ZeroIDIR: Zero-Reference Illumination Degradation Image Restoration with Perturbed Consistency Diffusion Models

2026-05-11

AI Summary

This work addresses the challenge of restoring illumination-degraded images under no-reference conditions by proposing a zero-reference diffusion framework that enables unsupervised training without requiring high-quality ground-truth images. The method decouples restoration into two stages: adaptive illumination correction and diffusion-based detail reconstruction. It innovatively integrates spatially varying gamma correction with a perturbation-consistent diffusion mechanism and introduces a histogram-guided illumination correction loss alongside a perturbation-diffusion consistency loss to enhance restoration fidelity and stability. Extensive experiments demonstrate that the proposed approach significantly outperforms existing unsupervised methods on multiple public benchmarks, achieving performance comparable to supervised counterparts while exhibiting strong generalization across diverse real-world scenes.

Abstract

In this paper, we propose a zero-reference diffusion-based framework, named ZeroIDIR, for illumination degradation image restoration, which decouples the restoration process into adaptive illumination correction and diffusion-based reconstruction while being trained solely on low-quality degraded images. Specifically, we design an adaptive gamma correction module that performs spatially varying exposure correction to generate illumination-corrected only representations to mitigate exposure bias and serve as reliable inputs for subsequent diffusion processes, where a histogram-guided illumination correction loss is introduced to regularize the corrected illumination distribution toward that of natural scenes. Subsequently, the illumination-corrected image is treated as an intermediate noisy state for the proposed perturbed consistency diffusion model to reconstruct details and suppress noise. Moreover, a perturbed diffusion consistency loss is proposed to constrain the forward diffusion trajectory of the final restored image to remain consistent with the perturbed state, thus improving restoration fidelity and stability in the absence of supervision. Extensive experiments on publicly available benchmarks show that the proposed method outperforms state-of-the-art unsupervised competitors and is comparable to supervised methods while being more generalizable to various scenes. Code is available at https://github.com/JianghaiSCU/ZeroIDIR.

Problem

• illumination degradation
• image restoration
• zero-reference
• unsupervised learning
• diffusion models

Innovation

• zero-reference
• diffusion model
• illumination correction
• perturbed consistency
• unsupervised image restoration