Road Network Construction and Hot Routes Inferring with Vehicular Trajectories

This work proposes a novel hot routes inferring approach without the support of real road network. Discovery of hot routes is important to the applications that requiring classifies the traffic flow or profiles the dynamic of the city, such as targeted advertising, traffic management and control. The advances of location-acquisition technologies have led to a huge collection of objects’ trajectories in the road network, which give us the chances to finding hot routes conveniently. However, it is difficult to effectively detect hot routes without the support of the available road map. To address this issue, we first develop a Road Network Constructing Algorithm (RNCA) that extract road network from vehicular trajectories using image processing methods, and then propose a Hot Route Inferring Algorithm (HRRA) based on the extracted road network. Meanwhile, a novel road matching operation is also developed to match trajectory points onto roads. We have conducted extensive experiments on real dataset of taxi trajectories. Simulation results show that the proposed RNCA and HRRA are both effective and efficient.