A daytime sea fog detection algorithm combining LinkNet network and coordinate attention mechanism
LIU Ao
JI Yonggang
XIAO Yanfang
Abstract：Sea fog is an important weather phenomenon that affects horizontal visibility at sea.The re-mote sensing monitoring of sea fog is of great significance to marine activities such as maritime traffic and oil development.Based on the FY-3D polar-orbiting meteorological satellite data,this paper pro-posed a deep learning-based daytime sea fog remote sensing detection model CA-LinkNet.The model used the LinkNet network as the backbone,and introduced the coordinate attention mechanism(CA)in the skip connection part of the encoder and decoder to make full use of the position and channel infor-mation of the node feature map,and to enhance the localization and recognition capabilities of the net-work.The experimental results showed that compared with FCN,U-Net and LinkNet segmentation networks,the CA-LinkNet network proposed in this paper had higher accuracy in sea fog detection on the test set,with the mean intersection over union,probability of detection,false alarm rate,critical success index and Heidke skill score reaching 0.911,0.900,0.046,0.862,0.905,respectively.The model was used to detect sea fog in the Yellow Sea,the Sea of Japan and the Sea of Khozik and sea fog near the Chiba Islands,and the detection results were in good consistent with the CALIOP data.
Keywords：FY-3DLinkNetcoordinate attention mechanismsea fog detection
Publication Date：2025-12-30
Online Publishing Date：2026-05-22(First online date of this platform, not the publication date of the document)
Pages：8( 63-70 )