Improved voxel-based volume estimation and pruning severity mapping of apple trees during the pruning period

Abstract

Estimating tree volume is crucial for managing apple orchards, as it reflects the nutritional status and vigor of the trees. However, accurate assessment of tree volume in apple orchards is challenging due to their complex structure. This study introduces a novel method for three-dimensional volume calculation of individual apple trees. Our approach facilitates the determination of pruning severity by analyzing the limb-to-trunk volume ratio and enables the creation of detailed pre- and post-pruning maps. We utilized a lightweight multi-camera system to reconstruct 3D point clouds of the trees and developed a voxel-based algorithm for tree volume calculation. This algorithm includes steps for interior filling, edge voxel thinning, and interior refilling. We validated our algorithm on seven apple trees by comparing the calculated volumes with the ground truth, determined using the water displacement method. The results showed that our voxel-based algorithm was highly effective in accurately calculating individual tree volumes from 3D point clouds. The algorithm also demonstrated a high coefficient of determination (0.994) and a mean absolute percentage error of 2.919% in a linear regression analysis against the ground truth. Furthermore, we produced detailed tree volume and pruning severity maps for individual trees, both before and after pruning. In conclusion, this study offers an effective solution combining 3D imaging and volume calculation techniques to accurately estimate individual apple tree volumes, providing a quantitative assessment of pruning severity.