Agent-based Generalization Algorithm

In this paper, we apply the theory of "Nature objects existed being excellent and washed out being of feebleness" to automated map generalization basing on thinking of the conditions of Map Use, Characters of Map Regions and Map Scale. Two generalization models which are called "Constrained Delaunay Triangle Net Generalization Model(CDTNGM)" and "Circle Generalization Model(CGM)" are created. During the process of integrating these models, techniques of Agent are adopted to monitor and control generalization user's operation. The main subjects of this paper are as follows:(1) The conditions of Map Use, Characters of Map Regions and Map Scale are considered during map generalization. We also build several tools to dynamic sum up and pack up generalization rules and parameters and put them to generalization knowledge library.(2) Imitating the process of evolvement of nature to generalize digital maps using the theory of "Nature objects existed being excellent and washed out being of feebleness".(3) Two generalization models are created, the names are called "CDTNGM" and "CGM". "CDTNGM" can operate map objects optionally and easily through its great capacities of analysis and dealing with relations. "CGM" operators map objects relying on the characters of circle.(4)Agent model is briefly introduced in paragraph five. The generalization system monitors user's operation and records user's action information to operation log library using Agent techniques. Computers can also judge the users whether they are operating correct or wrong basing on log library.(5) We pack up the generalization models using the technique of Microsoft COM. In order to manager geo-spatial data,- ORACLE database is used. And finally we integrate these distributed programs to the intact system using the language of visual C++6.0.