Research On License Plate Character Recognition Algorithm

With the development of the society,there are more and more vehicles.More and more vehicles bring convenience to the traffic,but also bring new challenges to the traffic department.In this paper,three important algorithms are studied,the training set and test set are obtained by template matching algorithm,and the license plate characters are recognized by support vector machine and convolutional neural network.The template matching algorithm obtains the recognition result by calculating the similarity between the character to be detected and the character template.In this paper,the license plate character detection is to detect a single character.Therefore,the original vehicle image is preprocessed to obtain a single character before template matching.The saved recognition results provide training set and test set for support vector machine and convolutional neural network to recognize license plate characters.Support vector machine(SVM)is a binary classification model,and license plate is composed of Chinese characters,letters and Numbers.Using the training set obtained by template matching,the multi-dimensional features of license plate characters are extracted to obtain the training data of support vector machine,and the nonlinear SVM is obtained by kernel function.The correct recognition rate of is improved compared with template matching.Convolutional neural network makes full use of the characteristics of license plate characters for learning.The convolutional neural network takes the whole grayscale image as input,extracts the local and global features of characters through multi-layer convolution and pooling operations,and achieves the goal of recognizing characters.The convolutional neural network has a high recognition rate when it is applied to character recognition.