Abschlussarbeiten
3D Modeling and Visualization of the Gomantong Cave System (Borneo) Using Combined Data of Terrestrial Laserscanning and UAV Imagery (DTM)
Art der Abschlussarbeit
Diplomarbeit
Autoren
- Hautz, Stefan
Betreuer
- Prof. Dr.phil.habil. Manfred Buchroithner
Weitere Betreuer
Dr. Bernd Hetze, Zentrum für Informationsdienste und Hochleistungsrechnen der TU Dresden
Abstract
The main objective of this diploma thesis was to show an appropriate way of processing raw point cloud data of arncomplex cave system to a comprehensive 3D model, which can be used for further scientific analyses as well asrnfor visualization purposes. There are multiple approaches for each of the necessary steps, depending on thernprovided software packages. After experimenting which programs are compatible to each other and whichrnprocessing methods are most constructive, a full workflow of the Gomantong project has been given.rnFor the successful creation of a 3D point cloud model, including the step of point cloud registration, FAROrnSCENE has been used. The resulting model consists of 5.3 billion points. It is possible to apply bat and bird nestrncount algorithms, which are based on the analysis of the scanner taken intensity values. Another possibility is tornestimate the cave's surface area and volume with high accurateness.rnFurthermore the point cloud model has been converted to a meshed 3D geometry model and integrated with arndigital elevation model of the surface of Gomantong Hill at unprecedented precision. For the creation of thernconsistent polygon mesh, the necessary steps of sub-sampling, noise reduction and hole filling were performedrnwith the provided software Geomagic Studio. Another option, giving even more features, would be the use ofrnPolyWorks IMInspect software. Geometries like handrails have been reconstructed semi-automatically withrnCINEMA 4D. For creating a 3D surface model, different approaches of processing aerial taken data have beenrntested and compared. Finally an elevation grid has been used. A combination of the cave model and the terrainrnmodel based on GPS references could not be performed due to incorrect references. Thus the datasets have beenrnadjusted manually to give a visual match.rnThe presented method for texturing the cave model with scanner taken photos and colored point clouds givesrnpleasant results but is still improvable. Especially the unattractive mixing of different color values needs to bernfixed. A possible solution might be the development of an algorithm balancing the color values.rnThe final combined 3D model can be used for different scientific and public purposes. Regarding the visualizationrnthere are several possibilities. First very accurate maps of the cave can be derived. This responsibility has beenrnaccepted by several international expedition members. Furthermore this diploma thesis results in a digital virtualrnfly trough. To reach this, the cave model has been rendered with CINEMA 4D. The video editing has been donernwith Pinnacle Studio giving an animation with a transition effect between terrain and cave data. To use the fullrnpotential of the 3D model and modern display techniques, the data has been prepared for stereoscopicrnvisualization. It can be applied to any softcopy non-autostereoscopic method, for instance back and frontrnprojections, as well as autostereoscopic multi view displays (five view directions). Theoretically the model couldrnalso be used in an immersive virtual reality environment (CAVE) allowing an individual cave expedition. Alsornaugmented reality applications can be derived. Finally, the point cloud data has been released using the FAROrnWebShare assistant. With this the scan files can be viewed online and it is possible to measure distances and areasrnin the cave model. Besides this there is also the possibility for a hardcopy visualization in form of laser subsurfacernengraving.rnSecond, further ecologic studies and statistics can be applied and visualized directly on the model. The advantagernof the subsampled mesh is that it is still high accurate but not as heavy and difficult to process like the full pointrncloud data. The consistence of the model, meaning a closed surface without any holes, enables hydrologicalrnanalyses like flow behavior simulations. In combination with the distribution data of birds and bats also areas withrnhigh erosion potential could be predicted.rnRegarding the precision, a second scan of the Gomantong Caves is not required in near future. The processing ofrnpoint clouds is bottlenecked by hardware and software limits. There is no need for scanning with higherrnresolutions because the complete number of points can not be processed into a mesh geometry yet. Nevertheless, arnsecond laser scan expedition would be necessary to cover the complete Gomantong cave system. The expeditionrnin 2012 only covered 30% of the area of Simud Putih, so there is still potential for further scanning research.
Zugeordnete Forschungsschwerpunkte
- Echt-3D-Visualisierung (raumbezogener Daten)
- Geoinformationssysteme für Umweltmonitoring
Schlagwörter
3D Modeling, Gomantong cave system, Borneo, Laserscanning
Berichtsjahr
2013