EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

MSAFNet: Multiscale Successive Attention Fusion Network for Water Body Extraction of Remote Sensing Images

Xin Lyu, Wenxuan Jiang, Xin Li, Yiwei Fang, Zhennan Xu, Xinyuan Wang

Remote Sensing · 15.06.2023 · 15 (12) · 3121

DOI: 10.3390/rs15123121 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Remote Sensing, online erschienen am 15.06.2023. Verfasst von Xin Lyu und 5 weiteren Autor:innen. Die Arbeit wurde bisher 22-mal zitiert (Crossref) und nennt 61 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Excellent Post-doctoral Program of Jiangsu Province, Fundamental Research Funds for the Central Universities, Project of Water Science and Technology of Jiangsu Province.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Water body extraction is a typical task in the semantic segmentation of remote sensing images (RSIs). Deep convolutional neural networks (DCNNs) outperform traditional methods in mining visual features; however, due to the inherent convolutional mechanism of the network, spatial details and abstract semantic representations at different levels are difficult to capture accurately at the same time, and then the extraction results decline to become suboptimal, especially on narrow areas and boundaries. To address the above-mentioned problem, a multiscale successive attention fusion network, named MSAFNet, is proposed to efficiently aggregate the multiscale features from two aspects. A successive attention fusion module (SAFM) is first devised to extract multiscale and fine-grained features of water bodies, while a joint attention module (JAM) is proposed to further mine salient semantic information by jointly modeling contextual dependencies. Furthermore, the multi-level features extracted by the above-mentioned modules are aggregated by a feature fusion module (FFM) so that the edges of water bodies are well mapped, directly improving the segmentation of various water bodies. Extensive experiments were conducted on the Qinghai-Tibet Plateau Lake (QTPL) and the Land-cOVEr Domain Adaptive semantic segmentation (LoveDA) datasets. Numerically, MSAFNet reached the highest accuracy on both QTPL and LoveDA datasets, including Kappa, MIoU, FWIoU, F1, and OA, outperforming several mainstream methods. Regarding the QTPL dataset, MSAFNet peaked at 99.14% and 98.97% in terms of F1 and OA. Although the LoveDA dataset is more challenging, MSAFNet retained the best performance, with F1 and OA being 97.69% and 95.87%. Additionally, visual inspections exhibited consistency with numerical evaluations.

Förderung

  • Excellent Post-doctoral Program of Jiangsu Province
  • Fundamental Research Funds for the Central Universities
  • Project of Water Science and Technology of Jiangsu Province
  • National Natural Science Foundation of China
  • Joint Fund of Ministry of Education for Equipment Pre-research
  • Research Fund from Science and Technology on Underwater Vehicle Technology Laboratory
  • Qinglan Project of Jiangsu Province

Zitieren

APA

Lyu, X., Jiang, W., Li, X., Fang, Y., Xu, Z., & Wang, X. (2023). MSAFNet: Multiscale Successive Attention Fusion Network for Water Body Extraction of Remote Sensing Images. Remote Sensing, 15(12), 3121. https://doi.org/10.3390/rs15123121

Vancouver

Lyu X, Jiang W, Li X, Fang Y, Xu Z, Wang X. MSAFNet: Multiscale Successive Attention Fusion Network for Water Body Extraction of Remote Sensing Images. Remote Sensing. 2023;15(12):3121. doi:10.3390/rs15123121

BibTeX
@article{Lyu2023,
  author = {Lyu, Xin and Jiang, Wenxuan and Li, Xin and Fang, Yiwei and Xu, Zhennan and Wang, Xinyuan},
  title = {MSAFNet: Multiscale Successive Attention Fusion Network for Water Body Extraction of Remote Sensing Images},
  journal = {Remote Sensing},
  year = {2023},
  volume = {15},
  number = {12},
  pages = {3121},
  doi = {10.3390/rs15123121}
}
Kostenlos starten →