| The multi-scale geospatial database series construction and updating task has become increasingly arduous. To meet the demand of real-time updates of the current geographic information data, map generalization, which is the core of modern cartography and map production, information extraction, visualization, and the important means of geo-spatial scale analysis, has become a key issue to be solved. Since the existing single-element-oriented or single-method-relied map generalization mechanisms cannot lead to a comprehensive solution to solve automated map generalization thoroughly. It is time to take multi-elements into account, to consider the interactions and restrictions between the multi-elements, and to establish a multi-elements-oriented mechanism.In this study, the roads and the residential areas which are closely related and significant on the map are chosen as the research object. The collaborative relations and the process of the map generalization between this pair of elements are discussed. The collaborative relations are divided into three types:the topological relations, the proximal relations and the structural relations are studied. Computational geometry data structures such as constrained Delaunay triangulation and its dual Voronoi diagram are used to construct the collaborative relations between the two elements. The construction of the collaborative relations is the main task of cooperative automated map analysis. The collaborative relations play the role of spatial constraints during the cooperative map generalization process. The road is picked as the controlling element which launches the generalization process. The cooperative generalization process of roads and settlements consists of two cooperative generalization models, one is the road selection and settlement aggregation model, and the other is the road simplification and settlement displacement model. In the first model, the result of road selection becomes the spatial boundaries of settlements to avoid over-aggregation and combination mistakes. For the requirement of map clarity, a polygon simplification is carried out progressively for the combined settlements. The simplification is based on a concave structure recognition algorithm. In the second model, for the conflicts which are caused by the road curves simplification between roads and settlements, the roads are set as fixed objects, and the settlements are displaced relatively to their neighboring roads. The settlements are classified into four types according to their nearby roads, each corresponding to different types of displacement operations. Finally, experiments are performed. The results validates the above multi-elements-oriented map generalization method is reasonable and effective. |
