摘要
[目的]基于无人机激光雷达(LiDAR)点云数据提取杉木树冠上部结构参数(树冠顶点、树高、冠幅和上部冠长),并进行树冠上部外轮廓模拟与可视化,为树种识别提供树木冠形特征。[方法]利用LASTools开源工具从激光雷达点云数据生成无孔洞的冠层高度模型,使用LiDAR360软件,采用局部最大值法检测树冠顶点,基于CHM种子点对点云进行单木分割,并在ArcGIS下手动选取杉木单株点云样本,用Python编程对“欠分割”样本进行单木纯化(之后全部编程方式自动化处理);提取纯化后单株样本的树冠上部结构参数(树冠顶点、树高、冠幅和上部冠长),再对单木点云按照一定高度间隔进行分层切片,使用宽度百分位数法提取单木树冠上部的相对着枝深度、枝条长度作为模型变量,以相对着枝深度分层分别建模与验证样本按照3倍标准差法剔除异常外轮廓点,选取二次多项式、幂函数和指数函数3个基础模型进行模型拟合与验证,最后采用最优拟合模型进行样地尺度的三维可视化。[结果]无人机激光雷达综合单木检测率为79.63%,结合实测参数与提取结果进行相关分析,树高线性回归R^(2)为0.8905,冠幅线性回归R^(2)为0.8456;二次抛物线、幂函数和对数函数拟合R^(2)分别为0.8070、0.8170、0.8060,幂函数对杉木树冠上部外轮廓的拟合效果更优。[结论]在高林分密度条件下,单木点云的有效提取纯化对客观描绘树冠形状非常重要;基于无人机激光雷达拟合的杉木树冠上部外轮廓反映了杉木的树冠上部形态,可为杉木的树种识别提供参考。
[Objective]To obtain the upper crown structure parameters(crown vertex,tree height,crown width and upper crown length)and to simulate and visualize the outer upper crown profile of Cunninghamia lanceolata based on Unmanned Aerial Vehicle(UAV)LiDAR data in order to provide reference for recognizing the individual tree crown characteristics and tree species.[Method]By means of local maximum method in LiDAR360 software,the Pit-Free CHM(canopy height model),which was generated from LiDAR point clouds by LASTools open source software,was used to detect the crown vertex.Then point clouds segmentation based on CHM seed points was carried out with the use of the previous crown vertex.After that,the individual tree point clouds was selected in ArcGIS software manually,and the samples of less segmentation were purified automatically by python programming(the entire following work was done automatically by programming),then the individual tree point clouds data was ready.Secondly,the upper crown structure parameters(crown vertex,tree height,crown width and upper crown length)were extracted from purified individual tree point clouds.Thirdly,according to certain interval,the purified individual tree point clouds were hierarchically sliced to get two outer upper crown profile parameters(relative depth into the crown of interest,branch length)as model variables by width percentile method.By the way of stratified relative depth into the crown of interest,the abnormal outer upper crown profile parameters of modeling and validation samples were removed according to the method of 3 times standard deviation respectively.Three candidate basic models,quadratic polynomial,power function and exponential function,were selected for model fitting and verification.Finally,the optimal fitting model was used to visualize the 3D scene at the sample plot scale.[Result]The results showed that the accuracy of tree detection was 79.63%.The correlation of the extracted values to the measured data was analysed.The R^(2) of tree height,crown radius w
作者
徐志扬
刘浩栋
陈永富
陈巧
李华玉
王娟
XU Zhi-yang;LIU Hao-dong;CHEN Yong-fu;CHEN Qiao;LI Hua-yu;WANG Juan(Research Institute of Forest Resource Information Techniques,Chinese Academy of Forestry,Key Laboratory of Forestry Remote Sensing and Information System,National Forestry and Grassland Administration,Beijing 100091,China;East China Inventory and Planning Institute,National Forestry and Grassland Administration,Hangzhou 310019,Zhejiang,China;College of Forestry,Southwest Forestry University,Kunming 650224,Yunnan,China)
出处
《林业科学研究》
CSCD
北大核心
2021年第4期40-48,共9页
Forest Research
基金
中央级公益性科研院所基本科研业务费专项资金项目“经营单位水平森林资源高效监测技术研究”,项目编号:CAFYBB2018SZ008。
关键词
无人机激光雷达
单木分割
分层切片
树冠上部外轮廓模拟
可视化
树种识别
UAV-LiDAR
individual tree segmentation
hierarchically sliced
outer upper crown profile simulation
visualization
tree species recognization
