Design Of Fuzzy License Plate Recognition System Based On Complex Environment

In recent years,with the continuous progress of science and technology,license plate recognition technology has been constantly improved,and the application in the field of intelligent transportation is becoming more and more widely.At present,the license plate recognition technology is mainly used in the "electronic eye" system of traffic intersection,the ETC detection of the expressway,the speed management system and the parking management system in the community.The purpose of the license plate recognition system is to supervise the vehicles intelligently,so as to ensure the safety of public transport and bring convenience to the development of the country.However,with the complexity of the urban traffic environment,the performance requirements of the license plate recognition system are also increasing.In complex environment,such as fog,rain and snow weather,cloudy day or night and so on,the vehicle license plate information captured by the camera is blurred,which will directly lead to the difficulty of vehicle information location and the low recognition accuracy.Therefore,how to correctly identify the fuzzy license plate captured by the equipment in the complex environment can improve the accuracy of the system recognition and meet the requirement of real time.All these problems are the research hot spots in recent years.There are still many problems to be further developed for the design of the license plate recognition system in complex environment.Dig and perfect.The fuzzy license plate recognition system for complex environment is designed in this paper,which includes four links: license plate image enhancement,preprocessing,license plate location and license plate character recognition,in which the license plate image enhancement is mainly enhanced by the improved multiscale Retinex(Multi-Scale-Retinex,MSR)algorithm of the image captured by the device.The image quality is further improved and the processing time is reduced by gray and two value.The Adaboost algorithm is used to extract the Haar features of the license plateand the license plate classifier is trained by the positive and negative samples.The algorithm for license plate character recognition uses the gradient iterative method based on the BP neural network.Line character recognition.In the aspect of system design,S3C6410 integrated circuit development board is used as the core processor of the embedded hardware,then the establishment of Linux embedded system,the cutting of the kernel,the transplantation of the Opencv image function library and the construction of the cross compiling environment are carried out.The license plate image is more clearly defined and the license plate location is improved.As well as the accuracy of character recognition,the whole system has good robustness and practical value.