| 摘要: |
| 为从遥感图像中获取海洋内波信息，提出了一种优化的TransUNet框架条纹分割算法，并验证了其在海洋内波合成孔径雷达（SAR）图像明暗条纹和明亮条纹分割中的有效性.相较于原始TransUNet和Unet模型，本模型的分割结果较为完整且没有断裂，且可以通过改变算法的复杂度，适应不同的数据规模，为海洋内波的反演研究奠定基础，具有良好的应用潜力. |
| 关键词: 内波条纹 合成孔径雷达（SAR） TransUNet 深度学习 分割 |
| DOI：10.20192/j.cnki.JSHNU（NS）.2025.02.012 |
| 分类号:P752 |
| 基金项目:国家自然科学基金（52201321） |
|
| Segmentation of internal ocean wave stripes based on deep learning network models |
|
WU Xuyun1, ZHANG Yu2, JIA Haiqing1
|
|
1.Shanghai Marine Monitoring and Forecasting Center, Shanghai 200062, China;2.College of Harbor, Coastal and Offshore Engineering, Hohai University, Nanjing 210024, Jiangsu, China
|
| Abstract: |
| An optimized TransUNet framework stripe segmentation algorithm was proposed to obtain ocean internal wave information from remote sensing image， and its effectiveness in segmenting bright and dark stripes in synthetic aperture radar （SAR） images of ocean internal waves was verified compared with the original TransUNet and UNet models. The segmentation results of the proposed model were relatively complete without breakage， and could adapt to different data scales by changing the complexity of the algorithm， which layed the foundation for the inversion research of ocean internal waves and retained good application potential. |
| Key words: internal wave stripe synthetic aperture radar （SAR） TransUNet deep learning segmentation |
