DMCF-Net: Dilated Multiscale Context Fusion Network for SAR Flood Detection

Synthetic aperture radar (SAR) imagery, with its all-weather, all-time capabilities, plays a critical role in flood detection. However, due to the diverse scattering mechanisms of water bodies, flood regions in SAR images typically exhibit high intraclass variance and low interclass variance. Additi...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhimin Wang, Lingli Zhao, Nan Jiang, Weidong Sun, Jie Yang, Lei Shi, Hongtao Shi, Pingxiang Li
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11059328/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849317823644434432
author Zhimin Wang
Lingli Zhao
Nan Jiang
Weidong Sun
Jie Yang
Lei Shi
Hongtao Shi
Pingxiang Li
author_facet Zhimin Wang
Lingli Zhao
Nan Jiang
Weidong Sun
Jie Yang
Lei Shi
Hongtao Shi
Pingxiang Li
author_sort Zhimin Wang
collection DOAJ
description Synthetic aperture radar (SAR) imagery, with its all-weather, all-time capabilities, plays a critical role in flood detection. However, due to the diverse scattering mechanisms of water bodies, flood regions in SAR images typically exhibit high intraclass variance and low interclass variance. Additionally, the complex shapes and blurred boundaries of flood regions make it challenging for single-scale convolution methods to accurately identify them. To address this issue, we propose a novel deep learning approach, DMCF-Net, to effectively capture the intricate characteristics of flood regions in SAR imagery. DMCF-Net consists of three main modules: multiscale feature aggregation (MSFA) module, cross-scale attention fusion (CSAF) module, and deep feature refinement (DFR) module. MSFA module extracts multiscale features using a dual-branch approach with dilated and depthwise separable convolutions. CSAF module combines contextual information from neighboring scales, using edge details from shallow features and semantic information from deep features. DFR module uses convolutions with varying kernel sizes to refine the deepest features, improving the accuracy of flood detection. The effectiveness of DMCF-Net is assessed on the Sen1Floods11 dataset. Experimental results show that DMCF-Net outperforms other deep learning models, achieving an F1 score of 81.6% and an intersection over union of 68.9%, while also having lower computational cost (97.4G) and fewer parameters (16.4M).
format Article
id doaj-art-882b46677a664c4f9f82fd7720b2a2fe
institution Kabale University
issn 1939-1404
2151-1535
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
spelling doaj-art-882b46677a664c4f9f82fd7720b2a2fe2025-08-20T03:51:04ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-0118165491656110.1109/JSTARS.2025.358428211059328DMCF-Net: Dilated Multiscale Context Fusion Network for SAR Flood DetectionZhimin Wang0Lingli Zhao1https://orcid.org/0000-0001-8838-224XNan Jiang2Weidong Sun3https://orcid.org/0000-0001-8718-1710Jie Yang4Lei Shi5https://orcid.org/0000-0001-7567-5510Hongtao Shi6https://orcid.org/0000-0003-4054-4240Pingxiang Li7School of Remote Sensing and Information Engineering, Wuhan University, Wuhan, ChinaSchool of Remote Sensing and Information Engineering, Wuhan University, Wuhan, ChinaSchool of Remote Sensing and Information Engineering, Wuhan University, Wuhan, ChinaState Key Laboratory of Information Engineering in Surveying, Mapping, and Remote Sensing, Wuhan University, Wuhan, ChinaState Key Laboratory of Information Engineering in Surveying, Mapping, and Remote Sensing, Wuhan University, Wuhan, ChinaState Key Laboratory of Information Engineering in Surveying, Mapping, and Remote Sensing, Wuhan University, Wuhan, ChinaSchool of Environment and Spatial Informatics, China University of Mining and Technology, Xuzhou, ChinaState Key Laboratory of Information Engineering in Surveying, Mapping, and Remote Sensing, Wuhan University, Wuhan, ChinaSynthetic aperture radar (SAR) imagery, with its all-weather, all-time capabilities, plays a critical role in flood detection. However, due to the diverse scattering mechanisms of water bodies, flood regions in SAR images typically exhibit high intraclass variance and low interclass variance. Additionally, the complex shapes and blurred boundaries of flood regions make it challenging for single-scale convolution methods to accurately identify them. To address this issue, we propose a novel deep learning approach, DMCF-Net, to effectively capture the intricate characteristics of flood regions in SAR imagery. DMCF-Net consists of three main modules: multiscale feature aggregation (MSFA) module, cross-scale attention fusion (CSAF) module, and deep feature refinement (DFR) module. MSFA module extracts multiscale features using a dual-branch approach with dilated and depthwise separable convolutions. CSAF module combines contextual information from neighboring scales, using edge details from shallow features and semantic information from deep features. DFR module uses convolutions with varying kernel sizes to refine the deepest features, improving the accuracy of flood detection. The effectiveness of DMCF-Net is assessed on the Sen1Floods11 dataset. Experimental results show that DMCF-Net outperforms other deep learning models, achieving an F1 score of 81.6% and an intersection over union of 68.9%, while also having lower computational cost (97.4G) and fewer parameters (16.4M).https://ieeexplore.ieee.org/document/11059328/Deep learningflood detectionmultiscale featuressen1floods11synthetic aperture radar (SAR)
spellingShingle Zhimin Wang
Lingli Zhao
Nan Jiang
Weidong Sun
Jie Yang
Lei Shi
Hongtao Shi
Pingxiang Li
DMCF-Net: Dilated Multiscale Context Fusion Network for SAR Flood Detection
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Deep learning
flood detection
multiscale features
sen1floods11
synthetic aperture radar (SAR)
title DMCF-Net: Dilated Multiscale Context Fusion Network for SAR Flood Detection
title_full DMCF-Net: Dilated Multiscale Context Fusion Network for SAR Flood Detection
title_fullStr DMCF-Net: Dilated Multiscale Context Fusion Network for SAR Flood Detection
title_full_unstemmed DMCF-Net: Dilated Multiscale Context Fusion Network for SAR Flood Detection
title_short DMCF-Net: Dilated Multiscale Context Fusion Network for SAR Flood Detection
title_sort dmcf net dilated multiscale context fusion network for sar flood detection
topic Deep learning
flood detection
multiscale features
sen1floods11
synthetic aperture radar (SAR)
url https://ieeexplore.ieee.org/document/11059328/
work_keys_str_mv AT zhiminwang dmcfnetdilatedmultiscalecontextfusionnetworkforsarflooddetection
AT linglizhao dmcfnetdilatedmultiscalecontextfusionnetworkforsarflooddetection
AT nanjiang dmcfnetdilatedmultiscalecontextfusionnetworkforsarflooddetection
AT weidongsun dmcfnetdilatedmultiscalecontextfusionnetworkforsarflooddetection
AT jieyang dmcfnetdilatedmultiscalecontextfusionnetworkforsarflooddetection
AT leishi dmcfnetdilatedmultiscalecontextfusionnetworkforsarflooddetection
AT hongtaoshi dmcfnetdilatedmultiscalecontextfusionnetworkforsarflooddetection
AT pingxiangli dmcfnetdilatedmultiscalecontextfusionnetworkforsarflooddetection