Hi Peda, yes indeed, yours is a very pertinent question. For the problem of differentiating between low-detail complex buildings and high-detail simple buildings currently I have two ideas. One, check the spatial distribution of the level of detail. This is based on the assumption that buildings in a spatial cluster are mapped in similar level of detail. I'm aware that there are several exceptions to this that might make my assumption invalid. Secondly, use an authorative data set to compare the detail of the footprints. Again, assuming that the authorative data set is consistent in its level of detail.
I admit that I was not aware of your second point, so thank you for bringing it to my attention. Which also helps me understand the limitations of such a method to enrich the OSM database. I will still work with OSM data in my thesis, however my project for GSoC 2017 if off the table. In general, my work is focused on the statistical comparison of footprints or 3D buildings by using a machine learning approach. However I'm still in the middle of developing the method, and ultimately I would like to know that until what extent is the automatic inference of LOD possible. Thank you very much for your feedback! Balázs -- Balázs Dukai MSc Geomatics engineer candidate | MEng Landscape architect The Hague, Netherlands LinkedIn: www.linkedin.com/in/balazsdukai _______________________________________________ dev mailing list [email protected] https://lists.openstreetmap.org/listinfo/dev
