Ordnance Survey – Great Britain's national mapping agency
![Ordnance Survey home page [logo]](/images/gui/logo-os.gif){kind=logo}
Generalisation
Automating and customising the generalisation process, which derives a small-scale map from large-scale data.
Main objective
We are working on a long-term plan to provide the business with a strategy and the appropriate tools to build on-demand generalisation applications, running as automatically as possible.
Challenges
The process of generalisation lacks formalisation. We need to be able to deduce from the specifications of the target product which transformations need to be applied to the initial data. We are currently working on:
- the algorithms needed to perform the transformations required;
- the rules that link data and specifications to the triggering of the algorithms; and
- the specification themselves: they must be designed to allow the user to select the product they want, and to provide the application with all the information required by the generalisation rules.
Strategy
In order to achieve the vision, we are building a development environment that can gradually be enriched to produce more and more complete generalisation solutions. We have chosen to base our developments on Clarity, which is a platform dedicated to generalisation developed by Laser-Scan. Clarity has been built on the knowledge acquired during the European project AGENT, which was itself an attempt to model a generic system to perform different generalisations depending on the user requirements (map specifications). One of the challenges of AGENT was to put together results of years of research in the field of generalisation around the world. Its main originality was in modelling the decisional entities of the system as agents (software agents) that have some level of autonomy to generalise the features under their control.
We are helping Laser-Scan to develop Clarity, and we are working on developing the tools to customise the system for performing the generalisation we want. This includes spatial analysis tools (like triangulation), generalisation algorithms, evaluation measures, rules for triggering algorithms depending on the context, and setting up the behaviour of different geographic entities (like cities, roads or rivers).
Future impact
The results of this research should gradually improve the cost efficiency of the current Ordnance Survey map production lines and ultimately allow the company to produce custom maps to respond to specific user needs.
Publications
REGNAULD, N., 2003. Algorithms for the Amalgamation of Topographic Data. Proceedings of the 21st International Cartographic Conference, Durban, South Africa.