Learning Structural Illumination for Unsupervised Low-light Enhancement

2026-08-08

AI Summary

Existing unsupervised low-light enhancement methods struggle to disentangle spatially varying relative illumination structure from absolute exposure levels and are often compromised by low signal-to-noise ratio regions; moreover, their reliance on fixed exposure targets limits adaptability across diverse scenes. This work proposes RISE, a novel framework that explicitly decouples relative illumination structure and absolute exposure for the first time, and generates dual photometric exposure references directly from the input image to enable scene-adaptive unsupervised enhancement. By integrating reliable bright-region-guided illumination estimation with spatial propagation modeling, RISE achieves state-of-the-art performance across multiple benchmarks and real-world scenarios, significantly improving interpretability, robustness, cross-scene generalization, and producing visually natural enhanced images.

Abstract

Existing unsupervised low-light image enhancement (LLIE) methods often estimate illumination directly from the entire low-light input, without separating its spatially varying illumination pattern, termed relative illumination structure, from the absolute exposure level or preventing unreliable low signal-to-noise ratio regions from biasing the estimate. Moreover, fixed exposure targets impose a scene-agnostic enhancement criterion, limiting adaptation across diverse lighting conditions. Inspired by the spatial propagation of light, we propose a Relative Illumination Structure Estimation (RISE) framework that decouples relative illumination structure from absolute exposure and infers it from reliable bright regions, enabling interpretable and robust enhancement. For scene-adaptive exposure adjustment, we further propose a Dual-Metering Exposure Reference derived from each input, allowing RISE to adapt the enhancement strength to individual scenes and generalize across diverse lighting conditions. Extensive benchmark and real-world generalization experiments show that RISE achieves state-of-the-art performance among unsupervised LLIE methods while producing visually natural results.

Problem

• low-light image enhancement
• relative illumination structure
• unsupervised learning
• exposure adaptation
• spatially varying illumination

Innovation

• Relative Illumination Structure
• Unsupervised Low-light Enhancement
• Dual-Metering Exposure Reference
• Scene-Adaptive Enhancement
• Illumination Decoupling