Automated detection of branch dimensions in woody skeletons of leafless fruit tree canopies

Light driven physiological processes of tree canopies need to be modelled based on detailed 3Dcanopy structure – we explore the possibilities offered by terrestrial LIDAR to automatically represent woody skeletons of leafless trees as a basis for adequate models of canopy structure. The automatic evaluation method for LIDAR data of fruit trees is based on a previously developed skeletonization algorithm. Branch length was chosen as example parameter to test the performance of the algorithm with manually measured data. The extraction of the brach length utilizes a graph splitting procedure to extract the individual branches from the skeleton. The algorithm is validated against six leafless apple trees and one cherry tree with small blossoms. The validation against a manually measured ground truth resulted in a good correlation up to 0.81.