2-Shots in the Dark: Low-Light Denoising with Minimal Data Acquisition

2025-12-02

AI Summary

In low-light imaging, photon scarcity results in extremely low signal-to-noise ratios (SNR) in raw images, while existing deep denoising methods rely on large-scale paired clean-noisy datasets that are prohibitively difficult to acquire. To address this, we propose a novel data synthesis framework requiring only a single real noisy image and its corresponding dark frame. Our method first models signal-dependent noise via a Poisson distribution, then performs spectral sampling in the Fourier domain to generate realistic, signal-independent noise that preserves both authentic statistical properties and spatial structure. Unlike conventional approaches, it avoids oversimplified noise assumptions and eliminates the need for large-scale paired training data. Evaluated on multiple low-light denoising benchmarks, our method achieves state-of-the-art performance. It significantly reduces data acquisition cost while maintaining superior denoising fidelity.

Technology Category

• Computer Vision: Computational Photography, Image & Video Synthesis
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Security and Privacy: Data transparency and provenance
• Web Mining and Content Analysis: Web data generation and simulation

Abstract

Raw images taken in low-light conditions are very noisy due to low photon count and sensor noise. Learning-based denoisers have the potential to reconstruct high-quality images. For training, however, these denoisers require large paired datasets of clean and noisy images, which are difficult to collect. Noise synthesis is an alternative to large-scale data acquisition: given a clean image, we can synthesize a realistic noisy counterpart. In this work, we propose a general and practical noise synthesis method that requires only one single noisy image and one single dark frame per ISO setting. We represent signal-dependent noise with a Poisson distribution and introduce a Fourier-domain spectral sampling algorithm to accurately model signal-independent noise. The latter generates diverse noise realizations that maintain the spatial and statistical properties of real sensor noise. As opposed to competing approaches, our method neither relies on simplified parametric models nor on large sets of clean-noisy image pairs. Our synthesis method is not only accurate and practical, it also leads to state-of-the-art performances on multiple low-light denoising benchmarks.

Problem

• Develops a noise synthesis method using minimal data
• Models signal-dependent and independent noise accurately
• Achieves state-of-the-art low-light denoising without large datasets

Innovation

• Uses single noisy image and dark frame per ISO
• Models noise with Poisson distribution and Fourier sampling
• Generates diverse noise realizations without clean pairs