A video-based traffic monitoring system based on the novel gradient-edge and detection window techniques

This paper presents the novel algorithms for a video-based traffic monitoring system. In the video-based traffic monitoring system, the vehicle detection and traffic parameter extraction are the main processes of the traffic monitoring system. The vehicles will be extracted in the first process, and the result will be fed as an input to the next process in order to evaluate the traffic information. The proposed vehicle detection technique employs the gradient-based adaptive threshold estimation (GATE) to estimate the proposed threshold values, which are automatically adapted to the changes of various illumination conditions throughout the day. In addition, the horizontal moving edge detection (HMED) is proposed for traffic parameter extraction. The experimental results indicate that the proposed GATE algorithm yield the proper threshold value for each illuminated condition in the vehicle detection process while the proposed HMED algorithm can effectively resolve the uneven edge density of vehicles.