License Plate Recognition Based On Support Vector Machine (SVM)

With the rapid development of computer software and hardware,license plate recognition systems have also developed rapidly.However,many license plate recognition systems that have been put into use can only be applied to one or several specific environments.When the external environment changes significantly,the recognition success rate of these systems will rapidly drop.This is also the main reason why many license plate recognition systems that have been put into production have limited application scope.Based on the problems encountered in the application of license plate recognition system,this article starts with the theory and algorithm related to license plate recognition,and adopts a series of methods to successfully achieve the license plate image acquisition,license plate character positioning,license plate character segmentation and license plate characters.Recognition and other functions.In the acquisition phase of license plate characters,this article first introduces the composition of the vehicle image acquisition system.After the vehicle image is collected,a series of preprocessing is performed on the vehicle image,Including image grayscale,image binarization and so on.Then the license plate part in the image was positioned.In the license plate positioning stage,the processing method of the inclined license plate and the determination method of the license plate character boundary were introduced.In the character segmentation phase,this article describes in detail the character width of each character in the license plate and the characteristics of the spacing between characters.Based on the existing method of character segmentation of various license plates,this paper proposes the characteristics of these known license plate characters.A vertical projection segmentation algorithm with prior knowledge constraints can greatly improve the speed and accuracy of license plate character segmentation.In the recognition phase of license plate characters,this paper proposes a character recognition method based on the combination of statistical features and structural features of characters,and combines the support vector machine algorithm(SVM)to identify characters,and introduces the support vector machine(SVM).The principles and algorithms used in the selection of some kernel functions and the construction of classifiers and so on.After all the functions of the license plate recognition system were successfully implemented,this article has carried on thedesign of the operation interface and the database creation to the system.This article reviews the various methods used in the above process,and performs a series of optimizations on related algorithms.Then a series of tests are performed using the images of the vehicles collected in the laboratory.The experimental results show that this paper uses the support vector machine.(SVM)algorithm for license plate recognition can effectively identify the license plate number in different working environments.