The Intelligent Traffic System (ITS) is the fundamental way to solve the traffic problem produced along with the economic development.The collection of traffic information is a basic problem in Intelligent Traffic. Traditionally,the data are usually measured by buried magnetic loops. But the loops can not meet with the requirement of the theory and practical application because of their limited scope. So the video-based method is presented in the thesis to solve the problem, which can supply with real-time data in the large-scale traffic scene. Video-based traffic information collection system can get a lot of information available, such as: traffic load,travel speed, lane occupancy, vehicle type, license plate number, traffic accidents, etc.. This paper focuses on the research of the collection of traffic flow. The main contents include: vehicle detection algorithm, vehicle counting algorithm, the camera calibration algorithm and vehicle queue length detection algorithm. With the research subject, the main contributions of the thesis include the following aspects: 1. In consideration of the lack of existing vehicle counting algorithms, we achieved a virtual loops based vehicle counting system。In variety of lighting and traffic conditions the vehicle counting system can count vehicles accurately. We exploited different counting methods at daytime and nighttime. In the the daytime we improved the existing algorithms to solve the missing counting for connected vehicles and repeated counting for congestion situation.In the night,the lighting condition is more complex , we achieved vehicle counting through detecting vehicle’s headlights.And a kNN classification was used to switch detection methods between daytime and nighttime. 2. For the lack of the existing vehicle queue length detection system, in this paper a complete vehicle queue length measurement system was achieved, consisted of vehicle queue detection and queue length measurement. Weexploited different detection methods at daytime and nighttime. At daytime, the queue detection algorithm consists of vehicle motion detection and vehicle presence detection, and a profile consisting of small profiles with variable sizes was used. In the night, a novel queue detection method through detecting vehicle’s headlights was achieved. And found a simple camera calibration method just using the geometry properties of road lane marking, then we could easily calculate the queue length. Experiment results indi...
修改评论