High-Resolution Boundary-Constrained and Context-Enhanced Network for Remote Sensing Image Segmentation
The technology of remote sensing image segmentation has made great progress in recent years. However, there are still several challenges which need to be addressed (e.g., ground objects blocked by shadows, higher intra-class variance and lower inter-class variance). In this paper, we propose a novel...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2022-04-01
|
Series: | Remote Sensing |
Subjects: | |
Online Access: | https://www.mdpi.com/2072-4292/14/8/1859 |
_version_ | 1797409495843864576 |
---|---|
author | Yizhe Xu Jie Jiang |
author_facet | Yizhe Xu Jie Jiang |
author_sort | Yizhe Xu |
collection | DOAJ |
description | The technology of remote sensing image segmentation has made great progress in recent years. However, there are still several challenges which need to be addressed (e.g., ground objects blocked by shadows, higher intra-class variance and lower inter-class variance). In this paper, we propose a novel high-resolution boundary-constrained and context-enhanced network (HBCNet), which combines boundary information to supervise network training and utilizes the semantic information of categories with the regional feature presentations to improve final segmentation accuracy. On the one hand, we design the boundary-constrained module (BCM) and form the parallel boundary segmentation branch, which outputs the boundary segmentation results and supervises the network training simultaneously. On the other hand, we also devise a context-enhanced module (CEM), which integrates the self-attention mechanism to advance the semantic correlation between pixels of the same category. The two modules are independent and can be directly embedded in the main segmentation network to promote performance. Extensive experiments were conducted using the ISPRS Vahingen and Potsdam benchmarks. The mean F1 score (m-F1) of our model reached 91.32% and 93.38%, respectively, which exceeds most existing CNN-based models and represents state-of-the-art results. |
first_indexed | 2024-03-09T04:15:22Z |
format | Article |
id | doaj.art-be2de18977704203b27a91640f75d226 |
institution | Directory Open Access Journal |
issn | 2072-4292 |
language | English |
last_indexed | 2024-03-09T04:15:22Z |
publishDate | 2022-04-01 |
publisher | MDPI AG |
record_format | Article |
series | Remote Sensing |
spelling | doaj.art-be2de18977704203b27a91640f75d2262023-12-03T13:55:35ZengMDPI AGRemote Sensing2072-42922022-04-01148185910.3390/rs14081859High-Resolution Boundary-Constrained and Context-Enhanced Network for Remote Sensing Image SegmentationYizhe Xu0Jie Jiang1School of Instrumentation and Optoelectronic Engineering, Beihang University, Beijing 100191, ChinaSchool of Instrumentation and Optoelectronic Engineering, Beihang University, Beijing 100191, ChinaThe technology of remote sensing image segmentation has made great progress in recent years. However, there are still several challenges which need to be addressed (e.g., ground objects blocked by shadows, higher intra-class variance and lower inter-class variance). In this paper, we propose a novel high-resolution boundary-constrained and context-enhanced network (HBCNet), which combines boundary information to supervise network training and utilizes the semantic information of categories with the regional feature presentations to improve final segmentation accuracy. On the one hand, we design the boundary-constrained module (BCM) and form the parallel boundary segmentation branch, which outputs the boundary segmentation results and supervises the network training simultaneously. On the other hand, we also devise a context-enhanced module (CEM), which integrates the self-attention mechanism to advance the semantic correlation between pixels of the same category. The two modules are independent and can be directly embedded in the main segmentation network to promote performance. Extensive experiments were conducted using the ISPRS Vahingen and Potsdam benchmarks. The mean F1 score (m-F1) of our model reached 91.32% and 93.38%, respectively, which exceeds most existing CNN-based models and represents state-of-the-art results.https://www.mdpi.com/2072-4292/14/8/1859remote sensing imagesemantic segmentationattention mechanismboundary information |
spellingShingle | Yizhe Xu Jie Jiang High-Resolution Boundary-Constrained and Context-Enhanced Network for Remote Sensing Image Segmentation Remote Sensing remote sensing image semantic segmentation attention mechanism boundary information |
title | High-Resolution Boundary-Constrained and Context-Enhanced Network for Remote Sensing Image Segmentation |
title_full | High-Resolution Boundary-Constrained and Context-Enhanced Network for Remote Sensing Image Segmentation |
title_fullStr | High-Resolution Boundary-Constrained and Context-Enhanced Network for Remote Sensing Image Segmentation |
title_full_unstemmed | High-Resolution Boundary-Constrained and Context-Enhanced Network for Remote Sensing Image Segmentation |
title_short | High-Resolution Boundary-Constrained and Context-Enhanced Network for Remote Sensing Image Segmentation |
title_sort | high resolution boundary constrained and context enhanced network for remote sensing image segmentation |
topic | remote sensing image semantic segmentation attention mechanism boundary information |
url | https://www.mdpi.com/2072-4292/14/8/1859 |
work_keys_str_mv | AT yizhexu highresolutionboundaryconstrainedandcontextenhancednetworkforremotesensingimagesegmentation AT jiejiang highresolutionboundaryconstrainedandcontextenhancednetworkforremotesensingimagesegmentation |