Intelligent Detection Of License Plate-Related Illegal Vehicles Using Bayonet Images

Currently, with the accelerated process of urbanization, car ownership also greatly increased, as well as the license plate-related violations or brand-related illegal behaviors. As the "identity card" of the motor vehicle, license plate number recognition has been the foundation and key to the intelligent traffic. In order to avoid the tracking of various electronic devices, license plate obscured, defaced, painting, reflective even no-linked to the license plate, using fake license, clone vehicle and other types of violations involving license plate appears one after another. License plate-related offense has been willful violations, a blatant challenge to traffic safety laws and regulations, which has a significant harm.Detection of all kinds of involving licensing violations mainly through the traditional manual inventory or others report completed inefficient. In order to strengthen road traffic management and improve the intelligence level of traffic, the country is vigorously carried out the bayonet system construction. For example, Jiangsu Province, in December2009,"320" project is a road monitoring network covering the whole province, traffic bayonet laid-intensive for license plate-related offenses detected increases reliability, real-time, comprehensive data source.This study is based on traffic bayonet HD images, combine with image processing technology, image recognition technology and GIS, detecting illegal vehicles of all kinds of involving licensing intelligently. In this paper, the main work is as follows:(1) Classification and characteristics involving licensing violations. Summarize and analyze all kinds of involving licensing violations currently exist, analysis geometry and color characteristics of license plate, analysis features of illegal,fake and clone license plate.(2) Obtain basic vehicle information based on segmentation of vehicle structure interval. This paper makes use of linear characteristics of car’s front window and the roof to divide vehicle structure interval, proposing a vehicle basic information acquisition method of vehicle type, license plate number and body color based on vehicle structure interval, and integration access to basic vehicle information has being achieved.(3) Construct license plate-related illegal vehicle detection decision tree. Using a variety of data such as bayonet real-time monitoring information, basic information of passing vehicles, database of basic motor vehicle information, the "blacklist", road network of monitoring area, and so on, creates a decision tree to detect illegal vehicle involving license. Detecting illegal license plate using vertical projection of license plate area, detecting cars using fake license plate and cars clone license plate by legality, information consistency logical consistency and other rules.(4) Develop prototype system of license plate-related to illegal vehicle detection. The prototype system of license plate-related has been built and it is based on Open Source Computer Vision Library (OpenCV) and ArcEngine10.0, it verifies the theory and methods in this paper.