Based On Opencv Traffic Vehicle Speed Detection System

The goal of this paper is to establish a traffic speed system based on these digital cameras which are placed in the traffic road, thus providing data for the background traffic control system to improve the efficiency of transport vehicles to pass. This traffic speed detection system is divided into four sections as follows:background modeling, foreground detection, vehicle tracking and traffic speed calculations. The purpose of background modeling is to extract an image that does not contain any foreground objects from the input image. Foreground detection is to be able to detect vehicles out from the input image. The specific method is calculating the difference of input image and background image, and then extracting the foreground objects from the difference image. In this step, the method of region of interest is also used to increase the processing speed. Vehicle tracking is to match the objects which are detecting from current foreground image with previous foreground objects, thus achieving tracking of vehicles. Finally, traffic speed calculating is calculating the moving distance of history by a distance calculating model had been building, and then dividing the interval of the first frame and the last frame, to obtain the speed of each vehicle. The entire system is based on OpenCV (Open Source Computer Vision Library) and C++.In this paper three major innovations has been proposed:First, proposed an improved Kalman Filter model can solve accuracy of the original predicted value of Kalman Filter is not high which led to the failure of vehicle tracking; Second, proposed a quadrant-based vehicles occlusions elimination algorithm; Third, proposed a moving distance detecting model based on the lane standard. Because of our country have some lane length standards in different traffic road.