Research on volume prediction of single tree canopy based on three-dimensional (3D) LiDAR and clustering segmentation

Abstract

ABSTRACT Canopy volume information of fruit trees is a very important biological parameter, which is of great significance to predict the yield of fruit trees, estimate the application amount of pesticides and fertilizers. This study proposes a novel volume prediction method of single tree canopy based on the three-dimensional (3D) Light Detection and Ranging (LiDAR) point cloud. The method involves several steps, mainly including point cloud pre-processing, spatial clustering segmentation based on K-dimensional tree (KD tree), acquisition of single tree structural parameters, calculation of tree canopy volume based on multiple regression analysis. This study tests the performance of the proposed method with a collected data set of Begonia forest. The average error and standard deviation between the predicted and manually measured heights to the canopy are 0.038 m and 0.030 m, respectively. As to the diameter of the trunk, the average error and standard deviation are 0.013 m and 0.008 m, respectively. The coefficient of determination (R 2) of the proposed canopy volume prediction method is 0.8610, and the F test result is significant. High correlation is found between the predicted canopy volumes and the R 2 value is 0.8223. The experimental results verify the validity of the proposed method. The research can provide a stable and accurate technical reference for the statistics on forest biomass.