Traffic Flow Detection Based On Video

This paper first introduce the importance and significance of traffic flow detection in ITS, then compare video detection technology with other detection technology, and propose superiority of video detection technology. Vehicle detection method based video and traffic flow detection system based fictitious loop are studied in details. Considering the characters of the moving vehicles image, the method of background subtraction is mostly studied to pick up the vehicle. A method of background estimate based on zone_distribution is adopt. A reliable update method is brought forward and the threshold confirmation in the division method is studied and compared, the Otsu method with better result is choosed. The after_treatment methods which are suitable for follow-up judge. By adopting a FNCC (Fast Normalized Cross_Correlation) algorithm in this paper, moving shadows were directly detected from grayscale image. A video traffic flow detection method are put forward and realized through programming with Visual C++, and we have tested about a large number of image arrays under the environment of real road. The result is good.