基于UAV-LiDAR马尾松人工林树高和胸径参数提取
Extraction of Tree Height and Diameter-at-Breast-Height Parameters of Pinus massoniana Plantation Forest Based on UAV-LiDAR
严小龙 1王书涵1
作者信息
- 1. 湖南科技大学地理空间信息技术国家地方联合工程实验室,湖南 湘潭 411100;湖南科技大学地球科学与空间信息工程学院,湖南 湘潭 411201
- 折叠
摘要
鉴于无人机激光雷达技术有着快速获取目标高精度的空间三维信息的特点,使用其采集的数据对马尾松进行树高的提取验证,且与点云提取出的单木参数信息,采用多元线性回归、随机森林、极端梯度提升等方法建立马尾松胸径拟合模型并验证结果.表明,在树高提取中,点云提取的树高与真实树高具有高度相关性,其拟合模型的决定系数R2为0.78,均方根差为0.66 m;在少量参数特征的情况下,多元线性回归拟合的胸径模型结果最好,其R2为0.64,均方根差为1.21 cm.机载激光雷达的点云提取拟合马尾松单木参数并拟合胸径具有可靠性,为实现大区域的机载激光雷达点云获取马尾松胸径信息提供依据.
Abstract
In view of the characteristics of UAV LiDAR technology to quickly obtain high-precision three-dimensional spatial infor-mation of targets,the collected data was used to extract and verify the tree height of Pinus massoniana,and the single tree parameter information extracted from the point cloud was extracted.Multiple linear regression,random forest,extreme gradient lifting and oth-er methods were used to establish the diameter fitting model of Pinus massoniana and verify the results.The results showed that the tree height extracted by point cloud is highly correlated with the real tree height,and the determination coefficient R2 of the fitting model is 0.78,and the root-mean-square difference is 0.66 m.In the case of a small number of parameter features,the multiple linear regression fitted DBH model has the best results with R2 of 0.64 and root-mean-square difference of 1.21 cm.Airborne LiDAR point cloud extraction and fitting of Pinus massoniana single tree parameters and DBH are reliable,which provides a certain basis for the realization of large-area airborne LiDAR point cloud to obtain the DBH information of Pinus massoniana.
关键词
激光雷达/马尾松/胸径/极端梯度提升/机器学习Key words
laser radar/Pinus massoniana/DBH/extreme gradient lifting/machine learning引用本文复制引用
基金项目
湖南省自然科学基金项目(2020JJ5164)
出版年
2024
NETL
NSTL
维普
万方数据