CMRT13 Keynote

Nicolas Paparoditis

Automated 3D data collection for 3D City Modeling from a NMA's point of view: some recent results, remaining locks and trends

We will show some recent results from the MATIS laboratory of IGN (the french national mapping agency) on automated data collection for 3D city modelling from different platforms (aerial, ground-based mobile mapping systems) and sensors (multi-spectral, lidar, hyperspectral, etc.). We will focus not only on 3D building reconstruction, but also on the 3D mapping of vegetation, roads, pavements, and land-occupation themes.

We will show that 3D models are on the one hand not always necessary, image-based solutions can be alternatively used for visualisation and to provide manual or semi-automatic 3D plotting tools and services to extract objects and reconstruct models on demand rather than on shelf. On the other hand 3D models when you have them are very useful to ease 3D interaction and navigation tasks and of course offer very interesting simulation possibilities.

We will also address some underlying issues for the generation and updating of 3D high quality city models such as relative registration between data sets for change detection purposes and registration of data on 2D or 3D existing databases for interoperability/integration purposes or to ensure an absolute sub-decimeter localisation accuracy necessary for autonomous navigation or topographic applications.

We will also show the limits of current methods and tools, and we will present some open issues (e.g. algorithm competition and self-diagnosis, multi-sensor data analysis, bigdata processing, etc.).