Geometric shape fitting on simulated and TLS-based leaning tree-trunk point cloud for precision forestry measurements

Abstract

Remote sensing and measurement methods have gained great importance in forest surveys, forest inventory, growth, and planning of assets in the last decade. In particular, it is important to obtain geometric parameters from dense point cloud data and estimate the diameter at breast height (DBH), which is a common parameter in forest inventory. In this way, "precision forestry" measurements come to the fore and leave traditional measurement methods behind. However, these processes are rather tedious and more complex than one might think. The accuracy of geometric estimations depends on the application of convenient methods. In estimating tree trunk diameters, 2D planar calculation of the diameter determinations of leaning trees is an important source of error. In order to eliminate this error, it is planned to prevent it with the robust principal components analysis (PCA) algorithm. For this purpose, the proposed methodology has been tested on simulation and real test datasets. The results show that the application of robust PCA algorithms prevents significant errors.