Abstract—With the development of economy, transportation is becoming more and more important, so a new method to control traffic is needed. Intelligent traffic system emerges as the time require. In this intelligent system, it is the basic that acquire traffic information instant. We used lots of way to get that information, such as induction coil, ultrasonic devices, microwave devices etc,. These tradition ways performed not as except. The cars on the road may block each other in different lane, so the result from the tradition devices may incorrect. Improving the tradition ways, we choose to use computer vision to statistic traffic flow. We use Hough Transform and Gaussian Mixture Model to detect road and cars. The whole system can be achieved on OMAP4460. We test the whole system on a real road, we captured all the cars and according to the test the road is detected accurately. The system performance is as we except.

Index Terms—Accurate, hough, real-time.

Liu Qiushi, Fu Ziyi, and Sun Yihang are with the Beijing Information Science & Technology University, Beijing, 100020 China (e-mail: lqs1990@ gmail.com, 465625854@qq.com, syh199011140 @qq.com).
Ray C. C. Cheung is with the City University of Hong Kong, Hong Kong (e-mail: cheung@ieee.org).

Cite:Liu Qiushi, Ray C. C. Cheung, Fu Ziyi, and Sun Yihang, "Traffic Flow Statistic Based on Computer Vision," International Journal of Computer and Communication Engineering vol. 2, no. 3, pp. 328-332, 2013.