Remote Sensing Technology and Application, Volume. 39, Issue 4, 897(2024)
High Spatial-Hyperspectral Tree Species Classification based on 3D-Octave Convolution
Figures & Tables(8)
Fig. 1. Regional location of Fuzhou University Ecological Botanical Garden with true color display of hyperspectral images
Fig. 6. Effect of the number of training samples on the classification ability of the model
Table 1.
Tree species categories and the number of pixels of training samples， test samples and total samples View tableView in ArticleTable 1.
Tree species categories and the number of pixels of training samples， test samples and total samples 序号 树种类别 训练样本 测试样本 标记样本总数 1 樟树 120 588 482 354 602 942 2 黄山栾树 45 436 181 744 227 180 3 盆架树 42 606 170 422 213 028 4 洋紫荆 7 946 31 784 39 730 5 南洋松 7 813 31 254 39 067 6 木棉 25 347 101 388 126 735 7 法国梧桐 7 212 28 846 36 058 8 海枣树 28 345 113 378 141 723 9 大王椰 10 305 41 220 51 525 10 大叶杜英 39 615 158 459 198 074 11 晚樱 3 834 15 334 19 168 12 金叶女贞 40 772 163 089 203 861 13 红花继木 18 478 73 911 92 389 14 黄皮 755 3 020 3 775 15 高山榕 52 865 211 459 264 324 16 紫薇 20 279 81 117 101 396 17 秋枫 10 632 42 526 53 158 18 鱼尾葵 24 622 98 487 123 109 19 三角枫 9 187 36 747 45 934 20 重阳木 158 252 633 010 791 262 21 芒果 6 192 24 768 30 960
Table 2.
Classification indexes （%） of different methods on the data set of ecological botanical gardens View tableView in ArticleTable 2.
Classification indexes （%） of different methods on the data set of ecological botanical gardens 类别 SVM ELM 2D-CNN 3D-CNN 3DOC-SSAM 类别 SVM ELM 2D-CNN 3D-CNN 3DOC-SSAM 樟树 83.70 80.78 85.26 94.24 99.56 红花继木 98.47 99.25 98.29 98.21 99.95 黄山栾树 79.36 75.07 83.24 92.23 99.12 黄皮 87.68 71.43 85.34 98.19 99.27 盆架树 93.99 89.12 94.11 97.98 99.71 高山榕 78.35 71.17 79.25 90.98 99.33 洋紫荆 93.72 86.69 93.09 98.24 99.74 紫薇 76.67 68.71 76.27 90.86 99.46 南洋松 91.42 89.14 94.28 98.36 99.76 秋枫 80.79 63.22 77.03 91.36 99.30 木棉 80.90 65.72 82.07 92.18 99.40 鱼尾葵 66.76 59.71 65.14 86.30 98.68 法国梧桐 75.18 71.10 94.77 95.47 99.44 三角枫 86.04 79.42 79.85 95.09 98.96 海枣树 94.68 88.13 94.77 97.95 99.65 重阳木 93.22 89.56 94.40 97.22 99.78 大王椰 60.23 54.26 64.22 83.10 99.03 芒果 60.64 69.66 68.50 89.92 98.22 大叶杜英 81.67 70.49 81.20 92.43 99.40 OA 85.67 81.04 86.63 94.17 99.53 晚樱 71.12 56.08 71.69 87.95 98.73 AA 78.51 75.98 83.84 93.64 99.38 金叶汝贞 97.94 96.80 97.96 98.33 99.98 Kappa 83.85 78.51 84.88 93.42 99.47
Tools
Get Citation
Copy Citation Text
Mingming WANG, Yunzhi CHEN, Yan DONG, Lei LIU, Yu Ke WANG. High Spatial-Hyperspectral Tree Species Classification based on 3D-Octave Convolution[J]. Remote Sensing Technology and Application, 2024, 39(4): 897
Paper Information
Category:
Received: May. 4, 2023
Accepted: --
Published Online: Jan. 6, 2025
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20