Moving Vehicle Detection And Violation Analysis Based On Surveillance Video

With the development of computer vision technology, intelligent video analysis as theirmain applications obtained more and more people’s attention, intelligent transportation systemsas one of the intelligent video analysis also has become a hot direction in the field. And the useof surveillance video to get the traffic parameters and determine whether the traffic participantsis illegally or not has become a research emphasis and difficultyies in the field of intelligenttransportation system. This article is based on monitoring the movement and tracking of thevehicle in the video to get the analysis of vehicle traffic incident and mainly detecting thevehicle speed and retrograde violations. It mainly uses the moving target detection and trackingalgorithms, use the extraction of the moving vehicle to calculate a reasonable representation ofthe state of moving object to determine moving vehicle traffic events and illegal behavior. Thisthesis mainly from the following several aspects to elaborate my ideas.First, this thesis introduces the commonly used moving object detection algorithm. Forbackground subtraction and background modeling algorithm adopted in this article, we have adetailed description and testing. While the background model obtained continuously updated tomake it more responsive to the changing background traffic video images. In addition tobinarization, filtering algorithms and morphological processing methods used in targetextraction here were tested.Secondly, on the target tracking phase, this thesis took the regional area-based Kalman filteralgorithm to track the vehicle. For the mutual integration of the vehicle, we took a different areaof the aspect ratio of the distinction between different vehicles. Such integration can achieveeffective separation of the target.Finally, in the case of illegal conduct analysis, we first need to get a description of thecharacteristics of the movement of the vehicle. In the main experiment, selected speed centroidand corner two features were carried out the speed of the vehicle. During centroid based on thevehicle speed, the data set taken trajectory centroid algorithm is analyzed by calculating thelength of passing through of the centroid time of the ROI, the calculation speed of the vehicle.In speed measurement based on the corner, the opposite corner detected in successive matchedframes, get in the corner of the frame adjacent the running time and distance, calculate the speedof a single corner, and finally calculate all the angles point instead of the average speed of the ROI by the vehicle speed. In retrograde detected by analyzing changes in the position of thecenter of mass of the vehicle information, get driving direction of the vehicle, therebydetermining whether the vehicle belongs to the reverse traveling violation. Throughexperimental verification, the experimental results of this show that the algorithm can detectvehicles speed and the vehicles retrograde effectively.