Video-Based Traffic Flow Monitoring Algorithm

Abstract

Intelligent traffic is the main trend of urban development at present. In view of the increasing number of vehicles and traffic congestion, this paper presents a video-based traffic flow monitoring algorithm. The proposed algorithm extracts the vehicles using characteristics of the video, detects the moving vehicles and the non-motor vehicles by the pixel sizes and positions. According to the results of vehicle detection, the number of vehicles is analyzed, speed is calculated by moving distance, and combined with traffic flow to judge current road conditions are smooth, general, or congested. This algorithm adopts the three-frame difference method to detect moving vehicles to ensure the real-time performance of the algorithm and displays the number of vehicles, speed information, the number of vehicles per lane in a direct visual way, and vehicles entering the region of interest can be identified. The experiment results show the proposed algorithm can judge the current traffic situations in real-time based on video, and vehicle detection is more accurate.