Recursive Multi-Scale Network With Enhanced Attention for Structure-Preserving Deraining

Rain streaks in outdoor images can severely affect the performance of high-level vision tasks, such as surveillance and autonomous driving, by obscuring important image details. The objective of this research is to develop a more effective deraining method that preserves the structural details of im...

Full description

Saved in:
Bibliographic Details
Main Authors: Shuang Jiao, Jianping Zhao, Hua Li, Yun Yang, Ning Gu
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10771756/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846110560835338240
author Shuang Jiao
Jianping Zhao
Hua Li
Yun Yang
Ning Gu
author_facet Shuang Jiao
Jianping Zhao
Hua Li
Yun Yang
Ning Gu
author_sort Shuang Jiao
collection DOAJ
description Rain streaks in outdoor images can severely affect the performance of high-level vision tasks, such as surveillance and autonomous driving, by obscuring important image details. The objective of this research is to develop a more effective deraining method that preserves the structural details of images while removing rain streaks. To address the limitations of existing methods, we introduce the Recursive Multi-Scale Network with Enhanced Attention for Structure-Preserving Deraining (RMASP-Net). This network integrates multi-scale feature processing, recursive architecture, attention mechanisms, and Rectified Local Contrast Normalization (RLCN). It utilizes wavelet transform’s scale-space capabilities to detect key features across different scales, improving the alignment of structural features and isolating rain pixels effectively. Additionally, RMASP-Net features a Mask-Guided Attention Module (Mask-GAM) for dynamic feature weighting and an Interactive Fusion Module (IFM) to enhance image clarity and structure. Extensive experiments on both synthetic and real-world datasets demonstrate that RMASP-Net significantly outperforms existing methods. Specifically, RMASP-Net shows noticeable improvements in both PSNR and SSIM across datasets like Rain100H, Rain100L, and Test1200, providing clearer and more structurally accurate derained images. This research sets a new benchmark for deraining performance under real-world conditions.
format Article
id doaj-art-b49ee9570ed24f5d93790ac83098ae7b
institution Kabale University
issn 2169-3536
language English
publishDate 2024-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-b49ee9570ed24f5d93790ac83098ae7b2024-12-24T00:01:14ZengIEEEIEEE Access2169-35362024-01-011218171818172710.1109/ACCESS.2024.350880510771756Recursive Multi-Scale Network With Enhanced Attention for Structure-Preserving DerainingShuang Jiao0https://orcid.org/0009-0009-9656-222XJianping Zhao1https://orcid.org/0000-0003-0824-538XHua Li2https://orcid.org/0000-0002-8566-1558Yun Yang3Ning Gu4School of Computer Science and Technology, Changchun University of Science and Technology, Chaoyang, Changchun, Jilin, ChinaSchool of Computer Science and Technology, Changchun University of Science and Technology, Chaoyang, Changchun, Jilin, ChinaSchool of Computer Science and Technology, Changchun University of Science and Technology, Chaoyang, Changchun, Jilin, ChinaChangchun Institute of Education, Luyuan, Changchun, Jilin, ChinaChangchun Institute of Education, Luyuan, Changchun, Jilin, ChinaRain streaks in outdoor images can severely affect the performance of high-level vision tasks, such as surveillance and autonomous driving, by obscuring important image details. The objective of this research is to develop a more effective deraining method that preserves the structural details of images while removing rain streaks. To address the limitations of existing methods, we introduce the Recursive Multi-Scale Network with Enhanced Attention for Structure-Preserving Deraining (RMASP-Net). This network integrates multi-scale feature processing, recursive architecture, attention mechanisms, and Rectified Local Contrast Normalization (RLCN). It utilizes wavelet transform’s scale-space capabilities to detect key features across different scales, improving the alignment of structural features and isolating rain pixels effectively. Additionally, RMASP-Net features a Mask-Guided Attention Module (Mask-GAM) for dynamic feature weighting and an Interactive Fusion Module (IFM) to enhance image clarity and structure. Extensive experiments on both synthetic and real-world datasets demonstrate that RMASP-Net significantly outperforms existing methods. Specifically, RMASP-Net shows noticeable improvements in both PSNR and SSIM across datasets like Rain100H, Rain100L, and Test1200, providing clearer and more structurally accurate derained images. This research sets a new benchmark for deraining performance under real-world conditions.https://ieeexplore.ieee.org/document/10771756/Recursivemulti-scalestructure-preservingderainingmulti-scale feature processingimage structure preservation
spellingShingle Shuang Jiao
Jianping Zhao
Hua Li
Yun Yang
Ning Gu
Recursive Multi-Scale Network With Enhanced Attention for Structure-Preserving Deraining
IEEE Access
Recursive
multi-scale
structure-preserving
deraining
multi-scale feature processing
image structure preservation
title Recursive Multi-Scale Network With Enhanced Attention for Structure-Preserving Deraining
title_full Recursive Multi-Scale Network With Enhanced Attention for Structure-Preserving Deraining
title_fullStr Recursive Multi-Scale Network With Enhanced Attention for Structure-Preserving Deraining
title_full_unstemmed Recursive Multi-Scale Network With Enhanced Attention for Structure-Preserving Deraining
title_short Recursive Multi-Scale Network With Enhanced Attention for Structure-Preserving Deraining
title_sort recursive multi scale network with enhanced attention for structure preserving deraining
topic Recursive
multi-scale
structure-preserving
deraining
multi-scale feature processing
image structure preservation
url https://ieeexplore.ieee.org/document/10771756/
work_keys_str_mv AT shuangjiao recursivemultiscalenetworkwithenhancedattentionforstructurepreservingderaining
AT jianpingzhao recursivemultiscalenetworkwithenhancedattentionforstructurepreservingderaining
AT huali recursivemultiscalenetworkwithenhancedattentionforstructurepreservingderaining
AT yunyang recursivemultiscalenetworkwithenhancedattentionforstructurepreservingderaining
AT ninggu recursivemultiscalenetworkwithenhancedattentionforstructurepreservingderaining