Chinese Handwriting Detection And Recognition For Intelligent Homework Review

The 21 st century is the era of artificial intelligence.With the rapid development of artificial intelligence,it has been widely used in many fields and achieved fruitful results.With the national strategy of strengthening the country through education,"artificial intelligence + education" is realizing the deep integration of artificial intelligence and education and teaching reform by combining the application requirements of artificial intelligence and education.This thesis,aiming at the application requirements of homework intelligent evaluation in primary and secondary schools,focuses on the task of handwritten Chinese character detection and recognition for user's homework image in intelligent teaching assistant system.The main work and contribution of this thesis are as follows:1.Geometric correction and submission validity detection.This part of the work is the basis of the detection and recognition of Chinese character handwriting.First of all,a remote user can take a photo of the completed paper-based handwriting homework,collect and upload it to the system according to instructions.By means of image processing,the uploaded image can be binarized and filtered,and some geometric features can be extracted.Then with the help of perspective transformation,the uploaded homework image can be geometric corrected.In order to ensure the effectiveness of the uploaded homework image,this thesis also proposes and implements a kind of detection mechanism based on histogram similarity measurement between the effective region of the uploaded homework image and the job template image.Finally,the effective homework image after geometric correction can be sent to the subsequent Chinese handwriting detection module to realize Chinese handwriting detection.2.Handwritten Chinese character detection based on PixelLink.Different from traditional border regression,this thesis implements Chinese handwriting detection based on PixelLink,in which deep learning strategy is effectively combined with the text detection based on the case segmentation.Finally,the segmentation results are obtained by means of text prediction and connection prediction,and handwriting border can be directly extracted from the segmentation results.3.Chinese character handwriting recognition based on ResNet.Handwritten Chinese character recognition is essentially a multi-class classification problem based on image data.As ResNet(Residual Network)can effectively solve the problem of gradient disappearance and gradient explosion when training traditional deep neural networks,by inputting every effective detection area from the result of Chinese handwriting detection,this thesis combines ResNet with Chinese handwriting recognition based on the deep learning model.Experimental results show that the Chinese character handwriting detection and recognition of homework image in this thesis can successfully detect and recognize Chinese character handwriting,which is helpful for further application of intelligent teaching assistant system.