融合SAR影像与辅助数据的图神经网络地表水体提取方法OA

In the extraction of surface water bodies from remote sensing imagery,synthetic aperture radar(SAR)images are often affected by speckle noise and complex background interference,leading to blurred boundaries and misjudgment.To address this issue,this paper proposed a method that integrated SAR imagery with auxiliary data using graph neural net-works.First,superpixel segmentation was applied to convert the image into a graph structure with superpixels as nodes,which helped to smooth out noise while reducing the number of processing units.Multi-source features related to water body extraction and identification were extracted based on SAR imagery and digital elevation model(DEM),and random forests were utilized for feature selection and dimensionality reduction to enhance feature discriminability.Subsequently,a deeper graph convolutional network(DeeperGCN)was introduced to classify the superpixel nodes,achieving precise water body identification.Experiments on real datasets demonstrate that the proposed method outperforms mainstream baseline models in terms of overall accuracy(0.992 1),precision(0.966 2),recall(0.964 4),and F1 score(0.965 3),showcasing good gen-eralization capabilities and providing a novel approach for the fusion of multi-source remote sensing data and the application of graph neural networks.