Diffusion-Based Low-Light Image Enhancement with Color and Luminance Priors

2026-02-27

AI Summary

Low-light images often suffer from poor visual quality and degraded performance in downstream tasks due to low contrast, high noise, and color distortion. This work proposes a conditional diffusion-based enhancement method that introduces a novel Structured Control Embedding Module (SCEM) to decompose the input into four physically meaningful priors: illumination, illumination-invariant features, shadow priors, and color-invariant cues. These structured priors guide the U-Net diffusion process to achieve coherent and perceptually faithful enhancement. Notably, this is the first approach to integrate decoupled illumination and color priors into a diffusion framework, enabling strong cross-dataset generalization without fine-tuning. Trained solely on LOLv1, the method achieves state-of-the-art quantitative and perceptual results across multiple benchmarks, including LOLv2-real, LSRW, DICM, MEF, and LIME.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Low-light images often suffer from low contrast, noise, and color distortion, degrading visual quality and impairing downstream vision tasks. We propose a novel conditional diffusion framework for low-light image enhancement that incorporates a Structured Control Embedding Module (SCEM). SCEM decomposes a low-light image into four informative components including illumination, illumination-invariant features, shadow priors, and color-invariant cues. These components serve as control signals that condition a U-Net-based diffusion model trained with a simplified noise-prediction loss. Thus, the proposed SCEM equipped Diffusion method enforces structured enhancement guided by physical priors. In experiments, our model is trained only on the LOLv1 dataset and evaluated without fine-tuning on LOLv2-real, LSRW, DICM, MEF, and LIME. The method achieves state-of-the-art performance in quantitative and perceptual metrics, demonstrating strong generalization across benchmarks. https://casted.github.io/scem/.

Problem

• low-light image enhancement
• color distortion
• noise
• low contrast

Innovation

• diffusion model
• low-light enhancement
• structured control embedding
• physical priors
• color and luminance decomposition