Context and objectives
The company TELE ATLAS produces and maintains geographic databases in vector format for road networks in several European countries, and is one of the market leaders in this field in Europe. The data for the databases are derived from large scale topographic maps, which involves a number of problems (lack of up-to-date maps, copyright fees). TELE ATLAS is therefore looking for alternative data sources. The present quality of the commercially available satellite images allows extraction of 3 to 5 road-classes. TELE ATLAS hopes to use the very high resolution images (VHR, resolution of 1 to 5 m) in the near future to extract all 8 road-classes.
In this view, it is advisable to start extending the in-house developed software with specific modules for image processing (geometric correction, contrast enhancement and change detection). The actual application will depend on the prices for VHR satellite imagery demanded by the distributors.
The technical and economic potential of KOSMOS images with a resolution of 2 m was evaluated in a TELSAT feasibility study. The results indicate that KOSMOS images perform well in comparison with topographic maps and orthophotoplans.
Expected scientific results
Algorithms and programmes for :
- geometric correction;
- contrast enhancement of satellite images for visual interpretation;
- change detection, allowing extraction of the linear elements in a semi-automatic way.
At the University of Liège (ULg), a test bank of very high resolution images (KVR, IRS-C, SPOT-5 simulation and others) was compiled in order to test the performance of newly developed software programmes.
Specific constraints of a production environment : the algorithms should be fast, efficient and simple to implement (not always the most accurate algorithm).