Research Of Vehicle Speed Detection Algorithm Based On Video

In recent years, Intelligent transportation system(ITS) is developing rapidly, which is considered as the most effective method to solve problems of the urban traffic and The traffic information detection technology based on video, is a very important part in ITS. The vehicle speed detection has become a hot spot in the research of video traffic information detection technology. In this paper, we study two kinds of velocity measurement methods based on video and achieve a vehicle velocity measurement system.The main work of this paper is as follows:1) A motion vehicle detection algorithm based on background subtraction and frame difference is proposed. First, the pixel of the image is classified and the initial background is extracted using statistics median. Second, the background is updated in the change region, which is identified as background pixels using the frame difference method. Finally, the accurate foreground object is obtained by background subtraction and image post-processing. For shadow elimination algorithms, an improved detection method based on HSV color space and texture features is proposed.2) An improved vehicle tracking algorithm based on moving regions and similarity matching operator is proposed.The vehicle velocity can be estimated on the basis of the tracking. First, the target vehicle’s center of gravity position is forecasted by the Gray Prediction Model GM(1,1). Second, a method is proposed, which consider image euclidean distance as the similarity matching operator. Then, matching the vehicle regional characteristics by the operator, so that the tracking is completed.The experiment shows the availability of the algorithm.3) An improved vehicle speed detection algorithm based on virtual coil is designed. In this paper, a coordinate transformation model between image distance and the vehicle actual motion displacement is researched. By setting up two virtual coil in the detection zone, the moment that the moving vehicle pass through the coil can be obtained, and then vehicle motion speed is estimated. Experimental results shows the feaibility of the algorithm. 4) A vehicle speed detection system is implemented. The system includes video player module, background extraction and update module, target vehicle extraction module, tracking and matching module and real-time speed estimation module.The paper has studied the vehicle detection and tracking algorithms. The vehicle measurement algorithms based on the target tracking and virtual coil are proposed. A vehicle measurement system is implemented. Experimental results show that the system can realize the vehicle speed measurement.