Design And Realization Of Embedded License Plate Recognition System

Since the reform and opening up more than 30 years,China's residents' income and consumption levels have increased significantly.The car has been from the previous luxury goods,gradually into daily consumer goods,many families have an even more cars.The rapid increase in car ownership has increased the burden of urban transport in China,especially in many large and medium-sized cities overwhelmed.The traffic management department is eager to ease the traffic pressure through a simple,efficient and intelligent way.License plate recognition system as a kind of intelligent traffic management aids began to attract people's attention and gradually become a hot issue.However,many products in the current domestic market are mostly having a certain degree of defects,and it is difficult to obtain satisfactory recognition results in the conditions of poor weather or complex shooting environment.Therefore,further study of better performance,more intelligent license plate recognition system has a very important practical significance.This paper mainly focuses on the recognition algorithms and the hardware system design of license plate.For the license plate recognition algorithms,firstly,the preprocessing algorithm of image gray scale,binarization and filtering is studied,and then the algorithm of license plate location,segmentation and recognition is discussed in depth.A rough-fine double stage license plate location algorithm based on boundary scan as well as a character segmentation algorithm based on continuous projection dichotomy is proposed.On this basis,an embedded license plate recognition system based on FPGA and DSP is also proposed.The design ideas and the implementation steps of the main hardware circuit and software program are given,including image acquisition module,FPGA and DSP application circuit module,LCD module and power supply module.Finally,some simple experiments are conducted to test the image capture,recognition accuracy and recognition speed of the license plate system.From the experimental results,it can be seen that the license plate recognition system studied in this paper has a good performance in terms of recognition accuracy,which can reach about 90%,and the system processing time is less than that of the existing system.Nevertheless,when some license plate characters are very similar,cannot get more satisfactory results,this will be the next step in the direction of research.