Study Of Technology On Vehicle License Plate And Logo Recognition

In recent years, with the rapid development of scieence and technology and intelligent traffic system, the vehicle identification system has becoma a hot topic. The vehicle identification system has widely applied in the automatic charging, parking management, intelligent residential area construction, vehicles peccancy automatic recording, traffic accident treatment and other fields.Two key technologies are included in the vehicle identification system: vehicle-license plate recognition and vehicle-logo recognition, now the methods of license plate recognition increasingly sophisticated, but it's difficult to meet the video surveillance requirements of intelligent transportation system and intelligent community by a technical means. With the increasing number of vehicles, the traffic accidents, illegal escape, robbery of vehicles and other cases also increased significantly. The criminals often use the methods of replacing vehicle licenses, changing the appearance of the vehicle, it's increasing the difficulty of detecting and solving the case. As the result, the vehicle identification and vehicle logo recognition technology are developing.The vehicle-logo recoginition technique includes two key technologies:vehicle--Logo location and vehicle-logo recognition. As it is affected by natural background, vehicle background, light condition, weather condition and so on, vehicle-logo location is a challenging topic in the vehicle-logo recognition system.In this paper, attention is paid to the research on vehicle-license location, vehicle-logo location and vehicle-logo recognition. Firstly, proposed a vehicle-license location method based on the license's structure and the color feature from coarse to fine; secondly, proposed a vehicle logo location method based on the priori knowledge of vehicle license location and morphology algorithm from coarse to fine for the shortcomings of meeting the real-time fecture based on statistical learning method. The algorithm used the position relationship between vehicle license and vehicle logo to locate the vehicle logo for coarse firstly, then pinpointed the location of vehicle logo by edge dection of using the Sobel operator's veitical and horizontal template and morphology operation.Finally, extracted the feature of Hu invariants moment according to the vehicle logo image, and achieved the vehicle logo recognition by the application of the minimum classifier of Euclidean distiance. For the interference in the image after location and cut, used the local threshold Niblack binarization, then recognized the vehicle logo based on Hu invariants moment and Euclidean distiance. Through testing, it showed that this method has good precision and faster processing speed, the average time of vehicle logo recognition is 1.12 seconds and the recognition accuracy is 93.75%.