Retrieving 3D Structural Parameters Of Maize Using LiDAR Data

Leaf area index,leaf area profile,leaf angle,and leaf inclination angle are important three-dimensional structural parameters of maize.They are closely related to maize light interception ability,evapotranspiration ability,and photosynthesis.They are also important parameters for building a crop structure model and radiation transmission model.As an emerging active remote sensing technology,Light Detection and Ranging(LiDAR)can be used to accurately obtain three-dimensional information of crops and achieve a leap from two to three dimensions.How to use LiDAR to accurately estimate the three-dimensional structural parameters of maize has become a hot issue in current research.In this paper,at the National Precision Agriculture Research Demonstration Base of the Beijing Academy of Agriculture and Forestry Sciences,the point cloud data of maize at the plot level was acquired based on unmanned aerial vehicle-LiDAR(UAV-LiDAR)in 2018,and the point cloud data of individual level of maize was acquired based on terrestrial laser scanning(TLS)in 2019.The main work and conclusions are as follows:(1)Data acquisition and preprocessing:Based on UAV-LiDAR and TLS systems,data collection methods were researched,and point cloud data of plot and individual level were carried out;and according to the characteristics of LiDAR data,data preprocessing including data solving,multi-station scanning data splicing,point cloud filtering and registration,and the classification of maize into upper,middle,and lower data according to the agronomic characteristics.(2)Estimation and analysis of plot level LAI based on UAV-LiDAR;Because the upper point cloud data of maize is least affected by the occlusion effect,based on the voxel method,the upper layer data was used to verify and analyze the optimal voxel size corresponding to each route with the measured values;According to the optimal voxel size corresponding to each route,the LAI of the entire plant was estimated,and the accuracy analysis was performed to obtain the optimal incidence angle.The LAI is more accurately estimated when the direction of UAV flight is perpendicular to the ridge direction of maize planting;the accuracy of the estimation of the upper,middle and lower layers of the maize obtained by analyzing the optimal incidence angle was obtained,and quantitatively evaluate the occlusion effect of maize point cloud obtained by UAV-LiDAR.(3)Estimation and analysis of leaf angle and leaf inclination angle based on TLS at an individual level:a method for estimating leaf angle and leaf inclination angle using a Support Vector Machine(SVM)model was proposed,including separation of stem and leaf,Density-Based Spatial Clustering of Applications with Noise(DBSCAN)clustering,and feature points extraction(Leaf tip point,leaf center point,leaf location point,the number of leaf points,and Projected area of leaf)and other data processing steps;and precision comparison with the skeleton extraction algorithm;and the leaf angle and leaf inclination angle distribution of different layers and different varieties were analyzed.(4)Estimation and analysis of Leaf Area Profiles(LAP)based on UAV-LiDAR:Based on the incidence angle and the leaf inclination angle,a method of hierarchically calculating correction factors was proposed,which improves the accuracy of LAP estimation using UAV-LiDAR;and changes of LAP and Average LAP(ALAP)of different planting densities and different growth periods were analyzed;finally,the horizontal thickness used in calculating the LAP was discussed.