On Video Processing Based Traffic Flow Tracking Statistics Technology

With the rapid development of social economy,urbanization and the growing population,the increasing demand for efficient transportation results the intelligent system to be new future research direction.As an important part of intelligent transportation system,vehicle traffic monitoring system has received a wider range of attention in terms of urban road traffic information collection.The real-time collection of traffic flow plays an important role in reducing urban traffic congestion,improving transportation efficiency and rationally allocating road resources.Based on the traditional vehicle detection and tracking algorithm,this paper proposes an improved motion vehicle detection algorithm to solve the problems of the existing algorithms.Meanwhile,this paper propose a vehicle tracking and counting algorithm based on Euclidean distance matching and Hash algorithm.The main work of this paper is presented below.In the aspect of moving vehicle detection,we compare the characteristics of various moving target detection algorithms.Based on the background difference algorithm,we propose an improved Surendra background updating algorithm.The improved Surendra algorithm combined with the Otsu algorithm based on least squares can adaptively update the background image;the counter added to the background to solve the stagnation of moving target in the video;the improved Surendra algorithm can extract the moving target more accurately.In the aspect of tracking and counting of moving vehicles,we analyze the current mainstream moving target tracking algorithm.Combined with the actual characteristics of vehicle videos,we propose a vehicle tracking algorithm based on Euclidean distance matching and hash algorithm.Considering the problem that the speed of the vehicle is too fast in the actual scene,which will lead to matching failure,we combined with the hash algorithm to perform secondary detection to avoid the repeated counting problem caused by the tracking failure being tracked again and the vehicle miss detection problem.The experimental results prove the effectiveness of the secondary detection,and finally we combine the results of vehicle tracking with the virtual detection line to carry out multi-lane counting of the vehicle.In the design of traffic flow statistics system,combined with the basic requirements of the traffic flow statistics system,the prototype of the traffic flow statistics system was designed by integrating the paper's algorithm.The system mainly includes three modules;vehicle detection,vehicle tracking and vehicle counting.We test the designed vehicle traffic tracking statistical system algorithm by test videos.The experimental results show that the proposed scheme is feasible,real-time and accurate.