Individual-Tree Parameters Measurement Based On TLS

Vegetation structure parameters are the key elements for the study of ecosystem function and ecosystem interaction.Due to the amount of plant matter in the trees,it is impractical to carry out a detailed search of many of these parameters by direct measurement.The Terrestrial Lidar Scanner（TLS）has been shown to have high potential as an indirect measure of highly detailed estimates of plant structure parameters,and some studies have identified some of the challenges inherent in this approach.In order to better understand and predict how terrestrial ecosystems respond to and influence environmental change,information on the structure of ecosystems within a small scale is needed.LiDAR remote sensing from ground-based instruments is a promising technology to provide this information.The ground-based laser radar is not affected by the weather,reducing the impact of leaf cover while splicing multi-view scanning data to ensure high density of scanned point cloud data.Accuracy Integrity,physical models are ideal for processing data and obtaining reliable vegetation parameters.In this study,we mainly studied the skeleton extraction and leaf area density inversion of tree point cloud data.We introduce the concept of ₁L-median value to obtain a robust point cloud live standing skeleton line,and iterate the point cloud data locally using a random sample point into a one-dimensional skeleton point,and regard it as the local center of the branch.The main advantage of this method is that there is no high requirement for the accuracy and quality of point cloud data.At the same time,the requirements for the topological structure of tree trunk space are not as stringent.In the process of extracting the skeleton of cherry tree and sapindilla,not only can Effective anti-noise,and there is a certain correction effect for outliers and point cloud data missing.At the same time,we use a voxel-based ray tracing algorithm to retrieve the leaf area distribution of individual trees.This method is based on the contact frequency method applied to ground-based laser scanning radar（TLS）from a single or multiple scanning positions.The size of the calculated voxels can be set to 10,20,and 30 mm.First,the normal vector is extracted,and the characteristics are added to realize the separation of the leaves of the point cloud data.Then,the effect of the laser beam size on the overlapping effect of the blades is simulated,and the light transmission is simulated through computer simulation.The model is used to mitigate the occlusion effect,based on the availability of light,using a correction factor to estimate the leaf area of the voxel where the overlap effect is too pronounced.We compared the leaf area estimates retrieved from the point cloud data with the direct measurements obtained by collecting the leaves.In addition to this,we focused on the impact of the Lambert-Beer law on the transmission of light.Through computer simulations of the affected distance traveled by the leaves in the voxel,a new model was developed to invert the leaf area density.The sampled trees had a measured leaf area of 5-30 m2 and a canopy value of 0.75-3.2.The average difference between the leaf area estimate and the reference measurement（relative deviation）is 13%.Our method provides the process of inverting the leaf area density of a single tree,and at the same time it can indirectly estimate the leaf area index and further estimate the structural parameters of the vegetation through the vertical distribution of the trees.