Research On Video-based Detection, Recognition And Tracking Method Of Vehicle In Intelligent Transportation System

With the development of the society and technology, the detection and trackingalgorithm of video motion objects is one of core topics in the field of computer vision. It’salso the key intelligent technology of video surveillance system. In many areas, it is theapplication of wide range of video surveillance system to monitor and keep track ofmoving targets needed. In this thesis, in the background of intelligent transportation videosurveillance application, it analyzes and researches the methods for the moving vehicletarget detection and tracking. First, in the moving vehicle target detection study, in order tosolve the video moving vehicle target detection accuracy and shadow interference, itsimply analyzes three typical background model and has proposed a Gaussian mixturemodel based on moving target detection method in the multiple moving targets. Thismethod adopts Gaussian mixture shadow model method and HSV color model to suppressshadow of moving vehicle. Secondly, in the research of the moving vehicle tracking, itanalyzes different models and current research focus on the tracking problem. Combinedwith the existing algorithm, the tracking algorithm based on Meanshift improvements isproposed. This algorithm is mainly refer to solve the tracking target color interference,blobcking and other issues. The thesis finally used OpenCV software to program theintelligent video vehicle detection and tracking algorithms and analyze the results of theprogram realization. And in the vehicle tracking, software simulation comparison betweenthe traditional meanshift tracking algorithm and improved camshift algorithm proves thesuperiority of the improved algorithm.