Dynamic graph computing: A method of finding companion vehicles from traffic streaming data

Abstract

For intelligent transportation research, the detection of companion vehicle patterns aids in excavating behavioral relation between vehicles. This paper builds a dynamic license plate corpus based on streaming automatic number plate recognition data and combines the monitoring cameras character division to mine companion vehicle groups in real time. First, we establish the traffic flow graph based on vehicle trajectory data, and the improved PageRank algorithm is used to obtain the influence of the monitoring cameras. Second, the time-sliding window mechanism in Spark Streaming is applied to the streaming data, and the creation and updating of vehicle dynamic license plate corpus are completed according to the driving trajectory data. Finally, the plate-number dynamic graph computing algorithm is proposed to establish the dynamic relation graph between vehicles based on dynamic corpus and character identification of monitoring cameras. The camera characters are used as the influencing factor to correlate with the graph formed by vehicle nodes. The companion groups are obtained by real-time calculation of the trim and weight in the graph. Experimental analysis on the real license plate recognition dataset shows that the proposed model can effectively reduce the complexity of data processing and can detect companion vehicle groups in real time.