A Spatiotemporal Fusion Network for Remote Sensing Based on Global Context Attention Mechanism

Spatial-temporal fusion algorithms commonly encounter difficulties in effectively striking a balance between the extraction of intricate spatial details and changes over time. To mitigate these problems, we propose a spatiotemporal fusion network for remote sensing based on a global context (GC) att...

Full description

Saved in:
Bibliographic Details
Main Authors: Weisheng Li, Yusha Liu, Yidong Peng, Fengyan Wu
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/10766929/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850164947240091648
author Weisheng Li
Yusha Liu
Yidong Peng
Fengyan Wu
author_facet Weisheng Li
Yusha Liu
Yidong Peng
Fengyan Wu
author_sort Weisheng Li
collection DOAJ
description Spatial-temporal fusion algorithms commonly encounter difficulties in effectively striking a balance between the extraction of intricate spatial details and changes over time. To mitigate these problems, we propose a spatiotemporal fusion network for remote sensing based on a global context (GC) attention mechanism. This network comprises an extracting feature network and a difference network. In the difference network, we introduce a GC attention mechanism that focuses on crucial details across various features, assuming different roles in different modules. A GC attention-based U-block (GCAU) employs a U-structured design and integrates GC attention mechanisms into every layer of its architecture. This enables the module to effectively process regions with pronounced spatial heterogeneity and to adeptly capture spatial differential information. A GC layer (GCL) block comprises five interconnected GC attention blocks interspersed with local residuals. During the training process, these residuals assist in rectifying missing data and bolstering feature transfer, thereby enabling the module to more effectively capture temporally dynamic changes in rapidly evolving areas. This allows the module to better capture time-based changes in fast-changing areas. Supervision of both the final output and intermediate difference images is facilitated by a composite loss function, which improves the fusion quality in the temporal, spatial, and visual domains. The model's robustness and superiority are validated through experimental testing on three datasets, accompanied by subjective and objective evaluations, as well as ablation experiments.
format Article
id doaj-art-c4d68aed897f4aefb59865dab17f0bd2
institution OA Journals
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-c4d68aed897f4aefb59865dab17f0bd22025-08-20T02:21:51ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-01181451147110.1109/JSTARS.2024.350587410766929A Spatiotemporal Fusion Network for Remote Sensing Based on Global Context Attention MechanismWeisheng Li0https://orcid.org/0000-0002-9033-8245Yusha Liu1https://orcid.org/0009-0007-1894-3989Yidong Peng2https://orcid.org/0000-0003-3779-0360Fengyan Wu3Chongqing Key Laboratory of Image Cognition, Chongqing University of Posts and Telecommunications, Chongqing, ChinaChongqing Key Laboratory of Image Cognition, Chongqing University of Posts and Telecommunications, Chongqing, ChinaChongqing Key Laboratory of Image Cognition, Chongqing University of Posts and Telecommunications, Chongqing, ChinaChongqing Key Laboratory of Image Cognition, Chongqing University of Posts and Telecommunications, Chongqing, ChinaSpatial-temporal fusion algorithms commonly encounter difficulties in effectively striking a balance between the extraction of intricate spatial details and changes over time. To mitigate these problems, we propose a spatiotemporal fusion network for remote sensing based on a global context (GC) attention mechanism. This network comprises an extracting feature network and a difference network. In the difference network, we introduce a GC attention mechanism that focuses on crucial details across various features, assuming different roles in different modules. A GC attention-based U-block (GCAU) employs a U-structured design and integrates GC attention mechanisms into every layer of its architecture. This enables the module to effectively process regions with pronounced spatial heterogeneity and to adeptly capture spatial differential information. A GC layer (GCL) block comprises five interconnected GC attention blocks interspersed with local residuals. During the training process, these residuals assist in rectifying missing data and bolstering feature transfer, thereby enabling the module to more effectively capture temporally dynamic changes in rapidly evolving areas. This allows the module to better capture time-based changes in fast-changing areas. Supervision of both the final output and intermediate difference images is facilitated by a composite loss function, which improves the fusion quality in the temporal, spatial, and visual domains. The model's robustness and superiority are validated through experimental testing on three datasets, accompanied by subjective and objective evaluations, as well as ablation experiments.https://ieeexplore.ieee.org/document/10766929/Global context (GC)global context attention-based U-block (GCAU)global context layer (GCL)remote sensingspatiotemporal fusion
spellingShingle Weisheng Li
Yusha Liu
Yidong Peng
Fengyan Wu
A Spatiotemporal Fusion Network for Remote Sensing Based on Global Context Attention Mechanism
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Global context (GC)
global context attention-based U-block (GCAU)
global context layer (GCL)
remote sensing
spatiotemporal fusion
title A Spatiotemporal Fusion Network for Remote Sensing Based on Global Context Attention Mechanism
title_full A Spatiotemporal Fusion Network for Remote Sensing Based on Global Context Attention Mechanism
title_fullStr A Spatiotemporal Fusion Network for Remote Sensing Based on Global Context Attention Mechanism
title_full_unstemmed A Spatiotemporal Fusion Network for Remote Sensing Based on Global Context Attention Mechanism
title_short A Spatiotemporal Fusion Network for Remote Sensing Based on Global Context Attention Mechanism
title_sort spatiotemporal fusion network for remote sensing based on global context attention mechanism
topic Global context (GC)
global context attention-based U-block (GCAU)
global context layer (GCL)
remote sensing
spatiotemporal fusion
url https://ieeexplore.ieee.org/document/10766929/
work_keys_str_mv AT weishengli aspatiotemporalfusionnetworkforremotesensingbasedonglobalcontextattentionmechanism
AT yushaliu aspatiotemporalfusionnetworkforremotesensingbasedonglobalcontextattentionmechanism
AT yidongpeng aspatiotemporalfusionnetworkforremotesensingbasedonglobalcontextattentionmechanism
AT fengyanwu aspatiotemporalfusionnetworkforremotesensingbasedonglobalcontextattentionmechanism
AT weishengli spatiotemporalfusionnetworkforremotesensingbasedonglobalcontextattentionmechanism
AT yushaliu spatiotemporalfusionnetworkforremotesensingbasedonglobalcontextattentionmechanism
AT yidongpeng spatiotemporalfusionnetworkforremotesensingbasedonglobalcontextattentionmechanism
AT fengyanwu spatiotemporalfusionnetworkforremotesensingbasedonglobalcontextattentionmechanism