Research Of License Plate Recognition Technology Based On Deep Learning And FPGA Implementation

The development of national industrialization and the popularization of automobiles have greatly facilitated people's daily life,but this has led to increasingly serious traffic problems,and the emergence of intelligent transportation has effectively solved this problem.As the basic component of intelligent transportation system,license plate recognition technology is of great significance to the rapid development of intelligent transportation system.Aiming at the problem that the existing license plate recognition system has low recognition accuracy under complex background,combined with the advantages of deep learning algorithm in the field of image processing,the license plate location and character recognition algorithms are studied.The main research contents are as follows:Through the study of deep learning algorithm,comparing the advantages and disadvantages of various algorithms and the scope of application,the convolutional neural network is selected as the algorithm used in vehicle license plate location and character recognition.Research on license plate location algorithm.Aiming at the problem of difficult license plate location in complex backgrounds such as license plate tilt and image quality degradation,a rough license plate location algorithm based on improved edge detection and morphological filtering is proposed.The license candidate area obtained after rough positioning is well trained.The Convolutional Neural Network(CNN)model judges and eliminates the pseudo license plate area to achieve precise positioning.Research on license plate character recognition algorithm.On the basis of considering the characteristics of Chinese license plate characters,the AlexNet network model of convolutional neural network is used to improve the model in order to realize the overall recognition of the license plate without character segmentation.Under the premise of ensuring the accuracy of the license plate recognition system,the real-time performance of the system is improved.Due to the rich hardware resources and powerful parallel computing capabilities of FPGA,this paper proposes an implementation scheme of license plate technology based on FPGA.The results show that the accuracy of the license plate recognition system in the complex background is more than 97%,the recognition rate is over 95%,and the running time of the system is about 0.5s,which meets the system performance requirements.