Research On Video Vehicle Tracking And Traffic Flow Length Detection Algorithm

Vehicle detection is an important aspect of Intelligent Traffic Systems(ITS),which is the basis for all other technologies to achieve.If the vehicles cannot be accurately detected,other kinds of treatment outcome is difficult to be guaranteed.Vehicle queue length is also one of the most important parameters in Intelligent Traffic Systems.Vehicle detection and vehicle queue length detection have been studied in this thesis,the main work is as follows:In order to coping with the condition of high traffic density,we bring in the Adaboost learning algorithm with Haar-like features and propose some afterwards process.Our experiment results show that this algorithm has an advantage over moving object detecting algorithm obviously.As there is much less feature of car at night compared to it is in daytime,we propose a night car detecting algorithm that focuses on Hough circle detecting.Our experiment results show that it is a fast and usable algorithm when facing the night car detecting problem except that the traffic density is too high.After these works,we propose a tracking algorithm named 4-states tracking algorithm based on Kalman filter.This algorithm not only links up the car information between different frames,but solves some problems in car detecting part like false hits and miss hits.At last,we propose a video-based method to detect the length of the queue.The method,which is based on vehicle detection and motion detection in video sequences acquired by a stationary camera,has received good results.This method firstly applys sobel edge-detection operator to detect the presence of vehicles,then use the robust Kanade-Lucas-Tomasi feature tracker to detect and track features.Experiments prove that this motion detection method is precise and reliable.