Research And Simulation On The License Plate Recognition Algorithm Based On Improved Neural Network

With the rapid development of economic environment, the intelligenttransportation system is facing a great challenge. In intelligent transportation, thelicense plate recognition system also need to continuously improve. Today, weestablish perfect road network traffic and intensify efforts to introduce intelligenttransportation system that are to solve the traffic problem, compared to buildmore infrastructure road. With the continuous development of intelligenttransportation, license plate recognition system has become a popular topic.In this paper, the license plate recognition is divided into four parts, imagepreprocessing, license plate orientation, license plate segmentation and characterrecognition. In license plate preprocessing part, license plate image is into a grayimage. Further processing the image using histogram equalization technology,finally using median filtering noise reduction, make image can improve thequality, for the next step in the identification system. In license plate locatingsection, comprehensive analysis of all kinds of localization algorithm andresearch, finally choose the template matching algorithm for license platelocalization, in order to further improve the efficiency, is the improvedtechnology, accelerate the matching method of template matching to locatelicense plate. In character segmentation part, first of all is to locate image binaryprocessing, and then use the corrosion and further processing is done in theexpansion technology, using the vertical projection of the license plate imageletters, Numbers, divided into single characters and normalized processing. Andit is an important work in the early period of the character recognition. Incharacter recognition, the first is to carry on comparative analysis to variouskinds of recognition algorithm, especially for the BP neural network is introduced in detail, this article uses the improved BP neural network learningalgorithm for license plate recognition, and compared with standard BP neuralnetwork, to the advantage of the improved BP neural network, and efficiency.In this article, also at various stages of the license plate recognition to do thecorresponding simulation, the simulation to have further understanding of licenseplate recognition technology, image processing techniques are used to get thefinal experimental results with the experimental data.