The Research On Moving Vehicle Detection And Tracking Technology In Video Surveillance

Intelligent Transportation System (Intelligent Transportation System, ITS) is becoming the development direction of traffic system nowadays. How to use the video surveillance technology to realize effective detection and tracking for moving vehicles has become a research hotspot and difficult problem. This is the premise of judging vehicle behavior and solving the problem of road congestion. This paper mainly studies moving vehicle detection and tracking algorithms based on video surveillance from a single fixed camera. It also improves the original algorithm and completes the corresponding application programming.In the practical intelligent video surveillance, whether real-time computer processing can be achieved is an important indicator of an algorithm’s practical value. This paper carries out in-depth study on the detecting and tracking algorithm for moving vehicles from the perspective of practical application. The main research work is divided on the following aspects:(1) ViBe algorithm is selected as the moving vehicle detecting algorithm adopted in this paper based on the analysis and comparison of several different detecting algorithms for moving target. As regard to the characteristic that the ViBe algorithm can generate ghost easily, the paper improves the ViBe algorithm, and the experimental results show that the improved algorithm can remove the ghost more quickly.(2)Aiming at eliminating the influence of vehicle’s shadow on target detection, the author puts forward an algorithm, in which the threshold can be adaptively updated in gray color space, to remove shadows. The method is verified by experiments, and the result shows that the method is able to remove the shadow effectively.(3)Through studying the general targets tracking methods, the author gets the vehicle tracking algorithm with real-time and accurate tracking capability. After improvement, the author finally developed a limited-window vehicle tracking algorithm which combines Camshift with Kalman filter. The experiment demonstrates this algorithm can achieve good results.(4) Programming to achieve the multi-objective moving vehicle detection and tracking. Combining with the OpenCV, this paper uses the trigger mechanism to realize the vehicle counting and multiple target tracking. Moreover, it achieves vehicle marking and automatic drawing vehicle moving trajectory.