Equivariant Sampling for Improving Diffusion Model-based Image Restoration

2025-11-13

AI Summary

Existing diffusion-based image restoration (DMIR) methods struggle to fully exploit diffusion priors, limiting their restoration performance. To address this, we propose EquS, a general-purpose framework that introduces equivariant structural priors via dual sampling trajectories and designs a temporal-aware scheduling (TAS) strategy to enhance prior utilization efficiency—without increasing computational overhead. EquS and TAS are plug-and-play modules compatible with mainstream DMIR approaches. Extensive experiments across multiple benchmark datasets demonstrate substantial improvements in denoising and super-resolution tasks, achieving state-of-the-art PSNR and SSIM scores. The results validate the effectiveness of jointly optimizing equivariant modeling and temporal scheduling for diffusion-based restoration.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Planning, Routing, and Scheduling: Temporal Planning
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Recent advances in generative models, especially diffusion models, have significantly improved image restoration (IR) performance. However, existing problem-agnostic diffusion model-based image restoration (DMIR) methods face challenges in fully leveraging diffusion priors, resulting in suboptimal performance. In this paper, we address the limitations of current problem-agnostic DMIR methods by analyzing their sampling process and providing effective solutions. We introduce EquS, a DMIR method that imposes equivariant information through dual sampling trajectories. To further boost EquS, we propose the Timestep-Aware Schedule (TAS) and introduce EquS$^+$. TAS prioritizes deterministic steps to enhance certainty and sampling efficiency. Extensive experiments on benchmarks demonstrate that our method is compatible with previous problem-agnostic DMIR methods and significantly boosts their performance without increasing computational costs. Our code is available at https://github.com/FouierL/EquS.

Problem

• Improving diffusion model-based image restoration performance
• Addressing limitations in leveraging diffusion priors effectively
• Enhancing sampling efficiency without increasing computational costs

Innovation

• Equivariant sampling with dual trajectories
• Timestep-Aware Schedule for efficient sampling
• Maintains computational efficiency while boosting performance