With laser-scanners and cameras mounted on cars and/or drones, it is possible to reconstruct automatically and quickly a 3D textured mesh of a city. While such models admittedly look great, in practice they cannot be used in several applications because the urban features (eg buildings, trees, bridges, etc.) are not available, we only have one gigantic mesh of triangles with textures.
The aim of the project is to develop a robust and automatic approach for generating semantic building models in boundary representation from textured meshes. The project is funded by the Dutch company CycloMedia and by an EU research project involving several companies. The research output of this project will be used to support traffic, lighting, and safety applications.