关键词: 道路网，结构模式，形态相似性，数据匹配

Abstract: This paper presents an approach to match nodes from heterogeneous road networks in the identical urban area. The heterogeneous road networks are often produced at different levels-of-detail (LoD) and coordinate systems, which leads to difficulties for matching and integrating between them. Firstly, dual carriageways and complex road junctions are identified as road patterns to eliminate the LoD effect of different datasets. Then the block patterns around nodes in heterogeneous road networks are extracted to make nodes comparable. The comparison is thereby made between two local networks derived from compared nodes. The original local networks are rotated and scaled according to each possible pattern correspondence, and the optimal one is measured with the maximum similarity value. After the pattern matching, two compared heterogeneous networks are approximately registered into the similar coordinate system, and the remaining nodes will be matched with geometrical and topological methods. An experiment of matching heterogeneous road networks is carried out without any other semantic information. The result shows the approach can get satisfactory nodes matching from heterogeneous road networks.

Key words: Road networks, Structure pattern, shape similarity, data matching