An electronic police system with multiple vehicle parts model

Abstract

With the development of urbanization, vehicle violations bring lots of problems for urban traffic. In this paper, we implement an electronic police system based on multiple salient vehicle parts for traffic surveillance. Vehicle is represented by its salient parts and its trajectory is obtained by tracking based on Kalman filter. First of all, multiple salient vehicle parts including the license plate and rear-lamps are localized using their distinctive features. Then vehicle tracking is performed using these parts with a Kalman filter to get vehicle motion trajectories. At last, traffic violations are detected by analyzing the vehicle trajectories and configuring various detection regions. Experiments show that our system is effective and it can achieve real-time performance for real traffic applications.