Vehicle classification system based on dynamic Bayesian network

Abstract

Vehicle classification system is an important part of intelligent transportation system (ITS), which can provide us the necessary information for autonomous navigation, toll systems, surveillance and security systems, and transport planning. In this paper, we introduce a vehicle classification system based on dynamic Bayesian network (DBN). Three main types of features are employed in our system: the geometrical characteristic of the vehicle, the location and shape of license plate, and the vehicle pose. Firstly, vehicle detection and tracking method are used to locate the vehicle. Then, we extract the features from video sequences. Gaussian Mixture Model (GMM) is used to construct the probability distribution of the feature. Finally, we classify a vehicle into one of four classes: sedan, bus, microbus, and unknown. The experiment shows the proposed method can achieve classification exactly and credibly.