The implementation can be divided into several steps:
- Ray casting (C++)
- Comparison of point clouds and city models (FME)
- Projecting textures to models (FME)
- Deriving probability scores (FME + R)
- Geometry and semantic modeling (FME)
full overview diagram - pending
For the in-depth understanding do not hesitate to check out the paper:
@article{WysockiRefiningByUnderpasses,
title = {Refinement of semantic 3D building models by reconstructing underpasses from MLS point clouds},
journal = {International Journal of Applied Earth Observation and Geoinformation},
volume = {111},
pages = {102841},
year = {2022},
issn = {1569-8432},
doi = {https://doi.org/10.1016/j.jag.2022.102841},
url = {https://www.sciencedirect.com/science/article/pii/S1569843222000437},
author = {Olaf Wysocki and Ludwig Hoegner and Uwe Stilla},
keywords = {MLS point clouds, Building reconstruction, Semantic 3D building models, Underpasses, Buildings refinement, Bayesian networks, Uncertainty},
}
Should you have any further questions do not hesitate to ask me: olaf.wysocki@tum.de