The Measurement Of Vehicle Speed And Vehicle Violation Detection In Tunnels Based On Machine Vision

In recent years,with the rapid development of the national economy,our country in the construction of the tunnel has made far-reaching achievements,which are more representative is the highest altitude permafrost tunnel in the world Fenghuoshan Tunnel of Qinghai Tibet Railway.Although Chinese has become the world’s fastest-growing tunnel and underground engineering construction market,but with the tunnel traffic safety problem to still can not be ignored.Unlike the general opening section,the tunnel is a special space structure,which has the characteristics of closure property,no accessibility,unknown,difficult contact and rescue.these characteristics easily lead to frequent traffic accidents in tunnels.And the reduce of tunnel traffic accident rate to a large extent depends on the driver’s standard driving behavior.Therefore,this paper proposes and studies the task of measurement of vehicle speed and vehicle violation detection in tunnels based on machine vision.First of all,the design of the whole system of this subject is introduced in detail.We analyze the composition and function of each module of the system,expound the key and the difficulty of the project and the feasibility.Secondly,it needs to be emphasized the rapid movement of vehicles in the camera will lead the collected video image is blurred.In this paper,the Wiener filter based motion blurred image restoration algorithm and the algorithm is based on the projection onto convex sets(POCS super-resolution reconstruction algorithm,and compare the two recovery results.In this paper,we use the Wiener filter based motion blurred image restoration algorithm and the projection onto convex sets algorithm,and compare the two recovery results.Once more,we introduce the license plate recognition module of this topic in detail.This paper employs three frame difference method for the detection of moving vehicles.if the vehicle is detected,then we conduct the subsequent license plate recognition.The license plate recognition includes license plate location,character segmentation and character recognition.This topic can achieve the coarse position of the license plate through the HSV color model and a priori knowledge of license plate,fine segmentation of the license plate according to the projection of the coarse positioning to the license plate region,each character of the vehicle license plate image through the projection method and the single plate character recognition by BP neural network method.Finally,we analyze the implementation of vehicle speed and violation detection and the design system of the database.This project creates a SQL Server 2008 R2 database for the detection of illegal vehicle in tunnels,and two data sheets be used to store the license plate information of the tunnel entrance and exit license plate information.Judging illegal vehicles and measuring vehicle speed,we just call the corresponding data in the table and through certain rules and standards to calculate the average speed in the interval tunnel and determine which car in tunnels occurred violations.Experimental results show that this license plate identification precision is high.And under the different environment in the tunnel,through the license plate recognition we can calculate the vehicle average speed in the tunnel and determine vehicle in the tunnel is illegal.