基于无人机高光谱影像的三江源草种精细识别研究
|
|
胡宜娜,安如,艾泽天,都伟冰
|
Researches on Grass Species Fine Identification based on UAV Hyperspectral Images in Three-River Source Region
|
|
Yina Hu,Ru An,Zetian Ai,Weibing Du
|
|
表1 基于SMLR与ASR的草地识别精度
|
Table 1 The grass species identification accuracy of SMLR and ASR
|
|
| 类别 | 训练样本/ 验证样本 | SMLR-EMAP | SMLR-EMAP-SA | ASR-EMAP | ASR-EMAP-SA |
|---|
| PA | UA | PA | UA | PA | UA | PA | UA |
|---|
| 矮火绒 | 8/83 | 0.891 6 | 0.840 9 | 1.000 0 | 0.846 9 | 0.879 5 | 0.973 3 | 0.963 9 | 0.833 3 | | 棘豆 | 17/163 | 0.975 5 | 0.649 0 | 1.000 0 | 0.795 1 | 1.000 0 | 0.610 5 | 1.000 0 | 0.871 7 | | 苔藓 | 18/168 | 0.982 1 | 0.896 7 | 0.994 0 | 0.933 0 | 0.988 1 | 0.873 7 | 0.928 6 | 0.725 6 | | 细叶亚菊 | 7/69 | 0.579 7 | 0.975 6 | 0.768 1 | 0.688 3 | 0.724 6 | 0.781 3 | 0.811 6 | 0.965 5 | | 苔草 | 19/182 | 0.978 0 | 0.983 4 | 0.923 1 | 1.000 0 | 0.989 0 | 0.923 1 | 0.972 5 | 1.000 0 | | 藏嵩草 | 11/1 111 | 1.000 0 | 0.848 7 | 1.000 0 | 0.886 0 | 1.000 0 | 0.952 8 | 1.000 0 | 0.776 9 | | 小嵩草 | 3/39 | 0.615 4 | 1.000 0 | 0.666 7 | 1.000 0 | 0.512 8 | 1.000 0 | 0.384 6 | 0.882 4 | | 针茅 | 3/33 | 0.636 4 | 1.000 0 | 0.666 7 | 1.000 0 | 0.636 4 | 1.000 0 | 0.424 2 | 0.823 5 | | 裸地 | 15/1 658 | 0.956 6 | 0.995 6 | 0.968 6 | 0.999 4 | 0.935 5 | 0.995 5 | 0.959 6 | 0.995 0 | | OA | 0.940 7 | 0.957 1 | 0.931 5 | 0.942 7 | | Kappa | 0.893 3 | 0.922 5 | 0.878 5 | 0.896 7 | | 平均用时/s | 111 | 152 | 646 | 692 |
|
|
|