期刊文献+

基于深度学习网络的手写英文自动化识别模型在机器英汉互译中的应用研究

Research on the application of automatic handwritten English recognition model based on deep learning network in machine English-Chinese translation
原文传递
导出
摘要 经济全球化使得语言翻译需求日益增加,实验针对手写英文字体易出现的写作不规范、存在涂改区域等问题提出一种结合深度学习网络的手写英文自动识别模型。该模型基于CNN和RNN形成的CRNN复合网络,首先在此基础上针对英文字体可能出现的书写不规范问题,对CRNN模型进行第一步改进得到CRNN-1模型;然后对CRNN-1模型引入注意力机制来增强提取文字前后之间的关联性,得到CRNN-A模型;最后针对手写英文字体可能出现的涂改问题建立分类模型,得到CRNN-C模型。实验对提出的CRNN-C模型进行性能验证,结果表明识别准确率能达到96.87%左右,为机器翻译高效无误地运行贡献力量。 The economic globalization has made the demand for language translation increase day by day.The experiment proposes a handwritten English automatic recognition model combined with deep learning network to solve the problems such as non-standard writing and the existence of altered areas.This model is based on the CRNN composite network formed by CNN and RNN.On this basis,the first step is to improve the CRNN model to obtain the CRNN-1 model for the possible non-standard writing of English fonts;Then the attention mechanism is introduced into CRNN-1 model to enhance the relevance between the text before and after extraction,and the CRNN-A model is obtained;Finally,a classification model is established for the possible erasure of handwritten English fonts,and the CRNN-C model is obtained.The experiment verifies the performance of the proposed CRNN-C model,and the results show that the recognition accuracy can reach about 96.87%,contributing to the efficient and error free operation of machine translation.
作者 何娟 HE Juan(Xi’an Siyuan University,Xi’an 710038,China)
机构地区 西安思源学院
出处 《自动化与仪器仪表》 2023年第7期191-195,共5页 Automation & Instrumentation
基金 陕西省教育科学“十四五”规划2022年度课题《高校公共体育课程思政建设的方法与路径研究》(SGH22Y1871)。
关键词 深度学习网络 CRNN LSTM 手写英文 自动识别 机器翻译 deep learning network CRNN LSTM handwritten English automatic identification MT
  • 相关文献

参考文献13

二级参考文献74

共引文献39

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部