The Design of Embedded Video-Based Vehicle Tracking System

According to the need of distributed intelligent video surveillance in intelligent transportation system (ITS), a design based on S5PV210 microprocessor and embedded Linux operating system is proposed for real-time dynamic video target tracking and display. The scheme of the whole system design and the technical solutions of software and hardware are given. The system makes full use of the efficient video processing capability of S5PV210. The algorithm extracts moving vehicle foreground by Gaussian mixture model and background difference approach, if the vehicles are blocked, both the perspective characteristics and contour characteristics of the vehicles are used to segment the blocked foreground. Camshift and Kalman filtering algorithm are used in combination to track moving vehicles, the tracking result will be displayed on the LCD panel of the embedded system and uploaded to the information center. Details of the system realization are introduced in this paper.