Stem Modeling From Terrestrial Laser Scanning Using the Block Minimum Approach

Abstract

Terrestrial laser scanner is widely used in forestry surveys to map and estimate tree dimensions. The main problem is still how to reconstruct the tree geometry from the laser scanner point cloud. This study introduces an approach to model individual stems in automatic manner from terrestrial laser scanning data. Unlike the traditional approaches, the proposed algorithm does not assume that the stem is circular and therefore adapts better to the point cloud, leading a more fl exible estimate of the geometry along the stem. The idea is to transform the point cloud to a linear problem and to apply fi ltering algorithms, similar to those used for DTM extraction in airborne laser scanning. The minimum-block algorithm was tested on laser scanning data collected in a pine forest in South Brazil.