A Transformer-Based Diffusion Model for All-in-One Weather-Degraded Image Restoration

Image restoration under adverse weather conditions is of great significance for the subsequent advanced computer vision tasks. However, most existing image restoration algorithms only remove single weather degradation, and few studies has been conducted on all-in-one weather-degraded image restorati...

Full description

Saved in:
Bibliographic Details
Main Author: QIN Jing, WEN Yuanbo, GAO Tao, LIU Yao
Format: Article
Language:zho
Published: Editorial Office of Journal of Shanghai Jiao Tong University 2024-10-01
Series:Shanghai Jiaotong Daxue xuebao
Subjects:
Online Access:https://xuebao.sjtu.edu.cn/article/2024/1006-2467/1006-2467-58-10-1606.shtml
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850284904868216832
author QIN Jing, WEN Yuanbo, GAO Tao, LIU Yao
author_facet QIN Jing, WEN Yuanbo, GAO Tao, LIU Yao
author_sort QIN Jing, WEN Yuanbo, GAO Tao, LIU Yao
collection DOAJ
description Image restoration under adverse weather conditions is of great significance for the subsequent advanced computer vision tasks. However, most existing image restoration algorithms only remove single weather degradation, and few studies has been conducted on all-in-one weather-degraded image restoration. The denoising diffusion probability model is combined with Vision Transformer to propose a Transformer-based diffusion model for all-in-one weather-degraded image restoration. First, the weather-degraded image is utilized as the condition to guide the reverse sampling of diffusion model and generate corresponding clean background image. Then, the subspace transposed Transformer for noise estimation (NE-STT) is proposed, which utilizes the degraded image and the noisy state to estimate noise distribution, including the subspace transposed self-attention (STSA) mechanism and a dual grouped gated feed-forward network (DGGFFN). The STSA adopts subspace transformation coefficient to effectively capture global long-range dependencies while significantly reducing computational burden. The DGGFFN employs the dual grouped gated mechanism to enhance the nonlinear characterization ability of feed-forward network. The experimental results show that in comparison with the recently developed algorithms, such as All-in-One and TransWeather, the method proposed obtains a performance gain of 3.68 and 3.08 dB in average peak signal-to-noise ratio while 2.93% and 3.13% in average structural similarity on 5 weather-degraded datasets.
format Article
id doaj-art-9896f0da2f7a4115afe58cca647f8d67
institution OA Journals
issn 1006-2467
language zho
publishDate 2024-10-01
publisher Editorial Office of Journal of Shanghai Jiao Tong University
record_format Article
series Shanghai Jiaotong Daxue xuebao
spelling doaj-art-9896f0da2f7a4115afe58cca647f8d672025-08-20T01:47:26ZzhoEditorial Office of Journal of Shanghai Jiao Tong UniversityShanghai Jiaotong Daxue xuebao1006-24672024-10-0158101606161710.16183/j.cnki.jsjtu.2023.043A Transformer-Based Diffusion Model for All-in-One Weather-Degraded Image RestorationQIN Jing, WEN Yuanbo, GAO Tao, LIU Yao0School of Information and Engineering, Chang’an University, Xi’an 710064, ChinaImage restoration under adverse weather conditions is of great significance for the subsequent advanced computer vision tasks. However, most existing image restoration algorithms only remove single weather degradation, and few studies has been conducted on all-in-one weather-degraded image restoration. The denoising diffusion probability model is combined with Vision Transformer to propose a Transformer-based diffusion model for all-in-one weather-degraded image restoration. First, the weather-degraded image is utilized as the condition to guide the reverse sampling of diffusion model and generate corresponding clean background image. Then, the subspace transposed Transformer for noise estimation (NE-STT) is proposed, which utilizes the degraded image and the noisy state to estimate noise distribution, including the subspace transposed self-attention (STSA) mechanism and a dual grouped gated feed-forward network (DGGFFN). The STSA adopts subspace transformation coefficient to effectively capture global long-range dependencies while significantly reducing computational burden. The DGGFFN employs the dual grouped gated mechanism to enhance the nonlinear characterization ability of feed-forward network. The experimental results show that in comparison with the recently developed algorithms, such as All-in-One and TransWeather, the method proposed obtains a performance gain of 3.68 and 3.08 dB in average peak signal-to-noise ratio while 2.93% and 3.13% in average structural similarity on 5 weather-degraded datasets.https://xuebao.sjtu.edu.cn/article/2024/1006-2467/1006-2467-58-10-1606.shtmlcomputer visiondiffusion modelimage restorationtransformerweather-degraded image
spellingShingle QIN Jing, WEN Yuanbo, GAO Tao, LIU Yao
A Transformer-Based Diffusion Model for All-in-One Weather-Degraded Image Restoration
Shanghai Jiaotong Daxue xuebao
computer vision
diffusion model
image restoration
transformer
weather-degraded image
title A Transformer-Based Diffusion Model for All-in-One Weather-Degraded Image Restoration
title_full A Transformer-Based Diffusion Model for All-in-One Weather-Degraded Image Restoration
title_fullStr A Transformer-Based Diffusion Model for All-in-One Weather-Degraded Image Restoration
title_full_unstemmed A Transformer-Based Diffusion Model for All-in-One Weather-Degraded Image Restoration
title_short A Transformer-Based Diffusion Model for All-in-One Weather-Degraded Image Restoration
title_sort transformer based diffusion model for all in one weather degraded image restoration
topic computer vision
diffusion model
image restoration
transformer
weather-degraded image
url https://xuebao.sjtu.edu.cn/article/2024/1006-2467/1006-2467-58-10-1606.shtml
work_keys_str_mv AT qinjingwenyuanbogaotaoliuyao atransformerbaseddiffusionmodelforallinoneweatherdegradedimagerestoration
AT qinjingwenyuanbogaotaoliuyao transformerbaseddiffusionmodelforallinoneweatherdegradedimagerestoration