Reserch On Offline Handwritten Chinese Character Recognition Based On CNN

With the development of science and technology,offline handwritten Chinese character recognition has become more and more widely used in daily life.However,the recognition of handwritten Chinese characters has become a very challenging problem in the field of pattern recognition due to the variety of offline handwritten Chinese characters,the randomness of writing and the similarity of Chinese characters.Therefore,how to quickly and accurately identify handwritten Chinese characters has become one of the most important research topics in the field of pattern recognition.In recent years,deep learning(especially convolutional neural network)has made a breakthrough in image recognition.As one of the representative algorithms of deep learning,convolutional neural network has the characteristics of automatic feature extraction,weight sharing.It performs well in handwritten Chinese character recognition.Therefore,using convolutional neural network to recognize handwritten Chinese characters has a good theoretical basis and practical value.The main achievements of this study are as follows:(1)aiming at the low recognition rate of offline handwritten Chinese characters,a handwritten Chinese character recognition method based on independent component analysis(ICA)and convolutional neural network(CNN)was proposed.This method improves the recognition rate of handwritten Chinese characters by adding image features.First,we extracted independent basis images and the projection vector of handwritten Chinese character recognition by using Fast ICA,and obtained the feature vector.Then,the feature vectors extracted by Fast ICA are put into the input layer of CNN for classification together with the original image.The project took full advantage of good feature extraction feature capability of ICA and strong classification ability of CNN.It was verified in CASIA-HWDB1.1 database that the recognition rate of handwritten Chinese characters was improved,which proved that ICA was helpful to optimize the convolutional neural network model and improve the recognition rate of handwritten Chinese characters.(2)To solve the problem of slow recognition speed of handwritten Chinese characters by convolutional neural network,a handwritten Chinese character recognition method based on two dimensional principal component analysis and convolutional neural network is proposed.This method improves the recognition speed of handwritten Chinese characters by reducing redundant data in the image.Firstly,the main features of handwritten Chinese characters were extracted by 2DPCA;Secondly,the extracted feature matrix is classified as the input of the convolutional neural network.Experiments show that the proposed method can greatly reduce the computational cost and running time of the convolutional neural network without losing the accuracy.