University of Natural Resources and Life Sciences, Vienna (BOKU) - Research portal
Gewählte Master / Diploma Thesis:
Sarah Witzmann
(2022):
Development and evaluation of algorithms for the automatic marker-free registration of forest point clouds obtained from Personal Laser Scanning.
Master / Diploma Thesis - Institut für Waldwachstum (WAFO),
BOKU-Universität für Bodenkultur,
pp 29.
UB BOKU
obvsg
FullText
- Abstract:
- The usage of data obtained from Personal Laser Scanning (PLS) for forest inventory purposes has increasingly gained recognition in the past few years. The reasons for the growing popularity of PLS in forestry are not far to seek: Laser scanning technology allows for an acquisition of highly precise individual tree and stand information in a relatively short amount of time. Nevertheless, the high labor cost efficiency of Personal Laser Scanning in forest inventory and monitoring has one limitation which is not to be underestimated when “wall-to-wall” data, as opposed to forest inventory data with realtively small sample plots, is desired: larger areas (>1-3 ha) cannot be scanned at one go, due to the scanning time limitation associated with software and data processing restrictions. Therefore, multiple scans must be conducted and co-registered to obtain one continuous point cloud. One way of doing this is to use easily recognizable artificial reference markers, like white spheres placed on tripods. These markers can afterwards be detected by semi-automatic software and used for the referencing of spatially adjacent scans. However, the transportation and deployment of such markers constitute great logistic and occupational efforts. Taking into account the additional time needed for the preprocessing and the matching of the point clouds, this semi-automatic registration can be considered as bottleneck for the otherwise high efficiency of PLS-based forest inventory and monitoring on larger areas. The goal of this master thesis was to develop and evaluate an algorithm for the automatic and marker-free registration of point clouds to eliminate this bottleneck and to pave the way for a more practical and time-efficient usage of PLS on larger areas in the future. To achieve this goal, a registration algorithm based on the tree locations identified within the point clouds was adapted and improved, enabling a fast and reliable coarse registration of forest point clouds.
- Beurteilende*r: Nothdurft Arne
- 1.Mitwirkender: Ritter Tim
- 2.Mitwirkender: Gollob Christoph