Aggregating Nearest Sharp Features via Hybrid Transformers for Video Deblurring

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Shang, Wei, Ren, Dongwei, Yang, Yi, Zuo, Wangmeng
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909408180568064
author Shang, Wei
Ren, Dongwei
Yang, Yi
Zuo, Wangmeng
author_facet Shang, Wei
Ren, Dongwei
Yang, Yi
Zuo, Wangmeng
contents Video deblurring methods, aiming at recovering consecutive sharp frames from a given blurry video, usually assume that the input video suffers from consecutively blurry frames. However, in real-world scenarios captured by modern imaging devices, sharp frames often interspersed within the video, providing temporally nearest sharp features that can aid in the restoration of blurry frames. In this work, we propose a video deblurring method that leverages both neighboring frames and existing sharp frames using hybrid Transformers for feature aggregation. Specifically, we first train a blur-aware detector to distinguish between sharp and blurry frames. Then, a window-based local Transformer is employed for exploiting features from neighboring frames, where cross attention is beneficial for aggregating features from neighboring frames without explicit spatial alignment. To aggregate nearest sharp features from detected sharp frames, we utilize a global Transformer with multi-scale matching capability. Moreover, our method can easily be extended to event-driven video deblurring by incorporating an event fusion module into the global Transformer. Extensive experiments on benchmark datasets demonstrate that our proposed method outperforms state-of-the-art video deblurring methods as well as event-driven video deblurring methods in terms of quantitative metrics and visual quality. The source code and trained models are available at https://github.com/shangwei5/STGTN.
format Preprint
id arxiv_https___arxiv_org_abs_2309_07054
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Aggregating Nearest Sharp Features via Hybrid Transformers for Video Deblurring
Shang, Wei
Ren, Dongwei
Yang, Yi
Zuo, Wangmeng
Computer Vision and Pattern Recognition
I.4.3
Video deblurring methods, aiming at recovering consecutive sharp frames from a given blurry video, usually assume that the input video suffers from consecutively blurry frames. However, in real-world scenarios captured by modern imaging devices, sharp frames often interspersed within the video, providing temporally nearest sharp features that can aid in the restoration of blurry frames. In this work, we propose a video deblurring method that leverages both neighboring frames and existing sharp frames using hybrid Transformers for feature aggregation. Specifically, we first train a blur-aware detector to distinguish between sharp and blurry frames. Then, a window-based local Transformer is employed for exploiting features from neighboring frames, where cross attention is beneficial for aggregating features from neighboring frames without explicit spatial alignment. To aggregate nearest sharp features from detected sharp frames, we utilize a global Transformer with multi-scale matching capability. Moreover, our method can easily be extended to event-driven video deblurring by incorporating an event fusion module into the global Transformer. Extensive experiments on benchmark datasets demonstrate that our proposed method outperforms state-of-the-art video deblurring methods as well as event-driven video deblurring methods in terms of quantitative metrics and visual quality. The source code and trained models are available at https://github.com/shangwei5/STGTN.
title Aggregating Nearest Sharp Features via Hybrid Transformers for Video Deblurring
topic Computer Vision and Pattern Recognition
I.4.3
url https://arxiv.org/abs/2309.07054