Li, Tatsuya Harada View a PDF of the paper titled Denoising the Deep Sky: Physics-Based CCD Noise Formation for Astronomical Imaging, by Shuhong Liu and 9 other authors View PDF HTML (experimental) Abstract:Astronomical imaging remains noise-limited under practical observing conditions. Standard calibration pipelines remove structured artifacts but largely leave stochastic noise unresolved. Although learning-based denoising has shown strong potential, progress is constrained by scarce paired training data and the requirement for physically interpretable models in scientific workflows. We propose a physics-based noise synthesis framework tailored to CCD noise formation in the telescope. The pipeline models photon shot noise, photo-response non-uniformity, dark-current noise, readout effects, and localized outliers arising from cosmic-ray hits and hot pixels. To obtain low-noise inputs for synthesis, we stack multiple unregistered exposures to produce high-SNR bases. Realistic noisy counterparts synthesized from these bases using our noise model enable the construction of abundant paired datasets for supervised learning. Ex...
