Luminance-Aware Statistical Quantization: Unsupervised Hierarchical Learning for Illumination Enhancement

2025-11-03

AI Summary

Low-light image enhancement (LLIE) faces a fundamental trade-off between reconstruction fidelity and cross-scene generalization, primarily because prevailing methods rely on deterministic pixel-wise mappings learned under paired supervision, neglecting the continuous physical nature of illumination variation. To address this, we propose a reference-free framework for luminance distribution modeling: we formulate illumination transformation as a hierarchical power-law distribution in intensity space and introduce a luminance-aware statistical quantization mechanism that converts deterministic mapping into probabilistic sampling across continuous brightness layers. Coupled with an unsupervised diffusion forward process, our method autonomously learns optimal luminance transition paths without requiring normal-light reference images. Enhancement is achieved via hierarchical power-function approximation and unsupervised distribution modeling. Experiments demonstrate state-of-the-art performance on both supervised and reference-free benchmarks, with significant improvements in generalization to real-world scenarios and restoration quality.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Calibration & Uncertainty Quantification
• 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
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Low-light image enhancement (LLIE) faces persistent challenges in balancing reconstruction fidelity with cross-scenario generalization. While existing methods predominantly focus on deterministic pixel-level mappings between paired low/normal-light images, they often neglect the continuous physical process of luminance transitions in real-world environments, leading to performance drop when normal-light references are unavailable. Inspired by empirical analysis of natural luminance dynamics revealing power-law distributed intensity transitions, this paper introduces Luminance-Aware Statistical Quantification (LASQ), a novel framework that reformulates LLIE as a statistical sampling process over hierarchical luminance distributions. Our LASQ re-conceptualizes luminance transition as a power-law distribution in intensity coordinate space that can be approximated by stratified power functions, therefore, replacing deterministic mappings with probabilistic sampling over continuous luminance layers. A diffusion forward process is designed to autonomously discover optimal transition paths between luminance layers, achieving unsupervised distribution emulation without normal-light references. In this way, it considerably improves the performance in practical situations, enabling more adaptable and versatile light restoration. This framework is also readily applicable to cases with normal-light references, where it achieves superior performance on domain-specific datasets alongside better generalization-ability across non-reference datasets.

Problem

• Reformulating low-light enhancement as statistical sampling over luminance distributions
• Replacing deterministic mappings with probabilistic luminance transition modeling
• Enabling unsupervised illumination enhancement without normal-light references

Innovation

• Statistical sampling over hierarchical luminance distributions
• Power-law distribution modeling for intensity transitions
• Unsupervised diffusion process for luminance path discovery