Road Network Conflation: An Iterative Hierarchical Approach

Abstract

Road Network Conflation is concerned with the unique identification of geographical entities across different road networks. These entities range from elemental structures such as crossings represented by nodes in the network to aggregated high-level entities such as topological edges or sequences of edges. Based on topological, geometrical and semantic information, the road networks to be conflated are investigated in order to identify similarities as well as differences. In this paper, we introduce a novel approach for conflating road networks of digital vector maps which iteratively employs multiple matching steps on different hierarchies of structures in order to progressively find, evaluate and refine possible solutions by recognizing and exploiting topological and geometrical relationships. The introduced algorithms are applied to real-world maps and validated against ground truth data retrieved from visual inspection. Validation shows that our approach leads to good results exhibiting high success rates in rural regions and provides a reasonable starting point for further refining in dense urban areas, where special heuristics are required in order to tackle difficult matching cases.