基于TransUNet的辽河口碱蓬草遥感提取研究
Fine-Grained Mapping of Suaeda salsa in the Liaohe River Delta Using a Hybrid TransUNet Model
DOI: 10.12677/aep.2026.163032, PDF,   
作者: 张晓哲：辽宁师范大学地理科学学院，辽宁 大连；张 翔：大连市国土空间规划设计有限公司，辽宁 大连
关键词: 碱蓬草；语义分割；TransUNet；Sentinel-2；辽河口；Suaeda Salsa； Semantic Segmentation； TransUNet； Sentinel-2； Liaohe River Delta
摘要: 辽河口湿地碱蓬草(Suaeda salsa)群落对于维护滨海生态平衡具有重要意义。针对该区域潮滩背景复杂、植被斑块破碎及边界模糊导致的提取难题，本文引入结合了Transformer全局建模能力与U-Net细节恢复优势的TransUNet模型，开展碱蓬草精细化遥感提取研究。基于Sentinel-2多光谱遥感影像构建辽河口碱蓬草数据集，并进行模型训练与测试。实验结果表明，TransUNet模型在辽河口复杂潮滩场景下表现优异，mIoU达到95.66%，F1分数达到94.82%，能够有效克服传统方法在细碎斑块识别上的局限，较好地刻画了碱蓬草的空间分布特征。研究结果验证了TransUNet在滨海湿地植被监测中的适用性，可为辽河口湿地保护与管理提供技术支撑。
Abstract: The Suaeda salsa community in the Liaohe River Delta wetland plays a crucial role in maintaining coastal ecological balance. Addressing extraction challenges caused by complex tidal flat backgrounds, fragmented vegetation patches, and blurred boundaries, this study introduces the TransUNet model—which integrates the global modeling capabilities of Transformers with the detail-recovery advantages of U-Net—to perform fine-grained remote sensing extraction of Suaeda salsa. A dedicated dataset was constructed using Sentinel-2 multispectral imagery for model training and validation. Experimental results demonstrate that TransUNet performs exceptionally well in the complex tidal flat scenarios of the Liaohe River Delta, achieving a mean Intersection over Union (mIoU) of 95.66% and an F1-score of 94.82%. The model effectively overcomes the limitations of traditional methods in identifying small, fragmented patches and accurately depicts the spatial distribution characteristics of the vegetation. These findings verify the applicability of TransUNet for coastal wetland vegetation monitoring and provide technical support for the protection and management of the Liaohe River Delta wetlands.
文章引用：张晓哲, 张翔. 基于TransUNet的辽河口碱蓬草遥感提取研究[J]. 环境保护前沿, 2026, 16(3): 312-317. https://doi.org/10.12677/aep.2026.163032

参考文献

[1] Dong, Q., Zhang, Q., Liao, A., Xu, C. and Liu, M. (2022) Plant Adaptability and Vegetation Differentiation in the Coastal Beaches of Yellow-Bohai Sea in China. International Journal of Environmental Research and Public Health, 19, Article 2225. [Google Scholar] [CrossRef] [PubMed]
[2] Ke, L., Lu, Y., Tan, Q., Zhao, Y. and Wang, Q. (2024) Precise Mapping of Coastal Wetlands Using Time-Series Remote Sensing Images and Deep Learning Model. Frontiers in Forests and Global Change, 7, Article 1409985. [Google Scholar] [CrossRef] 
[3] Ronneberger, O., Fischer, P. and Brox, T. (2015) U-Net: Convolutional Networks for Biomedical Image Segmentation. In: Navab, N., Hornegger, J., Wells, W. and Frangi, A., Eds., Medical Image Computing and Computer-Assisted Intervention—MICCAI 2015, Springer, 234-241. [Google Scholar] [CrossRef] 
[4] Yin, H., Hu, Y., Liu, M., Li, C. and Chang, Y. (2021) Evolutions of 30-Year Spatio-Temporal Distribution and Influencing Factors of Suaeda salsa in Bohai Bay, China. Remote Sensing, 14, Article 138. [Google Scholar] [CrossRef] 
[5] Chen, J., Lu, Y., Yu, Q., Luo, X., Adeli, E., Wang, Y., Lu, L., Yuille, A.L. and Zhou, Y. (2021) TransUNet: Transformers Make Strong En-coders for Medical Image Segmentation. arXiv: 2102.04306.
[6] Drusch, M., Del Bello, U., Carlier, S., Colin, O., Fernandez, V., Gascon, F., et al. (2012) Sentinel-2: ESA’s Optical High-Resolution Mission for GMES Operational Services. Remote Sensing of Environment, 120, 25-36. [Google Scholar] [CrossRef] 
[7] 吴涛, 李新荣, 赵士洞, 等. 辽东湾碱蓬草群落生物量遥感估算研究[J]. 生态学报, 2011, 31(9): 2457-2466.
[8] Frampton, W.J., Dash, J., Watmough, G. and Milton, E.J. (2013) Evaluating the Capabilities of Sentinel-2 for Quantitative Estimation of Biophysical Variables in Vegetation. ISPRS Journal of Photogrammetry and Remote Sensing, 82, 83-92. [Google Scholar] [CrossRef] 
[9] Tucker, C.J. (1979) Red and Photographic Infrared Linear Combinations for Monitoring Vegetation. Remote Sensing of Environment, 8, 127-150. [Google Scholar] [CrossRef] 
[10] Zhang, S., Tian, Q., Lu, X., Li, S., He, S., Zhang, X., et al. (2024) Enhancing Chlorophyll Content Monitoring in Coastal Wetlands: Sentinel-2 and Soil-Removed Semi-Empirical Models for Phenotypically Diverse Suaeda salsa. Ecological Indicators, 167, Article ID: 112686. [Google Scholar] [CrossRef] 
[11] Lin, X., Cheng, Y., Chen, G., Chen, W., Chen, R., Gao, D., et al. (2023) Semantic Segmentation of China’s Coastal Wetlands Based on Sentinel-2 and Segformer. Remote Sensing, 15, Article 3714. [Google Scholar] [CrossRef] 
[12] Xu, J., Xiong, Z. and Bhattacharyya, S.P. (2023) PIDNet: A Real-Time Semantic Segmentation Network Inspired by PID Controllers. 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Vancouver, 17-24 June 2023, 19529-19539. [Google Scholar] [CrossRef]