Automatic dendrometry: Tree detection, tree height and diameter estimation using terrestrial laser scanning

This study presents an automatic method to identify tree stems, and estimate tree heights and diameters from terrestrial laser scanning (TLS) data. The method is based on the isolation and vertical continuity of the stems. First, a height-normalized version of the point cloud is created. From this, stems are individualized, an iterative process is applied to the points at breast height for estimating diameters, and tree heights are calculated after denoising and clustering the points of each tree. The method was tested in three different sites. All the elements detected as trees were actual trees, and more than 99% of the trees in the plots were detected. Root mean square error (RMSE) of the estimated diameters at breast height (DBH) ranged from 0.8 to 1.3 cm in the test plots, and total tree height (TH) RMSE ranged from 0.3 to 0.7 m. In the cases studied, the algorithm showed robustness to the presence of steep or irregular terrain, the presence of low vegetation and artifacts at breast height, the indistinct use of individual or multiple scans, and tree density in the plot.