Hong-sheng Zhang ^1* JIYU SUN ¹ Jiao-Jiao LU ¹ Ying-Gang Zheng ² Ying-Gang Zheng ³

• ¹ School of Marine Science and Engineering, Shanghai Maritime University, Shanghai, Shanghai Municipality, China
• ² Translational Research Institute of Brain and Brain like Intelligence, School of Medicine, Tongji University, Shanghai, China
• ³ Shanghai Communications Construction Contracting Co, Ltd, Shanghai, 201306, China., Shanghai, China

The study of oceanic internal waves remains a critical area of research within oceanography. With the rapid advancements in oceanic remote sensing and deep learning, it is now possible to extract valuable insights from vast datasets. In this context, by building data sets using deep learning models, we propose a novel stripe segmentation algorithm for oceanic internal waves, leveraging Synthetic Aperture Radar (SAR) images based on the Segformer architecture. Initially, a hierarchical Transformer encoder transforms the image into multi-level feature maps. Subsequently, information from various layers aggregates through an MLP Decoder, effectively merging local and global contexts. Finally, an MLP (Multilayer Perceptron) layer is utilized to facilitate the segmentation of oceanic internal waves. Comparative experimental results demonstrate that Segformer outperforms other models such as U-Net, Fast-SCNN, ORCNet, and PSPNet, efficiently and accurately segmenting marine internal wave stripes in SAR images. Additionally, we discuss the results of oceanic internal wave detection under varying settings, further underscoring the algorithm's effectiveness.