Research On Motion Target Detection And Tracking Algorithm In Intelligent Transportation

With the constantly expanding scale of modern transportation road network, it is not realistic that simply using the manual labour to monitor all points online no longer, so it has became the development trend of future traffic monitoring system by information acquisition, pattern recognition and information fusion technology for network intelligent regulation. As the background of intelligent transportation, the paper use computer vision technology to do research in target detection, tracking, and realize continued stability of detection and tracking multi-object under monitoring area. Paper mainly work can be summarized the following several aspects:(1) In the target detection algorithm, aiming at the situation that existing the hole in the foreground image by the three frame difference method,this paper proposes an improved algorithm combines the three frame difference and Surendra, Surendra algorithm using dynamic threshold method to update the background model for adapting to scene changes, it is greatly improve the effect of prospect target detection, and good real-time performance.(2) In the single target tracking algorithm, aiming at the situation that complex background disturbance or the target moving fastly in monitor video result in the problem of tracking failures, this paper puts forward an improved Meanshift target tracking algorithm based on the ORB feature matching. This algorithm can track the target accurately and improve the robustness of target tracking.(3) In the multi-target tracking, for multiple target tracking technology in traffic monitoring requirements of the real scene, this paper proposes a framework of multiple target tracking, including four modules which are target detection, target tracking, target correlation and update tracking list, detection and tracking, it is realize that automatic detection and tracking of the multiple target vehicles in the monitoring area. And in tracking module, this paper adopt an improved Meanshift target tracking algorithm based on color and texture feature fusion to achieve the tracking, it improved the accuracy of tracking.To the new borned target and problematic target,the proposed method can carry on the good processing,achieved sustained and stable tracking.The research work of this paper is aimed at complex background, multi-target interference based on closed color situation, and the problems of poor anti-interference detection and tracking, this paper mainly studies the algorithm of detection and tracking, and proposes the corresponding improvement method, through the experiment, to achieve the multi-target vehicles tracking steadily in traffic monitoring area, improve the accuracy and robustness of detection and tracking.