N. Regnauld1, A. Fechir2, F. Lecordix3, D. Rejkjaer4

1 - Ordance Survey, Southampton, Great Britain

2 - Institut Geographique National, Brussels, Belgium

3 - Institut Geographique National, Paris, France

4 - Kort og Matrikelstyrelsen, Copenhagen, Denmark



A growing number of National Mapping Agencies (NMA) have produced topographic vector databases that they want to exploit to generate a wide range of cartographic products at different scales. In order to be able to use these data as efficiently as possible, complex generalisation problems need to be resolved.

Important research efforts on generalisation have been made during the last 15 years and many papers have been published on the subject at different conferences, for instance at the International Cartographic Conference. Research on automatic generalisation started to deliver some promising solutions, and some commercial companies started proposing packages dedicated to automatic generalisation. For example, with the European project AGENT (1997-2000) managed by COGIT laboratory (IGN France), Laser-Scan proposed a first generalisation prototype (Barrault, 2001), with the module AGENT in their GIS LAMPS2.

This AGENT prototype was evaluated for use in production environment by four NMA: Institut Géographique National (France) (Jahard 2003), Kort og Matrikelstyrelsen (Denmark) (West-Nielsen, 2001), Institut Géographique National (Belgium) and Ordnance Survey (Great Britain). They concluded that the Agent technology was a promising solution to solve their problems in production, but required further  improvements to be easily usable in a production environment. The main requirements for improvement were on the core of the system, to make it more flexible and generic, on the interface, to be more user friendly and on improving the way the system can be extended (by means of customer-developed tools) to perform customer-specific tasks.

These four NMAs also noticed that they have very similar problems, data and objectives. Working in isolation meant that a lot of similar tools were developed by the different NMAs, and acquiring knowledge about automating the generalisation process was a slow learning curve repeated in each organisation. They have therefore decided to examine the possibility of collaborating on generalisation. A first collaboration was established to share with Laser-Scan the development cost of a new version of the AGENT prototype, called Clarity. This new version was available for the four partners in 2003.

The collaboration was pushed to a new level when the four NMA negotiated and signed in February 2006 an agreement for sharing software (custom developments) and for coordinating some developments on Clarity. The collaboration covered by this agreement does not involve any funding between partners.

This paper will detail this specific agreement between the MAGNET consortium members and the achievements of the different partners enabled or made easier by this collaboration.



Barrault, M., Regnault, N., Duchêne, C. Haire, K., Baeijs, C. Demazeau, Y., Hardy, P., Mackaness, W., Ruas, A., Weibel, R. (2001). Integrating Multi-Agent, Object-Oriented and Algorithmic Techniques For Improved Automated Map Generalization, Proc. Of the 20th International Cartographic Conference, vol. 3, Beijing, China, 2001, pp. 2110-2116.

Jahard, Y., Lemarié, C., & Lecordix, F. (2003): The implementation of New Technology to Automate Map Generalisation and Incremental Updatin Processes, Proceedings of the 21st International Cartographic Conference (ICC), Durban, South Africa.

West-Nielsen, P. (2001): Generalisation, the Danish Experience. Laser-Scan User Group, June 2001, Cambridge UK.