In order to solve the problem of low recognition accuracy of similar handwritten Chinese characters by traditional off-line handwritten Chinese character recognition methods:the present study proposes a two-stage offline handwritten Chinese character recognition method that combines image processing and convolutional neural network,the first stage uses the traditional convolutional neural network for recognition,and the second stage uses the differences discrimination method based on image processing for more accurate secondary recognition.The experimental results show that the differences discrimination method based on image processing combining with convolutional neural network has better performance than the recognition method based on convolutional neural network alone,and thus the overall handwritten Chinese character recognition accuracy can be improved.In addition,the proposed method shows more stable recognition and can effectively deal with the case that training set contains mislabeled samples.
关键词
脱机手写汉字识别/相似汉字/卷积神经网络/差异辨别方法
Key words
Offline Handwritten Chinese Character Recognition/Similar Chinese Characters/Convolutional Neural Network/Method of Differences Discrimination
维普
万方数据