Geometric Modeling Of Individual Trees Considering The Distribution Of Canopy Leaf Area Index

Forests are an important part of terrestrial ecosystems that play a very important role in climate regulation,maintaining ecological balance and global environmental change.Single tree as an important part of the forest,its three-dimensional geometric model can be widely used in forestry,botany,remote sensing,ecology and virtual reality and other fields.In view of the high complexity and difference of the geometric structure of the trees,the traditional tree geometric modeling method has obvious shortcomings and limitations in the aspect of tree geometry information acquisition and modeling,which leads to tree geometry and actual tree geometry and there are large differences in form.Compared with the traditional tree geometric modeling method,the terrestrial lidar uses the non-contact method to obtain the massive cloud data of the canopy geometrical structure,which can overcome the current traditional tree geometric modeling method modeling results fidelity is not high and the limitations of large differences with the field trees,so as to achieve high-fidelity tree three-dimensional geometric modeling possible.Based on the collected tree point cloud data,this paper uses the level set,clustering and graph theory to extract the tree skeleton points of the individual trees,and on this basis,the tree branch geometry modeling was realized.The reconstructed model was analyzed and discussed.The main research contents and achievements are as follows:(1)Data collection and preprocessing.A reasonable set of observational programs was designed to simulate and collect the canopy point cloud data of single tree in lush and deciduous stages.Based on the three-dimensional geometric model of typical single trees,the point cloud simulation was carried out with the observation scheme and ILMSVP(In-situ LAI Measurement Simulation and And Validation Platform)software.The typical trees were selected as the observation targets,and based on the point cloud observation scheme and the Rigel-VZ400 scanner,the tree point cloud collection during the lush period and the deciduous stage was carried out.The geometric parameters such as tree height,DBH and branch radius were collected,and the geometric parameters of the leaves were selected,and leaves point cloud collection.According to the point cloud data collected in the field to carry out registration,denoising and other pretreatment.Finally,the main direction of the tree point data is calculated,and the classification is calculated.(2)Geometric modeling of single tree branches in deciduous stages.Based on the single tree point cloud data,the Dijkstra algorithm is used to solve the maximum path length of the geodesic graph.The tree point cloud of the single tree is stratified in the path of geodesic graph direction and the arithmetic mean value of the cloud data as a skeleton point.When the same layer exists branch,the European clustering method is used to determine the number of branches and calculate the branch skeleton points.According to the collection of isolated points,that is,with the individual trees canopy branch point of aggregation points do not converge the points gathered separately,the extraction method of tree branches and skeleton points was consistent with that of single trees.The skeleton points topology is constructed by using the minimum spanning tree,and the radius of the branches at the skeleton points is estimated based on the least squares algorithm.It is difficult to estimate the position of the branch radius accurately(such as the last branch,branch bifurcation and turning part).In this paper,the branch radius is estimated by using the pipe model.Single tree stem and branches were used as standard geometries to construct their individual tree model.The results show that the tree point cloud data with the angular resolution of 0.04 0 can meet the precision requirements of the individual geometric modeling of trees,and the different interval intervals only affect the details of the model,and the four kinds of observation schemes are effective for the model and geometric parameters less affected,The reconstructed deciduous tree model has a higher accuracy.(3)Geometry modeling of single tree canopy in the lush stage.The geometrical modeling of tree branches in the shrubs is the same as that of the tree branches in deciduous stages.For the geometrical modeling of tree canopy leaves in the lush stage,modeling is based on the classification of the leaf point cloud,and the leaf area of single tree was measured by the gapfraction model.The results of the measurement of the leaf area of each voxel were used as the constraint condition of the geometric modeling of the canopy leaf of single tree,finally realized the geometric modeling of canopy leaves of single tree.The results show that the method of point cloud classification can better realize the point cloud classification of single tree.The results of the canopy area measurement of single tree with 1 °zenith angle interval are the highest,and different observation programs had little effect on the leaf area of single tree.The results showed that the effect of the maximum leaf length on the leaf size was the highest and the precision was higher.The model of leaf area index of single tree was higher in this paper.Through the above analysis,it is concluded that the accuracy of single tree geometric modeling based on the distribution of leaf area index of single tree is high,which proves the feasibility of modeling method.