Research Of License Plate Character Automatic Recognition

Recently, with the rapid development of economy and technology of our country, the traffic is more and more convenient. The level of lives is improved, and the number of owning vehicles has increased. With the development of intelligent transportation system, traffic is also developed. The highway is the tie of economy development, charge for expressways are inevitable. Intelligent transportation system and expressway is complementary to each other, promoting mutually, develop mutually. Automatic recognition system of license plate character is very important in the intelligent transportation system. Intelligent transportation system has widely and profound development prospects in collecting information and monitoring illegal vehicle, tracking vehicle tracking etc.. Intelligent transportation system is very hot in recent years. For every vehicle, the license plate is the unique and unified mark. So the core of intelligent transportation system is the technology of automatic recognition of license plate characters. Character recognition technology is a hot research topic in recent years, and having practical and profound significance of research. The process of automatic license plate character recognition is:firstly, pretreating the vehicle image, automatically positioning to license plate of the vehicle. Secondly, cut apart the license plate character into single character. Lastly, using the BP neural network methods distinguish the license plate character. The automatic recognition system of license plate character installed in the expressway has considered the surrounding environment. And the automatic recognition system can weaken the influence of surroundings, can recognition license plate fast and accurately.This paper write with the basis of study the advanced and representative technology of vehicle license plate automatic recognition at home and abroad. And this paper research and design the automatic recognition system of license plate character, combining the reality of our country. The process of the license plate character recognition includes image preprocessing, license plate locating, character division dividing and recognizing. In the Image pretreatment, introduce the problems of image acquisition by the collection methods, equipment, the position of equipment. Firstly, after getting information of the vehicle license plate image, transform the color image into gray image with the methods of Maximum value, average or weighted average. Then, Smoothing the image by the way of median filtering. Then, we can get the Image pretreatment after detecting edge of image, and the method of two values. In the aspect of license plate location, this paper introduces the Color image positioning method that includes the methods based on gray image, color image and features, and methods of geometry and neural network systems. In the license plate character segmentation module, this article mainly aims at the problems that often occur, such as, the license plate tilt, characters and the characters adheres, character and frame, character adheres omission., and Put forward the method of that problems. We can use the methods combine horizontal projection and vertical projection to removal the influence of rivets and frame, and division characters. In the aspect of character recognition, this paper mainly used the BP neural network. First of all, normalize the segmented characters. Then, determine the number of neurons in the three layer neural network according to the previous experience and experiment of the senior. At last, establish the neural network system. We should train the BP neural network enough, and repeat and comparison the result, Improve defect Combined with the actual. After that we can get stable BP neural network. The results of experiment show that the method mentioned in the paper is feasible. The system can accurately and timely identify the license plate, and run well.