顾及几何特征和拓扑连续性的道路网匹配方法

已有的道路网匹配方法需要一定的人为干预和控制，仍无法实现匹配自动化，而且对于具有一定尺度差异的数据而言，受不相关道路对象的影响，匹配结果中会存在较多的误匹配。为此，本文提出一种顾及几何特征和拓扑连续性的由粗匹配到精匹配的分层匹配策略。粗匹配阶段，利用短边中位数Hausdorff距离计算匹配对象的相似度，再以道路网弧段端点的近邻分析结果作为匹配阈值，判断匹配对象是否为同名实体；精匹配阶段，将粗匹配阶段评价结果中离散的同名实体构建为完整的道路Stroke，即以追踪的方式剔除误匹配和添加漏匹配。利用宿城区不同尺度的道路网对所提算法进行验证，结果表明，该算法具有较好的匹配效果，能够实现匹配自动化。

Matching method for road networks considering geometric features and topological continuity

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.