Abstract: The existing road network matching methods still need some human intervention and control, and do not achieve automatic matching. Moreover, for the data with certain scale differences, there may be some mismatches in the matching results, due to the influence of unrelated road objects. In view of the above problems, this paper proposes a hierarchical matching strategy from rough matching to fine matching considering geometric features and topological continuity. In the rough matching stage, the short side median Hausdorff distance is used to calculate the similarity of the matching objects, and then the nearest neighbor analysis results of the end points of the road network arc segment are used as the matching threshold to determine whether the matching object is the same name entity. In the fine matching stage, the discrete entities with the same name in the evaluation results of the rough matching stage are constructed as a complete road stroke, that is, the false matching is eliminated and the missing matching is added by tracking. The proposed algorithm is verified by the road networks with different scales in Sucheng district. Experimental results show that the algorithm has good matching effect, which is conducive to the realization of matching automation.

Key words: geometric features, topological continuity, road network matching, hierarchical matching strategy, nearest neighbor analysis