Research And Application Of Image Recognition Technology In The ITS Of The License Plate Recognition

With the rapid development of social economy and transportation technology, Under the drive of economic development, city traffic faces tremendous pressure, people greatly concern about this issue. If we don’t deal with this problem, it will affect people’s travel and traffic of the city. Therefore, it is becoming a trend to development intelligent and informationization traffic management system.In recent years, many researchers have pay attention to how to better intelligent recognize the license plate. Because each car has its unique plate, so, the management of the vehicle can be equated to the management of vehicle plate, and this work can only be realized by intelligent license plate manage system. Therefore, we need this kind of system intelligent identification the license plate.So far, some of the methods are mainly based on the following aspects, For example, edge detection, texture analysis, the color information of the license plate, digital morphology, neural networks, etc. The license inclination issue can be corrected by the following method: Hough method, Based on the texture feature, etc. Character segmentation has the following methods: projection, the method of SVM clustering, region growing, etc. In the field of character recognition, it exist the following methods: Use of the neural network classification method, the method of template and based on the characteristics of the character classification, etc. However, in practice, some of the above methods are not very good adaptability.This paper references the previous studies of this kind of problem and analyzes the advantages and disadvantages of various methods, a part of methods are improved. For example, while positioning plate, in addition to using the inherent characteristics of license plate and the projection characteristics method, also using the Sobel y direction of the edge detection, the characteristics of wave and level difference method. While adjusting license plate, combining with the Hough method and texture features of license plate, so that we can handle plate polluted situation. While segmenting the license plate, in addition to using the inherent prior knowledge, also using the projection method and connected domain features of the characters. When identifying character, we use the BP(Back Propagation) method. In image preprocessing, using the image filtering and a binarization method based on the Otsu adaptive, etc.This paper first introduces the research background and the status, and explains the research content and the corresponding innovation. Secondly, introduces some related knowledge and ethics, For example, the image gray scale, digital morphology and adaptive control of the camera system, BP theory and the corresponding applications, etc. Then, we give a detailed analysis of the research contents of each module. Finally, this paper introduces the implementation of the system and analyzes the results; the experimental results show the effectiveness and accuracy of this method.