BluRef: Unsupervised Image Deblurring with Dense-Matching References

2026-03-14

AI Summary

This work addresses the reliance of image deblurring methods on paired blurry-sharp training data by proposing an unsupervised approach that requires only unpaired, scene-similar blurry and sharp images. The method leverages a dense matching model to establish pixel-level correspondences between a blurry input and a reference sharp image, generating high-quality pseudo-labels for self-supervised training of a deblurring network—without any paired data or pre-trained models. Experimental results demonstrate that the proposed approach achieves state-of-the-art performance across multiple benchmarks, significantly advancing unsupervised image deblurring. Furthermore, the framework exhibits strong scalability and holds promise for deployment in low-resource settings.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Multi-modal Vision
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

This paper introduces a novel unsupervised approach for image deblurring that utilizes a simple process for training data collection, thereby enhancing the applicability and effectiveness of deblurring methods. Our technique does not require meticulously paired data of blurred and corresponding sharp images; instead, it uses unpaired blurred and sharp images of similar scenes to generate pseudo-ground truth data by leveraging a dense matching model to identify correspondences between a blurry image and reference sharp images. Thanks to the simplicity of the training data collection process, our approach does not rely on existing paired training data or pre-trained networks, making it more adaptable to various scenarios and suitable for networks of different sizes, including those designed for low-resource devices. We demonstrate that this novel approach achieves state-of-the-art performance, marking a significant advancement in the field of image deblurring.

Problem

• image deblurring
• unsupervised learning
• unpaired data
• dense matching
• pseudo-ground truth

Innovation

• unsupervised image deblurring
• dense matching
• unpaired training data
• pseudo-ground truth
• reference-based deblurring