A Vehicle Path Tracking System With Cooperative Recognition of License Plates and Traffic Network Big Data

As the problem of urban transport is becoming serious, intelligent transportation systems have received great attention. In this paper, we present a system of vehicle route tracking. By using traffic network big data, our system is capable of cooperatively identifying the license plates of moving vehicles from videos recorded by multi real-time surveillance cameras, and then mining their driving routes. In other words, the system can infer the driving routes of vehicles from the videos of multi-peer high-definition cameras with crossroads' geographic data of urban traffic networks. For this system, we propose an improved algorithm for recognizing license plates and an algorithm for route tracking by means of fuzzy matching. The experiments have been conducted to validate our algorithms by comparing with others. The accuracy of moving vehicle extraction has been improved to 97.4% by using the improved visual background extractor (ViBe) algorithm. The accuracy of vehicle license plates detection is 98.3%, and the accuracy and time of average character recognition and time are 97.8% and 34.2ms. The results of our experiment have demonstrated the effectiveness of the algorithms in our system.