Remote Sensing Technology and Application, Volume. 40, Issue 2, 321(2025)
Comparison of Domain Adaptation Methods for Hyperspectral Image Classification
Figures & Tables(11)
Fig. 6. Overall classification accuracy of all methods with different training samples
Table 1.
Classes and samples of Pavia University and Pavia Center data View tableView in ArticleTable 1.
Classes and samples of Pavia University and Pavia Center data 类别 样本数量 No. 名称 帕维亚中心(源域) 帕维亚大学(目标域) 1 柏油路 9 248 6 631 2 草地 3 090 18 649 3 树木 7 598 3 064 4 裸地 6 584 5 029 5 沥青屋顶 7 287 1 330 6 砖块 2 685 3 682 7 阴影 2 863 947 总和 39 355 39 332
Table 2.
Land cover categories and sample in the HyRANK dataset View tableView in ArticleTable 2.
Land cover categories and sample in the HyRANK dataset 类别 样本数量 No. 名称 Dioni(源域) Loukia(目标域) 1 密集的城市结构 1 262 206 2 矿场 204 54 3 非灌溉耕地 614 426 4 果树 150 79 5 橄榄林 1 768 1 107 6 针叶林 361 422 7 茂密的苔藓植被 5 035 2 996 8 稀疏的苔藓植被 6 374 2 361 9 植被稀疏的地区 1 754 399 10 岩石和沙子 492 453 11 水体 1 612 1 393 12 沿海水域 398 421 总和 20 024 10 317
Table 3.
Classification results（<bold>%</bold>） for the Pavia University （5<bold>%</bold> sample） View tableView in ArticleTable 3.
Classification results（<bold>%</bold>） for the Pavia University （5<bold>%</bold> sample） 类别 领域自适应方法 CORAL JDA KBDIFS MTJDL-SLR GFK JGSA DANN TSTNET Kappa 33.73 42.68 56.83 68.54 42.95 48.16 63.78 70.56 1 0.2 65.37 79.78 51.83 70.81 72.02 68.03 91.84 2 48.82 46.41 78.97 71.92 35.35 39.91 89.85 95.13 3 79.13 85.76 73.52 89.73 84.43 96.47 68.33 78.02 4 44.51 77.83 75.13 94.84 87.36 68.52 49.24 53.56 5 31.16 7.65 25.12 83.35 33.74 15.38 75.20 60.07 6 95.57 23.01 23.64 10.40 37.91 78.19 65.95 85.34 7 72.33 94.24 80.88 89.11 96.02 93.73 97.73 98.08 OA 46.85 54.37 63.86 73.62 53.43 57.48 72.28 75.26 AA 53.07 57.14 62.39 70.14 63.63 66.27 73.49 80.29
Table 4.
Classification results （<bold>%</bold>） for the Loukia dataset （5<bold>%</bold> sample） View tableView in ArticleTable 4.
Classification results （<bold>%</bold>） for the Loukia dataset （5<bold>%</bold> sample） 类别 领域自适应方法 CORAL JDA KBDIFS MTJDL-SLR GFK JGSA DANN TSTNET Kappa 35.42 45.94 47.47 52.44 45.20 47.16 49.71 55.36 1 15.05 10.68 11.65 26.21 14.08 15.52 18.45 19.61 2 5.18 40.74 0 18.52 35.47 57.40 14.81 44.45 3 3.63 23.47 3.52 23.71 8.22 43.45 7.51 20.76 4 0 3.80 0 0 1.27 1.27 10.13 4.03 5 0.18 17.71 29.09 48.69 1.45 0.66 10.12 51.52 6 27.51 4.98 40.52 45.97 36.48 11.86 5.21 14.71 7 82.47 62.05 64.12 60.58 62.62 78.23 81.95 66.69 8 3.38 65.14 57.65 67.26 54.55 45.12 72.94 66.02 9 5.08 63.16 71.18 37.09 95.49 28.81 7.02 78.59 10 2.64 0 23.18 4.86 6.62 44.36 0.22 13.04 11 100 100 100 100 100 100 100 100 12 100 95.96 100 100 100 100 100 100 OA 43.01 56.31 58.32 60.92 54.73 56.67 60.47 63.32 AA 28.76 40.64 41.74 44.41 43.02 43.89 35.70 48.29
Table 5.
Comparison of running times of different methods View tableView in ArticleTable 5.
Comparison of running times of different methods CORAL JDA KBDIFS MTJDL-SLR GFK JGSA DANN TSTNET Pavia U 12 784 13 669 12 446 11 948 19 489 24 342 9 835 11 042 Loukia 15 864 16 436 15 391 14 712 21 146 26 741 11 247 13 795
Tools
Get Citation
Copy Citation Text
Wanglei WENG, Weiwei SUN, Kai REN, Jiangtao PENG, Gang YANG. Comparison of Domain Adaptation Methods for Hyperspectral Image Classification[J]. Remote Sensing Technology and Application, 2025, 40(2): 321
Paper Information
Category:
Received: Aug. 18, 2022
Accepted: --
Published Online: May. 23, 2025
The Author Email: Weiwei SUN (sunweiwei@nbu.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20