Dijkstra-Based Selection for Parallel Multi-lanes Map-Matching and an Actual Path Tagging System

摘要

Map-matching between a road network and a raw GPS trajectory must be done in order to analyze the urban traffic computing. A weight-based map-matching algorithm has proposed some important features to solve this problem, such as perpendicular distance between a raw GPS point and a road segment, bearing difference and connectivity. However, the connectivity of a map-matching problem becomes complex when the raw trajectory traveled a parallel multi-lanes road network segments, even humans will have difficulty selecting the correct road segment. To solve this problem, a dijkstra-based selection map-matching (DBSMM) algorithm is asserted by us. Candidate segment set formation, dijkstra-based selection and a friendly driver tagging system are presented in this paper. With the driver-tagged actual paths of our tagging system, it is possible to evaluate the DBSMM algorithm. Therefore, the precise map-matched network traffic data can be the basis for more further traffic researches.