Integrating Multi-Source POIs and Road Networks Based on Geometric Data

Abstract

With the popularity of mobile positioning devices, large numbers of multi-source Points of Interest (POI) and road data are increasingly collected resulting in a crucial problem when integrating these heterogeneous data. The paper thus proposes a geometric-based approach for integrating multi-source urban POI and road networks. The proposed method firstly extracts the linear features from POI data according to a DBSCAN-based linear clustering algorithm and then generates a graph structure, named POI Connectivity Graph (PCGraph). The matching nodes between PCGraph and road network are then selected by the probabilistic relaxation framework and then refined by the vector median filter to align POI data and road networks geometrically. The experiment shows that the linear extraction algorithm efficiently identifies the inexplicit structural patterns of POI data and the probabilistic relaxation matching approach correctly finds the corresponding points to efficiently accomplish the position adjustment of POI and road data.