SERNet: Spatially Enhanced Recalibration Network for Building Extraction in Dense Remote Sensing Scenes

The rapid development of urban and rural construction has accelerated the demand for segmentation in dense building scenes. However, the issue of inaccurate building localization in such scenes still lacks effective solutions. One of the causes of this problem is the loss of high-frequency informati...

Full description

Saved in:
Bibliographic Details
Main Authors: Kuikui Han, Yuanwei Yang, Xianjun Gao, Dongjie Yang, Lei Xu
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/11002701/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850131625558409216
author Kuikui Han
Yuanwei Yang
Xianjun Gao
Dongjie Yang
Lei Xu
author_facet Kuikui Han
Yuanwei Yang
Xianjun Gao
Dongjie Yang
Lei Xu
author_sort Kuikui Han
collection DOAJ
description The rapid development of urban and rural construction has accelerated the demand for segmentation in dense building scenes. However, the issue of inaccurate building localization in such scenes still lacks effective solutions. One of the causes of this problem is the loss of high-frequency information and spatial misalignment caused by repeated sampling. To address this, this article proposes the spatial enhancement and recalibration network (SERNet). SERNet divides the feature extraction process into three stages: spatial retention, enhancement, and recalibration. In the first stage, the designed parallel path feature extraction architecture is used to acquire deep semantic features by the spatial path to retain spatial information and contextual path to acquire deep semantic features. In the second stage, a spatial reinforcement module based on perceptual kernel is proposed. This module predicts the grouping kernel using spatial path features, obtains the local perceptual kernel by intrakernel similarity computation, and uses the kernel weights to strengthen the local detail information after grouping. In the third phase, the group bootstrap space calibration module was designed. Set the grouping according to the difference in scale variation. Provide guidance information for bias prediction by calculating the difference degree of direction and distance between features, and finally realize high-resolution reconstruction of building features through level-by-level calibration. Tested on three building datasets, Massachusetts, WHU Aerial, and WHU Satellite I, the IoU of this article's method reaches 75.42%, 91.42%, and 67.24%, respectively.
format Article
id doaj-art-8eddc74d132548e7be00947238d89ebf
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-8eddc74d132548e7be00947238d89ebf2025-08-20T02:32:22ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-0118134231343710.1109/JSTARS.2025.356965611002701SERNet: Spatially Enhanced Recalibration Network for Building Extraction in Dense Remote Sensing ScenesKuikui Han0https://orcid.org/0009-0003-2173-0686Yuanwei Yang1https://orcid.org/0000-0002-4221-4563Xianjun Gao2https://orcid.org/0000-0003-1144-8479Dongjie Yang3https://orcid.org/0000-0001-7815-3523Lei Xu4School of Geosciences, Yangtze University, Wuhan, ChinaSchool of Geosciences, Yangtze University, Wuhan, ChinaSchool of Geosciences, Yangtze University, Wuhan, ChinaSchool of Geosciences, Yangtze University, Wuhan, ChinaChina Railway Design Corporation, Tianjin, ChinaThe rapid development of urban and rural construction has accelerated the demand for segmentation in dense building scenes. However, the issue of inaccurate building localization in such scenes still lacks effective solutions. One of the causes of this problem is the loss of high-frequency information and spatial misalignment caused by repeated sampling. To address this, this article proposes the spatial enhancement and recalibration network (SERNet). SERNet divides the feature extraction process into three stages: spatial retention, enhancement, and recalibration. In the first stage, the designed parallel path feature extraction architecture is used to acquire deep semantic features by the spatial path to retain spatial information and contextual path to acquire deep semantic features. In the second stage, a spatial reinforcement module based on perceptual kernel is proposed. This module predicts the grouping kernel using spatial path features, obtains the local perceptual kernel by intrakernel similarity computation, and uses the kernel weights to strengthen the local detail information after grouping. In the third phase, the group bootstrap space calibration module was designed. Set the grouping according to the difference in scale variation. Provide guidance information for bias prediction by calculating the difference degree of direction and distance between features, and finally realize high-resolution reconstruction of building features through level-by-level calibration. Tested on three building datasets, Massachusetts, WHU Aerial, and WHU Satellite I, the IoU of this article's method reaches 75.42%, 91.42%, and 67.24%, respectively.https://ieeexplore.ieee.org/document/11002701/Building extractionfeature aggregationhigh-resolution remote sensing imageryspatial alignment
spellingShingle Kuikui Han
Yuanwei Yang
Xianjun Gao
Dongjie Yang
Lei Xu
SERNet: Spatially Enhanced Recalibration Network for Building Extraction in Dense Remote Sensing Scenes
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Building extraction
feature aggregation
high-resolution remote sensing imagery
spatial alignment
title SERNet: Spatially Enhanced Recalibration Network for Building Extraction in Dense Remote Sensing Scenes
title_full SERNet: Spatially Enhanced Recalibration Network for Building Extraction in Dense Remote Sensing Scenes
title_fullStr SERNet: Spatially Enhanced Recalibration Network for Building Extraction in Dense Remote Sensing Scenes
title_full_unstemmed SERNet: Spatially Enhanced Recalibration Network for Building Extraction in Dense Remote Sensing Scenes
title_short SERNet: Spatially Enhanced Recalibration Network for Building Extraction in Dense Remote Sensing Scenes
title_sort sernet spatially enhanced recalibration network for building extraction in dense remote sensing scenes
topic Building extraction
feature aggregation
high-resolution remote sensing imagery
spatial alignment
url https://ieeexplore.ieee.org/document/11002701/
work_keys_str_mv AT kuikuihan sernetspatiallyenhancedrecalibrationnetworkforbuildingextractionindenseremotesensingscenes
AT yuanweiyang sernetspatiallyenhancedrecalibrationnetworkforbuildingextractionindenseremotesensingscenes
AT xianjungao sernetspatiallyenhancedrecalibrationnetworkforbuildingextractionindenseremotesensingscenes
AT dongjieyang sernetspatiallyenhancedrecalibrationnetworkforbuildingextractionindenseremotesensingscenes
AT leixu sernetspatiallyenhancedrecalibrationnetworkforbuildingextractionindenseremotesensingscenes