关键词: 道路网, 地图综合, Delaunay三角网, Stroke特征

Abstract: In road feature generalization, the selection has to consider the grade in semantic description, the length shape in geometric characteristics, the distribution density and the accessibility in topological properties. The traditional method evaluates the road entity importance taking into account just one aspect, for example the segment length, node joint or mesh density, and then completes the selection process by ordering the importance “qualify”. Due to the lack of effective approach to take into account the spatial distribution information of road network, the method is not able to inherit the original spatial distribution characteristics. This paper presents a new method trying to keep the original spatial distribution characteristics. Firstly, concatenating the road segments into strokes and evaluating the importance of every stroke; secondly, selecting roads by the importance of stroke, but adding two types of constraint conditions to the above process, including class constraint conditions and spatial proximity constraint conditions. This proposed approach considers topological, geometric, semantic and the spatial distribution information of road network, and the experiments show that this method is convincing.

Key words: road network, map generalization, Delaunay triangulation, stroke properties