Closed Ylannl closed 4 years ago
I think there is a large overlap with https://github.com/tudelft3d/underground-buildings-filtering/issues/3 . Would it solve the issue if the underground part was cut off from the footprint before going into geoflow?
But in case we do an underground-part-filtering prior geoflow, probably we won't identify all the cases (but most I think), so you still need to handle these somehow. Or they will simply end up in the bunch of reconstructed buildings that are "problematic".
Which makes me wonder if https://github.com/tudelft3d/underground-buildings-filtering/issues/3 is redundant.
Or, since we do the whole underground building classification anyways, maybe better to keep this task there, so you don't need to deal with it in the reconstruction?
In the classification we label these partially underground buildings, would the labeling help?
I believe the method I propose above gives a more elegant and robust solution than cutting or replacing the footprint prior to reconstruction. It will be more data-driven and less dependent on errors in the BAG/BGT polygons. That is if it all works as well as I expect it to...
I think I would first try to simply perform this for all the buildings. In that case no prior labelling is needed. In fact it would be easy to label the building if a part of the footprint was cut off.
Labelling the floating buildings would still be useful however. I don't see how I can do that in the reconstruction method.
This should be fixed now with 4842e5b
This will fix seemingly arbitrary planes getting assigned to:
Idea for implementation: