Video-based Traffic Monitoring System for Traffic Flow Micro Data Collection

摘要

This paper proposes a system architecture for a video-based traffic monitoring system to collect traffic flow micro data. The proposed architecture was implemented through a software platform called "VideoTrack". The VideoTrack software platform was developed in Visual C++ 6.0 with OpenCV. After initialization, system frame calibrated the traffic scene. The coordinate transformation was done using the direct linear transformation algorithm. The feature point optical flow tracking algorithm was used to track the motion objects at the traffic scene.. The collected output data such as speed, acceleration and the trajectories were analyzed to validate the statistical significance. The experiment results have indicated that the coordinate transformation algorithm is valid, and the object tracking performance is relatively accurate and the proposed system architecture is stable and ready to be implemented in real world situations