Abstract
This article presents a three-level frame and a novel algorithm, which is based on a searching schematic analogous to a revolving door (RD), for automatic detection of individual trees from the point clouds acquired by mobile laser scanning (MLS). As a state-of-the-art technology, MLS in collecting the structural information of single trees is different from airborne laser scanners and terrestrial laser scanners in terms of sampling density, view angle and flexibility. To improve the efficiency and accuracy during extraction of trees, the three-level frame and the RD-schematic algorithm are proposed to fully employ the morphological characteristics of canopy surface models from coarse to fine resolutions. Methods of radius thresholding and symmetry judgement are also combined to remove disturbing objects, for example buildings and poles. Experiments based on the real-surveyed point clouds measured by the Sensei MLS system basically validate the proposed work, and the attributions introduced in this article serve as the fundamental procedures for further tree-relevant applications in regard to MLS.
Acknowledgements
The authors thank the Academy of Finland for financial support (Projects ‘Towards Improved Characterization of Map Objects’ and ‘Economy and Technology of a Global Peer Produced 3D Geographical Information System in the Built Environment’) and the Finnish Funding Agency for Technology and Innovation (Project ‘Development of Automatic, Detailed 3D Model Algorithms for Forests and Built Environment’). Thanks also to the anonymous reviewers for their instructive comments and suggestions.