Detection And Tracking Of Freeway Video Surveillance System Vehicles

The development of expressway has very important value and significance for urban traffic,The emergence of highways increased traffic flow,Improve driving efficiency,Reduce the cost of transportation to improve the economic and social benefits of transportation and reduce energy consumption.Everything has pros and cons,Due to its high speed and its high speed,large traffic characteristics make traffic accidents occur from time to time.Therefore,to avoid the occurrence of traffic accidents,maximize the benefits of freeways,Intelligent Traffic Management system came into being.Intelligent transportation system is a complex,multi-functional integrated system,and the intelligent video monitoring system has a pivotal position.The main theme of this paper is the intelligent expressway video monitoring system.Research on the moving target in video monitoring system,For video,the vehicle target needs to be processed through its detection,shadow elimination,and target tracking.This article includes the following sections:1.For moving objects detection and background modeling select background difference method and mixed Gauss background modeling method respectively.Background difference method is simple,the algorithm design is relatively easy,for real-time image processing has a very good effect.Excellent background model for the background differential effect has been greatly improved,General binarization target segmentation only need to set a threshold value of the entire image screen,this threshold can be either fixed or variable,However,in practice,it is found that how to adjust the threshold can not achieve the desired result.In fact,each part of the screen in the segmentation threshold are not the same,therefore,there is a flaw in the segmentation method of the whole image unified threshold,Partition-based algorithm can be a good solution to the problem,Gaussian mixture model is exactly this algorithm.Resulting in elimination with the shadow area,2.An improved algorithm based on traditional target shadow elimination is given.The elimination of the shadow of the moving target can improve the accuracy of the target detection of the moving vehicle and achieve better results.This article summarizes the HSV color space RGB color space advantages over the premise,For traditional HSV colors,the color characteristics of the HSV interior are too similar to those of the shaded area,By adding image area information to improve the algorithm,The improved algorithm makes up for the deficiencies of the previous algorithm,the detection accuracy is improved.3.Solve the multi-target occlusion vehicle target tracking method.Using combining the Kalman predictor with the target geometric features to deal with the multi-vehicle target tracking problem,by comprehensively applying geometric features such as centroid and area of the target to deal with occlusion of multi-targets,Occurrence of occlusion will lead to changes in the target number,Summarize the number of forecast target and the actual number of targets,Find the internal relations of different situations,Group discussions to develop specific tracking strategies.