CIMF-Net: A Change Indicator-Enhanced Multiscale Fusion Network for Remote Sensing Change Detection

Remote sensing change detection (RSCD) is essential for monitoring land use, urban expansion, and environmental changes. Despite advancements in deep learning, existing methods exhibit limitations such as discontinuous edges and internal gaps in detected change regions, primarily due to insufficient...

Full description

Saved in:
Bibliographic Details
Main Authors: Jiting Zhou, Pu Zhang, Zhihao Zhou
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10964253/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850179780752703488
author Jiting Zhou
Pu Zhang
Zhihao Zhou
author_facet Jiting Zhou
Pu Zhang
Zhihao Zhou
author_sort Jiting Zhou
collection DOAJ
description Remote sensing change detection (RSCD) is essential for monitoring land use, urban expansion, and environmental changes. Despite advancements in deep learning, existing methods exhibit limitations such as discontinuous edges and internal gaps in detected change regions, primarily due to insufficient feature fusion and limited contextual understanding. In this paper, we propose a novel change indicator-enhanced multiscale fusion network, called CIMF-Net, which integrates multiscale feature fusion modules (MFFM) and cascade guided attention modules (CGAM) to overcome these limitations. MFFM facilitates the extraction of semantic information through multiscale features, while CGAM employs hierarchical attention mechanisms to refine feature representation and enhance change region localization. Extensive experiments on three benchmark datasets highlight the superior performance of CIMF-Net compared to baseline network, achieving higher F1-scores, IoU, and overall accuracy. Furthermore, ablation studies confirm the effectiveness of MFFM and CGAM in enhancing feature fusion and hierarchical guidance. Our method not only delivers robust results in challenging scenarios but also presents a new perspective and technology for change detection.
format Article
id doaj-art-a22797fc95b44b26a71185d03b7b058b
institution OA Journals
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-a22797fc95b44b26a71185d03b7b058b2025-08-20T02:18:24ZengIEEEIEEE Access2169-35362025-01-0113668436685410.1109/ACCESS.2025.356059110964253CIMF-Net: A Change Indicator-Enhanced Multiscale Fusion Network for Remote Sensing Change DetectionJiting Zhou0https://orcid.org/0000-0001-7534-9993Pu Zhang1https://orcid.org/0009-0004-2429-4637Zhihao Zhou2https://orcid.org/0009-0008-9124-4364Shanghai University, Shanghai, ChinaShanghai University, Shanghai, ChinaShanghai University, Shanghai, ChinaRemote sensing change detection (RSCD) is essential for monitoring land use, urban expansion, and environmental changes. Despite advancements in deep learning, existing methods exhibit limitations such as discontinuous edges and internal gaps in detected change regions, primarily due to insufficient feature fusion and limited contextual understanding. In this paper, we propose a novel change indicator-enhanced multiscale fusion network, called CIMF-Net, which integrates multiscale feature fusion modules (MFFM) and cascade guided attention modules (CGAM) to overcome these limitations. MFFM facilitates the extraction of semantic information through multiscale features, while CGAM employs hierarchical attention mechanisms to refine feature representation and enhance change region localization. Extensive experiments on three benchmark datasets highlight the superior performance of CIMF-Net compared to baseline network, achieving higher F1-scores, IoU, and overall accuracy. Furthermore, ablation studies confirm the effectiveness of MFFM and CGAM in enhancing feature fusion and hierarchical guidance. Our method not only delivers robust results in challenging scenarios but also presents a new perspective and technology for change detection.https://ieeexplore.ieee.org/document/10964253/Remote sensing change detectionfeature fusionattention mechanismdeep learning
spellingShingle Jiting Zhou
Pu Zhang
Zhihao Zhou
CIMF-Net: A Change Indicator-Enhanced Multiscale Fusion Network for Remote Sensing Change Detection
IEEE Access
Remote sensing change detection
feature fusion
attention mechanism
deep learning
title CIMF-Net: A Change Indicator-Enhanced Multiscale Fusion Network for Remote Sensing Change Detection
title_full CIMF-Net: A Change Indicator-Enhanced Multiscale Fusion Network for Remote Sensing Change Detection
title_fullStr CIMF-Net: A Change Indicator-Enhanced Multiscale Fusion Network for Remote Sensing Change Detection
title_full_unstemmed CIMF-Net: A Change Indicator-Enhanced Multiscale Fusion Network for Remote Sensing Change Detection
title_short CIMF-Net: A Change Indicator-Enhanced Multiscale Fusion Network for Remote Sensing Change Detection
title_sort cimf net a change indicator enhanced multiscale fusion network for remote sensing change detection
topic Remote sensing change detection
feature fusion
attention mechanism
deep learning
url https://ieeexplore.ieee.org/document/10964253/
work_keys_str_mv AT jitingzhou cimfnetachangeindicatorenhancedmultiscalefusionnetworkforremotesensingchangedetection
AT puzhang cimfnetachangeindicatorenhancedmultiscalefusionnetworkforremotesensingchangedetection
AT zhihaozhou cimfnetachangeindicatorenhancedmultiscalefusionnetworkforremotesensingchangedetection