Image Motion Blur Removal in the Temporal Dimension with Video Diffusion Models

TL;DR

This paper introduces a novel single-image deblurring method that leverages pre-trained video diffusion models to effectively handle complex motion blur, outperforming existing techniques without explicit kernel estimation.

Contribution

It presents a new approach using video diffusion transformers to model motion blur as a temporal averaging process, improving deblurring performance over traditional spatial methods.

Findings

Outperforms existing deblurring methods on synthetic datasets

Effective in real-world complex motion blur scenarios

Utilizes a diffusion-based inverse problem framework

Abstract

Most motion deblurring algorithms rely on spatial-domain convolution models, which struggle with the complex, non-linear blur arising from camera shake and object motion. In contrast, we propose a novel single-image deblurring approach that treats motion blur as a temporal averaging phenomenon. Our core innovation lies in leveraging a pre-trained video diffusion transformer model to capture diverse motion dynamics within a latent space. It sidesteps explicit kernel estimation and effectively accommodates diverse motion patterns. We implement the algorithm within a diffusion-based inverse problem framework. Empirical results on synthetic and real-world datasets demonstrate that our method outperforms existing techniques in deblurring complex motion blur scenarios. This work paves the way for utilizing powerful video diffusion models to address single-image deblurring challenges.