Study On Recognition Method Of Handwritten Chinese Character Set Based On Deep Learning

In people's production and life,there is a large amount of information produced every day.With the development of technology,many fields begin to digitize information in order to save,query and manage information.The digitalization of information involves handwritten Chinese character recognition technology.However,because of the large number of Chinese characters and complicated structure,the traditional handwritten Chinese character recognition system adopts the method of manual extraction of features.There are still many difficulties in handwritten Chinese character recognition.Calligraphy is a traditional art form unique to the Chinese nation.The main way to evaluate calligraphy copying works is to obtain image features through artificial design and shallow learning.Then according to the similarity between the original paste and copying exercises,the copy evaluation is given.With the continuous development of artificial intelligence technology,how to apply it to calligraphy copy evaluation system will be a significant research direction.Deep learning is a new field in machine learning and one of the most advanced topics in machine learning.The biggest difference between deep learning and traditional pattern recognition is that it can automatically learn features from big data,which can effectively reduce the tedious process of artificial feature extraction and the shortcomings of artificial interference error.In this paper,the recognition of handwritten Chinese characters and the evaluation of calligraphy copying are studied in depth learning technology.The main work includes:(1)By consulting literature,the research background,significance and research status of related technologies are analyzed.(2)This paper proposes a character segmentation method based on standard peripheral contour method,selects alexnet network,handwritten Chinese character data set and software and hardware environment,trains alexnet network and analyzes the result by using the selected handwritten Chinese character data set.By adding the convolution layer of alexnet network,the feature extraction ability of the network is improved,and the recognition accuracy of handwritten Chinese characters is improved.(3)By analyzing the principle of calligraphy copying evaluation and the way of extracting features by convolution neural network,a method of calligraphy copying evaluation based on alexnet is put forward.The digital image processing technology is used to expand the calligraphy data set which is used to train the evaluation model of calligraphy copying,including image rotation,image scaling and image denoising.The expanded calligraphy data set is used to train the alexnet network,and then to evaluate the calligraphy copying works,which proves the feasibility and effectiveness of the calligraphy copying evaluation based on the deep learning.