An algorithm for robust tree detection in ground-based point clouds based on classical mechanics

Rubén Manso

doi:10.1016/j.compag.2024.109750

An algorithm for robust tree detection in ground-based point clouds based on classical mechanics

2025-02-01

0

PRE

AI

DOI：10.1016/j.compag.2024.109750

Original

Original request for help

Save

Abstract

En 中文

Tree detection in complex terrestrial point clouds, such as those produced by SLAM systems in environments of high branchiness, is often suboptimal when using geometrical approaches. An algorithm for robust detection in these situations is presented. The method is based on the assumption that applying Newton's law of universal gravitation and Newton's second law of motion to a horizontal point cloud slice, where each point is given a mass, will cause the points that belong to each tree to collapse towards the geometrical centre of their parent tree. To test this approach, three case studies of British-grown Sitka spruce were used, and the results were compared with those of a well-established detection algorithm. Our findings show that the method presented here matched or outperformed this latter method in terms of completeness, correctness and F-score (e.g. Fscore 0.74 vs 0.48, 0.98 vs 0.95 and 0.96 vs 0.92, respectively, for each site). A validation exercise with an independent dataset was also conducted. This confirmed the robustness of our approach but showed that the settings of our method may need to be adjusted according to the device's acquisition rate. On top of its usefulness in complex point clouds, our algorithm fundamentally differs from others of more geometrical nature and could therefore offer slightly different solutions to detection problems. These differing results could potentially be combined for more efficient global tree detection.

Keywords:

TLS

SLAM

Sitka spruce

Branchiness

An algorithm for robust tree detection in ground-based point clouds based on classical mechanics

Abstract

Journal

Organization