The Research Of The Eliminate Shadow And The Tracking Of Multi-objective In Video Speed Measuring System

The situation of today's traffic is getting more and more complicated and management is increasingly complex. Along with the rapid development of the computer technology and the picture processing technology, using real-time video image to the traffic management has become an important part in ITS.This paper focuses on the video speed system of vehicles target detection, eliminate of shadow and the problem of targets merger in multi_objective tracking.For the detection of vehicles target, this paper based on the methods of the gaussian mixture modeling for background extraction, and do improvement in set and update the model.For the eliminate of shadow, this paper considered not only all the geometric model of vehicle shadow in the highway but also the characteristics of shadow pixel to proposed a eliminate shadow algorithm which is based on the conversion of lane line coordinate system.For the problem of targets merger, this paper proposed two solutions. The one is based on the Camshift algorithm and combined with the kalman filter algorithm to deal with the problem.The other one is through the analysis of the geometric model of merge target, and considered the histogram of the number of pixels and the result of the calman filter's forecast to deal with the problem. Because of the first kind of solution's calculation is larger, this paper use the second kind of solution as the priority way and the first kind of solution as the service way in the algorithm which is proposed for solve the problem of targets merger.After the detail alalysis and a large number of experiments to several important problems of the video speed system, finally put forward a complete algorithm of multi target tracking and speed calculation scheme, and based on the system design, ultimately the computer platform to achieve a set of video measuring system.