License Plate Recognition System Used In Highway Toll Station

Aiming at the demand of the intellectualized traffic control for roadway toll station, this article carry out applied research on the system of license plate recognition. Based on the relevant domestic and overseas research results, profits from the image recognition knowledge and engineering methods, and take full of the ready-made facilities, through gathering and instant processing the video images, then achieve the recognition. The system provides important data and evidence for traffic management departments. Regarding the actual situations of toll station within Mianyang segment for the road between Mianyang and Jiuzhaigou, it realizes the practical application of license plate recognition at the road toll station according administrative requirements.The general system design use the embedded software processing mode, and take the auditting host computer which is for the original security system as the platform, send the pictures gathered from the camera to the platform to process, the recognition result will be delivered to back modules to carry on processing and the application.For the part of gathering images, Pre-burying the double induction coil be selected to be main triggering way, video mobile detection is a supplementary. We select the high definite and width dynamic and bright light inhibit product in the front segment. It carry on the snapshot for images with the call-back functions of the H series of video gathering card from Haikang, and completes the image gathering function.The recognition core is selected by PlateDSP embedded product from Shenzhen Pulide Co. , it process the images from the camera in the front segment by call-back function which got from OCX controls.We use Access2000 to build the database. The system connects with the controlling center. It realizes the applied processing, and completes the functions such as contrast, recording, statistics, inquiry, printing and network transmission and so on.After the practical application, the success ratio for capturing vehicles surpasses 99%, the accuracy ratio for recognition is more than 90%, the statistic error for traffic flow is lower than 5%, completely achieve the actual demand for users.