Funding Reference FCT - POCTI/AUR/48123/2002
Dates 2005-05|2008-06

The main goals of this project are related to the semi-automatic acquisition and maintenance of 3D virtual reality models of urban areas. It is intended to use registered aerial images and low altitude laser range scans to acquire 3D data of city structure. This data will be processed in order to segment relevant structures for urban planning (buildings, roads, green areas, etc). Range information provides a very rich description of 3D structure but lack photometric information. Aerial photos provide this information, allowing to pre-segment regions based on color and texture. The main scientific innovation of this project is the combined use of 2D (aerial images) and 3D (range scans) to simplify and improve the building extraction process. Most current approaches use one or the other types of data exclusively. The final result provides a computer model which stands for a mix geometry-image database that can interface to GIS software available (at CML), as well as the generation of real-time walkthrough with thematic information. The results of this project are to be integrated on Lisbon City Hall public computational facilities.

Research Groups Computer and Robot Vision Lab (VisLab)
Project Partners IDMEC-IST, Câmara Municipal de Lisboa
ISR/IST Responsible
Alexandre Bernardino
João Paulo Costeira
José Gaspar