Remote Sensing Technology and Application, Volume. 39, Issue 5, 1106(2024)
Research on Remote Sensing Intelligent Extraction Method of Tropical Rice Planting Area based on Deep Learning: A Case Study of Haikou City, Hainan Province
Figures & Tables(13)
Fig. 8. Comparison of model extraction effect between V2.0 （blue range line） and V4.0 （green range line）
Table 1.
Statistical table of sample image information View tableView in ArticleTable 1.
Statistical table of sample image information 地区 影像类型 影像分辨率 影像数量 影像时间 图斑个数 海口市美兰区 吉林一号 2.9 2 20221016 2 258 北京二号 3.2 2 20220828、20220901 499 WorldView 2 2 1 20230505 22 海口市琼山区 吉林一号 2、2.9、3.2 9 20221012、20220827、20220408、20221016、20221012、20220920 4 706 北京二号 3.2 4 20220901、20220828 2 497 WorldView 2 2 2 20230505 553 高景一号 2 1 20230324 171 海口市秀英区 吉林一号 2.9 3 20221012 2 144 儋州市 吉林一号 2.9 10 20221011 6 593 三亚市天涯区 吉林一号 2.9 1 20220917 771
Table 2.
Confusion matrix
Table 3.
Comparison of semantic segmentation network accuracy View tableView in ArticleTable 3.
Comparison of semantic segmentation network accuracy 指标 方法 切片1 切片2 切片3 平均值 精确度 Unet 0.965 0.952 0.921 0.946 SegNet 0.933 0.954 0.941 0.935 PSPNet 0.910 0.916 0.887 0.895 DeepLabV3+ 0.933 0.921 0.948 0.938 精准率 Unet 0.988 0.970 0.962 0.972 SegNet 0.932 0.933 0.915 0.925 PSPNet 0.881 0.826 0.866 0.854 DeepLabV3+ 0.942 0.944 0.926 0.931 召回度 Unet 0.877 0.871 0.854 0.869 SegNet 0.966 0.943 0.961 0.952 PSPNet 0.896 0.917 0.875 0.902 DeepLabV3+ 0.922 0.901 0.887 0.912 F1 score Unet 0.863 0.918 0.893 0.901 SegNet 0.892 0.861 0.897 0.888 PSPNet 0.811 0.826 0.809 0.821 DeepLabV3+ 0.923 0.951 0.964 0.941 IoU Unet 0.859 0.871 0.855 0.863 SegNet 0.796 0.810 0.806 0.803 PSPNet 0.743 0.739 0.721 0.732 DeepLabV3+ 0.902 0.897 0.891 0.895
Table 4.
Iterative training parameters of rice interpretation model View tableView in ArticleTable 4.
Iterative training parameters of rice interpretation model 模型版本 水稻图斑数量 样本切片数量 批大小 学习率 训练轮数 测试集比例 模型精度 V1.0 18 000 2 721 12 0.000 1 50 10% 61% V2.0 18 000 2 721 12 0.000 05 100 10% 75% V3.0 18 000 2 721 12 0.000 05 200 10% 74% V4.0 20 170 3 301 12 0.000 1 200 10% 87%
Table 5.
Statistics of rice planting area extraction results View tableView in ArticleTable 5.
Statistics of rice planting area extraction results 指标 红旗镇 三门坡镇 云龙镇 平均值 准确率 0.84 0.803 0.813 0.819 精确率 0.821 0.807 0.79 0.806 召回率 0.877 0.871 0.854 0.867 F1 得分 0.783 0.851 0.843 0.826 平均交并比 0.859 0.821 0.855 0.845
Tools
Get Citation
Copy Citation Text
Chunxiao WANG, Zengzhao XING, Jinsha LU, Fei CAO, Jianxin SUN, Xiaojing CAI, Xiaojuan LIU, Xiaoqing XIONG. Research on Remote Sensing Intelligent Extraction Method of Tropical Rice Planting Area based on Deep Learning: A Case Study of Haikou City, Hainan Province[J]. Remote Sensing Technology and Application, 2024, 39(5): 1106
Paper Information
Category:
Received: Feb. 4, 2024
Accepted: --
Published Online: Jan. 7, 2025
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20