Single-Stage Signal Attenuation Diffusion Model for Low-Light Image Enhancement and Denoising

2026-04-07

AI Summary

This work addresses the limitations of existing diffusion-based low-light image enhancement methods, which often rely on two-stage pipelines or auxiliary correction networks that decouple enhancement from denoising, leading to inconsistent optimization objectives. To overcome this, the authors propose the Signal Attenuation Diffusion Model (SADM), which for the first time incorporates a signal attenuation coefficient into the diffusion process to explicitly model low-light degradation. SADM unifies brightness restoration and noise suppression within a single-stage, end-to-end framework based on DDIM, leveraging multi-scale pyramid sampling and physically guided reverse denoising—eliminating the need for additional correction modules or multi-stage training. Experiments demonstrate that SADM achieves superior enhancement quality, enhanced detail preservation, and improved result consistency while maintaining computational efficiency, offering both interpretability and strong performance.

Abstract

Diffusion models excel at image restoration via probabilistic modeling of forward noise addition and reverse denoising, and their ability to handle complex noise while preserving fine details makes them well-suited for Low-Light Image Enhancement (LLIE). Mainstream diffusion based LLIE methods either adopt a two-stage pipeline or an auxiliary correction network to refine U-Net outputs, which severs the intrinsic link between enhancement and denoising and leads to suboptimal performance owing to inconsistent optimization objectives. To address these issues, we propose the Signal Attenuation Diffusion Model (SADM), a novel diffusion process that integrates the signal attenuation mechanism into the diffusion pipeline, enabling simultaneous brightness adjustment and noise suppression in a single stage. Specifically, the signal attenuation coefficient simulates the inherent signal attenuation of low-light degradation in the forward noise addition process, encoding the physical priors of low-light degradation to explicitly guide reverse denoising toward the concurrent optimization of brightness recovery and noise suppression, thereby eliminating the need for extra correction modules or staged training relied on by existing methods. We validate that our design maintains consistency with Denoising Diffusion Implicit Models(DDIM) via multi-scale pyramid sampling, balancing interpretability, restoration quality, and computational efficiency.

Problem

• Low-Light Image Enhancement
• Diffusion Models
• Image Denoising
• Signal Attenuation
• Single-Stage Restoration

Innovation

• Signal Attenuation Diffusion Model
• Single-Stage Restoration
• Low-Light Image Enhancement
• Denoising Diffusion Implicit Models
• Physical Prior Integration