Collaborative Map Generalization Method Of Roads And Buildings Based On Multi-agent

Map generalization has always been one of problems in the research of geographical science over years, and so far no more relatively complete automatic generalization solution has been put forward yet. Map is a common carrier of multiple subject inforrmation, Chen Shupeng academician once raised. Map generalization is involved with various fields such as cartography, mathematics, graphic image, computation geometry, graph theory, pattern recognition, artificial intelligence and computer vision, to achieve it theoretical knowledge and technical methods of the fields must be comprehensively utilized. Aiming at realizing automatic generalization by computer, earlier scholars decomposed complex task into sub-task generalizations contrapose single map element and futhur resolve it into single target obect and even multi parts generalization. Took the maintaining of entire morphologies and cluster features of multi-object on map, researchers later put forward some methods from overall situation. Recently collaborative map generalization which intends to conduct multi-element generalization from the whole collaborative spatial relations of map features was proposed on the basis of traditional map generalization, and has become an emerging direction of map generalization research. As an important part of map data, roads and buildings on the large scale topographic map occupies a large proportion in both numerical and areal load, what’s more, roads and buildings are connected with each other no matter in their local shape or spatial distribution. Thereout, to study the collaborative generalization of roads and buildings is imperative.Multi-element collaborative generalization is more complex than traditional as that the generalization of singal map feature and between different features must maintain effective coordination and cooperation. In addition, roads and buildings constructed by human bings is so miscellaneous and diverse in their spatial pattern and distribution characteristics, differing with contours, rivers and other natural features which are relatively taken shape in natural causes and obviously show their spatial distribution pattern on maps. As a consequence, the key of generalizing roads and buildings collaboratively is to eliminate spatial conflict, maintain spatial relationship, and get commond of generalization process. And now studies of methods taking collaborative spatial relations into account to process spatial conflict between double map features mostly specific into some featured area type or distribution character, also no more comprehensive and effective solutions have turned out yet. But, it should be noted, agent technique is a new appropriate choise to solve the problem because of its autonomy, interaction, reasoning and decision-making that can simulate human thinking well. Agents with map generalization knowledges can accomplish missions act on their own which may improve the level of automation and intellectualization of map generalization a lot.This study implemented roads and buildings double-feature collaborative generalization by taking advantage of agenttechnology. To begin with, domestic and overseas theories and methods of research about roads and buildings generalization were investgated and summarized. Then classification of map generalization knowledge and rules was established named descriotive knowledge, propositional knowledge and procedural knowledge, also expression of them is discussed besed on further analysis. The next step in-depth research is the classification and model of agents for collaborative generalization of roads and buildings, what’s more, behavior reasoning of different map agents on the strength of knowledge and rules were considered, to manage spatial conflicts generated during the generalization interaction mechanism of agents was presented. In the end, prototype system for test of roads and buildings collaborative generalization besed on multi-agent was constructed and multi-group experiments were done. The results showed that spatial conflicts were avoided effectively and collaborative relationships of roads and buildings on map were preserved. The introduction and application of agent technique successfully fulfilled roads and buildings two-feature collaborative generalization.