Non-Homogeneous Image Dehazing Model Based on Vision Transformer Shunt Self-Attention Aggregation

This study introduces an innovative dehazing technique utilizing a Vision Transformer to mitigate the image quality degradation caused by non-homogeneous haze in real-world environments. Initially, the model employs convolutional layers to augment the channel dimensions of the input image, enabling...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhigang Zhang, Byung-Won Min, Zijiao Zhang
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11028993/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849421807244804096
author Zhigang Zhang
Byung-Won Min
Zijiao Zhang
author_facet Zhigang Zhang
Byung-Won Min
Zijiao Zhang
author_sort Zhigang Zhang
collection DOAJ
description This study introduces an innovative dehazing technique utilizing a Vision Transformer to mitigate the image quality degradation caused by non-homogeneous haze in real-world environments. Initially, the model employs convolutional layers to augment the channel dimensions of the input image, enabling the extraction of basic local features. Subsequently, we incorporate a Multi-Scale Channel-Pixel Joint Attention Module, which refines the attention on various haze-affected areas, ensuring accurate capture of the complex characteristics associated with non-homogeneous haze. The cornerstone of our approach is the proposed improved Vision Transformer, which integrates a Shunt Self-Attention Aggregation Module. This module, leveraging a multi-head parallel self-attention mechanism, facilitates the efficient fusion of features across multiple scales, thereby enhancing feature reuse and integration. We conducted extensive experimental evaluations using both real-world and synthetic non-homogeneous haze datasets, rigorously validating the robustness and effectiveness of our model. The results demonstrate that our approach surpasses existing benchmark techniques across various evaluation metrics, affirming its preeminence in the field of image dehazing.
format Article
id doaj-art-5cbcb939e4074353ab633194f4cdaaaa
institution Kabale University
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-5cbcb939e4074353ab633194f4cdaaaa2025-08-20T03:31:21ZengIEEEIEEE Access2169-35362025-01-011310114310115910.1109/ACCESS.2025.357819511028993Non-Homogeneous Image Dehazing Model Based on Vision Transformer Shunt Self-Attention AggregationZhigang Zhang0https://orcid.org/0000-0002-4929-2169Byung-Won Min1Zijiao Zhang2College of Artificial Intelligence, Jiaozuo University, Jiaozuo, ChinaDepartment of IT Engineering, Mokwon University, Daejeon, South KoreaSchool of Telecommunications and Intelligent Manufacturing, Sias University, Zhengzhou, ChinaThis study introduces an innovative dehazing technique utilizing a Vision Transformer to mitigate the image quality degradation caused by non-homogeneous haze in real-world environments. Initially, the model employs convolutional layers to augment the channel dimensions of the input image, enabling the extraction of basic local features. Subsequently, we incorporate a Multi-Scale Channel-Pixel Joint Attention Module, which refines the attention on various haze-affected areas, ensuring accurate capture of the complex characteristics associated with non-homogeneous haze. The cornerstone of our approach is the proposed improved Vision Transformer, which integrates a Shunt Self-Attention Aggregation Module. This module, leveraging a multi-head parallel self-attention mechanism, facilitates the efficient fusion of features across multiple scales, thereby enhancing feature reuse and integration. We conducted extensive experimental evaluations using both real-world and synthetic non-homogeneous haze datasets, rigorously validating the robustness and effectiveness of our model. The results demonstrate that our approach surpasses existing benchmark techniques across various evaluation metrics, affirming its preeminence in the field of image dehazing.https://ieeexplore.ieee.org/document/11028993/Features reusenon-homogeneous dehazingmulti-scale channel-pixel joint attentionvision transformer shunt self-attention aggregation
spellingShingle Zhigang Zhang
Byung-Won Min
Zijiao Zhang
Non-Homogeneous Image Dehazing Model Based on Vision Transformer Shunt Self-Attention Aggregation
IEEE Access
Features reuse
non-homogeneous dehazing
multi-scale channel-pixel joint attention
vision transformer shunt self-attention aggregation
title Non-Homogeneous Image Dehazing Model Based on Vision Transformer Shunt Self-Attention Aggregation
title_full Non-Homogeneous Image Dehazing Model Based on Vision Transformer Shunt Self-Attention Aggregation
title_fullStr Non-Homogeneous Image Dehazing Model Based on Vision Transformer Shunt Self-Attention Aggregation
title_full_unstemmed Non-Homogeneous Image Dehazing Model Based on Vision Transformer Shunt Self-Attention Aggregation
title_short Non-Homogeneous Image Dehazing Model Based on Vision Transformer Shunt Self-Attention Aggregation
title_sort non homogeneous image dehazing model based on vision transformer shunt self attention aggregation
topic Features reuse
non-homogeneous dehazing
multi-scale channel-pixel joint attention
vision transformer shunt self-attention aggregation
url https://ieeexplore.ieee.org/document/11028993/
work_keys_str_mv AT zhigangzhang nonhomogeneousimagedehazingmodelbasedonvisiontransformershuntselfattentionaggregation
AT byungwonmin nonhomogeneousimagedehazingmodelbasedonvisiontransformershuntselfattentionaggregation
AT zijiaozhang nonhomogeneousimagedehazingmodelbasedonvisiontransformershuntselfattentionaggregation