The Research Of Video-based Moving Vehicle Detection And Tracking

Video-based detection and tracking of highway vehicles has a crucial role in intelligent transportation systems.The detection and tracking of the vehicle is to identify and extract the moving vehicle in the video,and then track the trajectory of the vehicle.Researching better detection and tracking algorithms can not only improve the efficiency and accuracy of detection,but also improve the efficiency of video analysis,and can also prepare for follow-up research.This article first introduces the experimental environment in detail,extracts frames from existing highway video,and prepares for the detection and tracking of moving vehicles.Based on the analysis of the detection algorithms of the moving vehicle,the paper focuses on the frame difference method and background difference method.The method of combination of frame difference method and background difference method is used.The frame difference method can make up for the background difference method sensitive to changes in light intensity.The disadvantages will not affect the shadow problem.The background difference rule can make up for the problem that the frame difference method can not completely extract the relevant points of the target image,and dynamically update the background so as to realize the extraction of the color static background of the multi-adjacent frame fusion.For the problem of incomplete extraction of vehicles similar to the color of the road surface,the complete segmentation detection of the vehicle can be achieved through regional growth.Because the shadow of the vehicle itself will affect the experimental results,this article eliminates the shadow of the vehicle’s own shadow;according to the difference of the shadow and the road surface,the gray value is not large,the vehicle area texture is much richer than the shadow area,and the vehicle area gradient value is also much larger The characteristics of the shadow area can be achieved by multi-scale wavelet transform to remove shadows.After classifying the tracking algorithm of the moving vehicle,for the tracking algorithm of the vehicle,this paper uses a matching method based on the spatial movement characteristics of the vehicle.In each frame of the video,the vehicle’s most complex frontal area feature basically does not occur.Change,gradient transform the obtained difference image,and then use multi-scale dynamic wavelet transform method to locate the complex area of the front of the vehicle,determine the direction and distance of vehicletravel in two adjacent frames to obtain each frame in each After the position coordinates of the car,a coordinate series is obtained,the position coordinates in each frame experienced by the car are recorded,and the vehicle running trajectory is drawn.This article is based on the study of moving vehicles on highways.Under Windows operating system,C++ and OpenCV programming are used to detect and track the moving vehicles.The captured images of freeway vehicles are tested.The results show that this method has better performance,High accuracy and real-time performance.