Silvilaser 2019 - Poster Presentations »
Terrestrial Point Cloud Processing to Estimate Individual Tree Parameters
In following abstract we will introduce our developed approach for individual tree parameters estimation from terrestrial point clouds but not restricted to. The whole workflow is fully implemented in DendroCloud software which is free and developed at Technical University in Zvolen (Slovakia). The main parameters estimated by DendroCloud are diameters of tree trunk at desired height, position of trunk, height, stem curve, biomass. By-products are digital surface model (DSM) and digital terrain model (DTM). The most important parameter estimated by our approach is a diameter of tree trunk, especially the diameter at breast height. Which is an important attribute in forestry. Multiple circle fitting approaches are used to estimate the trunk diameter. Three simple ones are minimum bounding box, centroid, maximum distance. Furthermore, two robust circle fitting methods are implemented: monte carlo and optimal circle. The performance of the methods was tested at plot with monoculture of European beech with size 50 × 50 m. We have used terrestrial laser scanner FARO 3D 120. The most accurate results were achieved by the robust methods and there was not significant difference between them when multi-scan method was used. In overall the best result was achieved by optimal circle method. The root mean square error of DBH estimation was in this case 0.77 cm . We were also part of “International benchmarking of terrestrial laser scanning approaches for forest inventories”  under International society for photogrammetry and remote sensing (ISPRS) where our approach provided similar results as approaches which used more complex workflows. The results from terrestrial photogrammetry are not as accurate as results achieved by terrestrial laser scanning (TLS). In our experiment we have achieved RMSE 4.41 cm  on plot with European beech with size 35 × 35 m. This is mainly caused by the data acquisition approach which provided point clouds with higher noise and lower density that the point clouds from TLS. The further research should focus on data acquisition which will provide point clouds with lower noise and higher density. DendroCloud is a free software available here: http://gis.tuzvo.sk/dendrocloud/ References 1.Koreň, M., et. al., (2017). Accuracy of tree diameter estimation from terrestrial laser scanning by circle-fitting methods. International Journal of Applied Earth Observation and Geoinformation, 63, 122-128. 2. Liang, X., et al., (2018). International benchmarking of terrestrial laser scanning approaches for forest inventories. ISPRS journal of photogrammetry and remote sensing 144 (2018): 137-179. 3. Mokroš, M., et. al., (2018). Evaluation of close-range photogrammetry image collection methods for estimating tree diameters. ISPRS International Journal of Geo-Information, 7(3), 93.