Detection And Tracking Of Moving Vehicle And Detection Of Specific Events

With the rapid urbanization, all countries are vigorously building high-speed transportation network. The scale of traffic management system becomes increasingly large than the original system. Therefore, governments and academics are attaching more importance to intelligent traffic management system for coordinating transportation, allocating resources, and alleviating the waste of human resources. Meanwhile, the intelligent traffic management system based on machine vision has become a hotspot.In the intelligent traffic management system based on machine vision technology, detection and tracking of vehicles from video surveillance is the basic functional module. This paper mainly discusses several key technologies about the detection and tracking of vehicles in traffic video, the main research contents include: (1) In the part of detection of vehicles, this paper analyses the basic algorithm of the vehicle detection and presents an adaptive algorithm called background updated difference combined the three frame difference method. This method segments the moving vehicles by background subtraction and uses three frame difference to improve accuracy of the detection of vehicles. Experiment result shows that this method can accurately detect moving vehicles; (2) In the section of tracking of the vehicles, the paper analyzes two solutions commonly used in the tracking of vehicles, namely, data-driven solution and the best guess solution based on Bayesian framework, then considering the moving distance of vehicles in two adjacent frames is very small and area change in the image is not too large, this paper improves the CAMShift algorithm, which is based on the data-driven solution, to achieve the tracking of moving vehicles .The improved algorithm has fewer parameters and a small amount of data, the experiment result shows that the improved algorithm has a good real-time and accuracy than the original algorithm; (3) The paper implements the incident detection of the moving vehicles based on tracking of the vehicles, including detection of vehicles reverse driving, moving into a specific area, and pressing lines; (4) Based on Intel's open source computer vision library OpenCV, this paper implements algorithm previously discussed, and experiment results show that the proposed algorithm in this paper has a good performance for detection and tracking of the vehicles.