Research On Traffic Vehicle Flow Based On Video Detection Technology

In recently years, more and more researchers are attracted the study about traffic information parameter based on video which in intelligent transportation system. This study lets the traffic department know the traffic operation more intuitively and clearly. At present, the methods of detecting traffic information parameter are various. But the vehicle video detection are developing quickly and used widely which based on image processing technology. And this method has got more advantage than before. So it has been a important research object in ITS.Traffic parameter detection system based on the video has so many advantages such as intuition, easy installation and low cost. As a result it stands for a new direction of vehicle detection system. The vehicle is driven along the roadway at the crossing of city traffic. And the vehicle detection is based on analyzing and disposing the traffic picture of each roadway. Therefore this paper detects the traffic flow parameters through setting virtual detection region and extracting the picture of detection region.The paper takes a video of city traffic crossing road as the research object. Firstly, extract the background model from the video image, then get the move target by image differences. At the same time we can have the image pretreatment just like binarization and getting rid of noise. And then we can achieve the vehicle count of virtual detection region. As a result, we get the vehicle flow information.In this paper, it mainly includes the following points:(1) Studied and analyzed the methods of the traffic vehicle flow in recent video detection technology. And studied the several algorithm of vehicle object detection.(2) Analyzed several frequently-used background extract algorithm. We finally choose the averaging method to extract the background. And we have made some improvement on the basis of this.(3) Realized the recognition and extract for the moving object in the traffic video.(4) In view of the specific traffic video, we can set up reasonable virtual detection area, and research basing on the change rule of characteristics of the image pixels in area of study. Then we can put forward the vehicle flow detection algorithm. Finally, the intersection traffic calculation will be realized.(5) According to the vehicle flow information at each road, we had a dynamic regulation for the traffic lights recent allocation.Finally, for the algorithm used and involved in this article, we mainly design and implement by using matlab on the PC. Luckily, we obtain a satisfactory results.