Research On License Plate Recognition Method Based On Image Processing

The intelligent transportation system applies various aspects of technology to the entire transportation management system effectively,and establishes an integrated transportation and management system with full-scale,wide-ranging functions and highefficiency,real-time and accuracy.License plate recognition technology,as one of the core technologies in the intelligent transportation system,has high research value,and its development determines and restricts the development of national intelligent transportation.In addition,the research on license plate recognition technology can be extended to various fields such as industrial applications,biometrics,medical detection,and artificial intelligence,providing a good technical reserve for our follow-up and social connection.This paper uses image processing technology to solve the license plate recognition problem.The image processing method is used to perform a series of operations on the original car picture.The specific operation steps include image preprocessi ng,license plate positioning,license plate character segmentation,and character recognition,and finally the license plate identification information is obtained.The research content of this paper is mainly from these four aspects.After theoretical an alysis and experimental verification,the optimal image processing method is obtained,and the recognition rate of license plate recognition is further improved.The research content is as follows:(1)Explained in detail the related research background and significance of the project,as well as the domestic and foreign research status related to license plate recognition based on image processing.(2)Discussed the license plate recognition technology based on indirect method and direct method based on recognition,and set up and tested the experimental environment of the thesis.It was clear that this paper was based on the image processing,studied the license plate recognition method,and determined The research on the license plate recognition method is discussed in detail from the four parts of the car image preprocessing,license plate positioning,license plate character segmentation and license plate character recognition.(3)A series of pre-processing of the images collected by the original car,t hrough theoretical analysis combined with experimental comparison,to get a better pretreatment program,lay the foundation for the positioning of the rear license plate.(4)Discusses the commonly used license plate location method,and proposes an approximate positioning method based on license plate features and a license plate accurate positioning method based on jump method and SVM on the basis of commonly used methods.It describes in detail the license plate of the original car image.The approximate positioning and precise positioning of the specific process,and through the original car graphics using different license plate precise positioning method for experimental comparative analysis,demonstration of the proposed approximate positioning and precise positioning of the license plate positioning program effectiveness.(5)Discussed the commonly used license plate character segmentation and character recognition methods.Based on this,a license plate character segmentation method based on the character contour method was proposed.The detailed process of this method was experimentally demonstrated,and good results were obtained and used.Based on the method of BP neural network,the vehicle license plate is identified.Based on the original BP neural network,the momentum factor is added,and the characters are grouped and identified.Two sub-networks are trained to improve the license plate recognition efficiency and recognition rate.Finally,it summarizes the work content of this topic,and poin ts out some of the shortcomings in the work,and points out the direction for subsequent improvement and improvement.