Clustering And Genetic Hybrid Optimization Research On License Plate Recognition Based On RBF Neural Network

Vehicle license plate recognition system is an important part of intelligent transportation system. At present, this technology has been widely used in car park management system, the vehicle management system of community, highway automatic toll system, intersection vehicle monitoring system and other fields. License plate recognition system is a comprehensive subject which involves image processing, neural network, pattern recognition and so on,the technology can greatly improve the management level and efficiency of vehicles, and can save a lot of human resources and material resources, so it has a wide range of research value and application prospectsThe license plate recognition system consists of three parts: license plate location, character segmentation and character recognition. Based on the theory of digital image processing and neural network, pretreatment of license plate image, license plate location, character segmentation, character feature extraction, RBF neural network character recognition process are studied in this paper and simulated in Matlab environment.In the part of license plate location,firstly, the mathematical morphology operations are applied to enhance the license plate image, and then the methord of edge detection and mathematical morphology are combined to locate the license plate, finally according to the license plate with a certain ratio of width and height to cut off false vehicle license plate area.In the part of character segmentation, firstly, the Radon transform is used for the license plate tilt correction. Then the license plate frame is removed based on the methord of gray level jump. Finally, the template matching method is used to find the interval between two and three characters in order to divide the license plate into two parts, then, the two parts of the license plate are segmented by the method of connected domain and template matching.In the part of character recognition, firstly the individual charactr is normalized.Then, histogram of oriented gradient features of gray character image is extracted. Finally the structure of RBF neural network is designed by the methord combining clustering and genetic algorithm, and the recognition of license plate characters is realized by this network.In the end, a license plate recognition GUI interface is designed by using MATLAB software. The experimental results show that the license plate recognition algorithm designed in this paper has high accuracy.