Um algoritmo para filtragem do tronco em nuvem de pontos laser terrestre de árvores de Pinus spp

Abstract

The terrestrial laser scanner has been tested and pointed as non-destructive alternative for measuring trees. Dendrometric variables can be obtained with threedimensional modeling techniques. However, data filtering is necessary to eliminate points that do not represent the surface to the trunk. This paper aimed at presenting a proposal of an algorithm, based on methods already established in the literature for automatic filtering of point clouds of tree trunk. The study was conducted from twelve trees inserted in a stand of Pinus spp with different ages. Three laser stations were used for surveying each tree. The proposed algorithm, called Maximum Distance Filter (Dmax Filter), filters iteratively the trunk by sections of the point cloud. For the best automation performance, we chose a parametric algorithm. The Dmax Filter was experimentally tested and the results showed that the surfaces of the trunks were effectively filtered up to approximately two thirds of the total height of the trees studied. The high density of leaves prevented the surface representation of the trunks in the upper third of the trees. The application of Dmax Filter did not change the geometric characteristics of the tree trunk and provided data to the threedimensional modeling.