【摘要】 为建立适合省级尺度的海洋藻类养殖区高精度智能提取方法，推进海洋藻类养殖面积的准确测算和变化监测，本文基于国产高分辨率遥感影像，对海洋藻类养殖区遥感智能提取方法进行研究。对比U-Net模型、DeepLab V3+模型、MSUResUnet模型在典型海洋藻类养殖区提取的结果，MSUResUnet模型提取的准确率（accuracy）、召回率（recall）、平均交并比（m IoU）、F1分数（F1-score）较U-Net模型提高0.14%、0.84%、0.34%、0.32%，较DeepLab V3+模型提高0.18%、0.88%、0.40%、0.36%，因此，选择MSUResUnet模型进行福建全省海洋藻类养殖区自动化提取。经提取结果统计，2022年7月至2023年5月，福建全省海洋藻类养殖面积约为345.6912 km~2。更多还原

【Abstract】 In order to develop a high-precision intelligent extraction method suitable for provincial-level marine algae aquaculture areas and to promote accurate calculation and change detection of these areas, this paper studies the remote sensing intelligent extraction method for marine algae aquaculture areas based on domestic high-resolution remote sensing images. Comparing the extraction results of the U-Net model, DeepLab V3+model, and MSUResUnet model in typical marine algae aquaculture areas, the accuracy(Accuracy), recall,mean intersection over union(mIoU), and F1 score of the MSUResUnet model were improved by 0.14%,0.84%, 0.34%, and 0.32% compared to the U-Net model, and were improved by 0.18%, 0.88%, 0.40%, and0.36% compared to the DeepLab V3+model. Therefore, the MSUResUnet model was chosen for automated extraction of marine algae aquaculture areas in Fujian Province. According to the extraction results, the total area of marine algae aquaculture in Fujian Province from July 2022 to May 2023 was approximately345.6912 km~2.更多还原