基于激光雷达点云数据的树种分类 
下载： 1554次
Classification of Tree Species Based on LiDAR Point Cloud Data
1 南京林业大学信息科学技术学院, 江苏 南京 210037
2 南京林业大学图书馆, 江苏 南京 210037
图 & 表
图 1. 基于LiDAR系统的点云样本树示例。(a)水杉;(b)柳树;(c)女贞树;(d)竹子;(e)苹果树
Fig. 1. Examples of point cloud sample trees based on LiDAR system. (a) Metasequoia glyptostroboides; (b) Salix babylonica; (c) Ligustrum lucidum; (d) bamboo; (e) Malus pumila Mill.tree
下载图片 查看原文
图 2. 实验流程图
Fig. 2. Experimental flow chart
下载图片 查看原文
图 3. 杭州钱江新城森林公园LiDAR点云预处理图
Fig. 3. Preprocessing diagram for point cloud of Qianjiang new town Forest Park in Hangzhou collected by LiDAR
下载图片 查看原文
图 4. 新疆维吾尔自治区阿克苏地区红旗坡农场LiDAR点云预处理图
Fig. 4. Preprocessing diagram for point cloud of Hongqipo farm in Akesu Prefecture, Xinjiang Uygur Autonomous Region collected by LiDAR
下载图片 查看原文
图 5. 3D空间中中心体素和13个方向相邻体素的位置关系
Fig. 5. Location relationship between central voxels and adjacent voxels in 13 directions of 3D space
下载图片 查看原文
图 6. 划分示意图。(a)样本树的初始扫描数据;(b)使用枝叶分离算法提取的叶子点;(c)扫描叶点的体素化;(d)划分为8个部分的叶子点云顶视图;(e)划分后的第1个分段和第8个分段
Fig. 6. Segmentation diagram. (a) Initial scanned data of sample tree; (b) extracted leaf points using branch leaf segment algorithm; (c) voxelization for scanned leaf points; (d) top view for leaf point cloud divided into 8 parts; (e) 1st part and 8th part after division
下载图片 查看原文
图 7. 总体分类精度结果
Fig. 7. Overall classification accuracy results
下载图片 查看原文
表 10 基于所有参数的树种分类结果
Table1. 0 Classification results of tree species based on all parameters
| Tree specie | Metasequoiaglyptostroboides | Salixbabylonica | Ligustrumlucidum | Bamboo | Maluspumila Mill. | Leakagerate /% | Misjudgedrate /% |
|---|
| Metasequoia glyptostroboides | 15 | 0 | 2 | 0 | 0 | 16.67 | 11.76 | | Salix babylonica | 1 | 11 | 0 | 1 | 2 | 21.43 | 26.67 | | Ligustrum lucidum | 1 | 2 | 15 | 1 | 0 | 25 | 21.05 | | Bamboo | 0 | 1 | 1 | 25 | 0 | 7.41 | 7.41 | | Malus pumila Mill. | 1 | 0 | 2 | 0 | 19 | 9.52 | 13.63 | | Total | 18 | 14 | 20 | 27 | 21 | — | — | | Correct rate /% | 83.33 | 78.57 | 75 | 92.59 | 90.48 | — | — | | Overall accuracy /% | 85 | — | — | | | | |
|
查看原文
表 1样本树木参数统计表
Table1. Parameter statistics for sample trees
| Specie | Number of trees | Average treeheight /m | Average crownwidth /m | Average crownvolume /m3 | Average number ofscanned points |
|---|
| Metasequoia glyptostroboides | 54 | 25.09±2.36 | 6.67±0.64 | 468.85±38.49 | 42884 | | Salix babylonica | 42 | 12.97±1.17 | 9.11±1.18 | 430.24±36.34 | 39861 | | Ligustrum lucidum | 60 | 10.31±1.08 | 5.26±0.63 | 131.22±12.45 | 12854 | | Bamboo | 81 | 12.95±1.53 | 2.42±0.41 | 33.37±3.36 | 4324 | | Malus pumila Mill. | 63 | 8.76±0.74 | 7.62±0.86 | 233.97±21.17 | 24339 |
|
查看原文
表 3训练样本与验证样本的数量
Table3. Number of training samples and validation samples
| Specie | Number oftrainingsamples | Number ofvalidationsamples | Totalnumber |
|---|
| Metasequoiaglyptostroboides | 36 | 18 | 54 | | Salix babylonica | 28 | 14 | 42 | | Ligustrumlucidum | 40 | 20 | 60 | | Bamboo | 54 | 27 | 81 | | Maluspumila Mill. | 42 | 21 | 63 | | Total | 200 | 100 | 300 |
|
查看原文
表 4基于SF参数的树种分类结果
Table4. Classification results of tree species based on SF parameters
| Tree specie | Metasequoiaglyptostroboides | Salixbabylonica | Ligustrumlucidum | Bamboo | Maluspumila Mill. | Leakagerate /% | Misjudgedrate /% |
|---|
| Metasequoia glyptostroboides | 10 | 2 | 2 | 0 | 3 | 44.44 | 41.18 | | Salix babylonica | 2 | 10 | 7 | 3 | 4 | 28.57 | 61.54 | | Ligustrum lucidum | 0 | 0 | 6 | 3 | 0 | 70 | 33.33 | | Bamboo | 2 | 2 | 0 | 18 | 0 | 33.33 | 18.18 | | Malus pumila Mill. | 4 | 0 | 5 | 3 | 14 | 33.33 | 46.15 | | Total | 18 | 14 | 20 | 27 | 21 | — | — | | Correct rate /% | 55.56 | 71.43 | 30 | 66.67 | 66.67 | — | — | | Overall accuracy /% | 58 | — | — | | | | |
|
查看原文
表 5基于TF参数的树种分类结果
Table5. Classification results of tree species based on TF parameters
| Tree specie | Metasequoiaglyptostroboides | Salixbabylonica | Ligustrumlucidum | Bamboo | Maluspumila Mill. | Leakagerate /% | Misjudgedrate /% |
|---|
| Metasequoia glyptostroboides | 11 | 2 | 2 | 0 | 3 | 38.89 | 38.89 | | Salix babylonica | 1 | 9 | 5 | 4 | 3 | 35.71 | 59.09 | | Ligustrum lucidum | 1 | 3 | 11 | 4 | 0 | 45 | 42.11 | | Bamboo | 3 | 0 | 0 | 18 | 0 | 33.33 | 14.29 | | Malus pumila Mill. | 2 | 0 | 2 | 1 | 15 | 28.57 | 25 | | Total | 18 | 14 | 20 | 27 | 21 | — | — | | Correct rate /% | 61.11 | 64.29 | 55 | 66.67 | 71.43 | — | — | | Overall accuracy /% | 64 | — | — | | | | |
|
查看原文
表 6基于CF参数的树种分类结果
Table6. Classification results of tree species based on CF parameters
| Tree specie | Metasequoiaglyptostroboides | Salixbabylonica | Ligustrumlucidum | Bamboo | Maluspumila Mill. | Leakagerate /% | Misjudgedrate /% |
|---|
| Metasequoia glyptostroboides | 11 | 1 | 2 | 3 | 1 | 38.89 | 38.89 | | Salix babylonica | 3 | 8 | 5 | 2 | 3 | 42.86 | 61.9 | | Ligustrum lucidum | 3 | 3 | 10 | 4 | 1 | 50 | 52.38 | | Bamboo | 0 | 0 | 2 | 17 | 2 | 37.04 | 19.05 | | Malus pumila Mill. | 1 | 2 | 1 | 1 | 14 | 33.33 | 26.32 | | Total | 18 | 14 | 20 | 27 | 21 | — | — | | Correct rate /% | 61.11 | 57.14 | 50 | 62.96 | 66.67 | — | — | | Overall accuracy /% | 60 | — | — | | | | |
|
查看原文
表 7基于SF和TF参数的树种分类结果
Table7. Classification results of tree species based on SF and TF parameters
| Tree specie | Metasequoiaglyptostroboides | Salixbabylonica | Ligustrumlucidum | Bamboo | Maluspumila Mill. | Leakagerate /% | Misjudgedrate /% |
|---|
| Metasequoia glyptostroboides | 14 | 1 | 1 | 3 | 1 | 22.22 | 30 | | Salix babylonica | 2 | 10 | 3 | 0 | 2 | 28.57 | 41.18 | | Ligustrum lucidum | 2 | 2 | 12 | 2 | 0 | 40 | 33.33 | | Bamboo | 0 | 0 | 3 | 21 | 2 | 22.22 | 19.23 | | Malus pumila Mill. | 0 | 1 | 1 | 1 | 16 | 23.81 | 15.79 | | Total | 18 | 14 | 20 | 27 | 21 | — | — | | Correct rate /% | 77.78 | 71.43 | 60 | 77.78 | 76.19 | — | — | | Overall accuracy /% | 73 | — | — | | | | |
|
查看原文
表 8基于SF和CF参数的树种分类结果
Table8. Classification results of tree species based on SF and CF parameters
| Tree specie | Metasequoiaglyptostroboides | Salixbabylonica | Ligustrumlucidum | Bamboo | Maluspumila Mill. | Leakagerate /% | Misjudgedrate /% |
|---|
| Metasequoia glyptostroboides | 14 | 0 | 2 | 2 | 1 | 22.22 | 26.32 | | Salix babylonica | 2 | 9 | 3 | 1 | 3 | 37.71 | 50 | | Ligustrum lucidum | 1 | 2 | 11 | 3 | 0 | 45 | 35.29 | | Bamboo | 0 | 1 | 2 | 21 | 2 | 22.22 | 19.23 | | Malus pumila Mill. | 1 | 2 | 2 | 0 | 15 | 28.57 | 25 | | Total | 18 | 14 | 20 | 27 | 21 | — | — | | Correct rate /% | 77.78 | 62.29 | 55 | 77.78 | 71.43 | — | — | | Overall accuracy /% | 70 | — | — | | | | |
|
查看原文
表 9基于TF和CF参数的树种分类结果
Table9. Classification results of tree species based on TF and CF parameters
| Tree specie | Metasequoiaglyptostroboides | Salixbabylonica | Ligustrumlucidum | Bamboo | Maluspumila Mill. | Leakagerate /% | Misjudgedrate /% |
|---|
| Metasequoia glyptostroboides | 15 | 0 | 3 | 0 | 2 | 16.67 | 25 | | Salix babylonica | 1 | 10 | 2 | 0 | 1 | 28.57 | 28.57 | | Ligustrum lucidum | 1 | 2 | 13 | 2 | 0 | 35 | 27.77 | | Bamboo | 0 | 0 | 1 | 23 | 2 | 14.81 | 11.54 | | Malus pumila Mill. | 1 | 2 | 1 | 2 | 16 | 23.81 | 27.27 | | Total | 18 | 14 | 20 | 27 | 21 | — | — | | Correct rate /% | 83.33 | 71.43 | 65 | 85.19 | 76.19 | — | — | | Overall accuracy /% | 77 | — | — | | | | |
|
查看原文
陈向宇, 云挺, 薛联凤, 刘应安. 基于激光雷达点云数据的树种分类[J]. 激光与光电子学进展, 2019, 56(12): 122801. Xiangyu Chen, Ting Yun, Lianfeng Xue, Ying'an Liu. Classification of Tree Species Based on LiDAR Point Cloud Data[J]. Laser & Optoelectronics Progress, 2019, 56(12): 122801.
PDF全文
