基于深度学习的汉盲转换系统  

Chinese-braille conversion system based on deep learning

作　　者：严泉勇 吴跃成 YAN Quanyong;WU Yuecheng(School of Mechanical Engineering,Zhejiang Sci-Tech University,Hangzhou 310018,China)

机构地区：[1]浙江理工大学机械工程学院,杭州310018

出　　处：《智能计算机与应用》2024年第6期183-187,共5页Intelligent Computer and Applications

基　　金：浙江省基础公益研究计划(LGF19E050005)。

摘　　要：汉盲转换是将中文电子文本信息自动转换成盲文数字化信息,这将会对视障人士的学习方式和生活水准产生巨大的影响。本文为解决汉盲转换中的分词连写问题,基于深度学习技术构建了一种双向门限循环单元条件随机场(BIGRU-CRF)网络模型。该模型通过双向门限循环单元获取上下文信息,再结合条件随机场模型序列标注特性,从而得到较为准确的分词连写结果,极大地提升了汉盲转换的准确率。测试结果表明,所设计的汉盲转换系统能够精确地将汉字转换成盲文,具有一定的实用性。Chinese-braille translation is the automatic translation of Chinese electronic text information into Braille digital information,which will have a huge impact on the learning style and living standards of the visually impaired.In this paper,a Bidirectional Gated Recurrent Unit Conditional Random Field(BIGRU-CRF)network model is constructed based on deep learning technology to solve the problem of Chinese-blind translation.The model obtains the context information through the bidirectional gated recurrent unit,and combines with the characteristics of conditional random field model sequence labeling,so as to obtain more correct word segmentation and sequential writing results,and greatly improves the accuracy of Chinese-braille translation.The experimental results show that the Chinese-blind translation system designed in this paper can accurately convert Chinese characters into braille,and has certain practicability.

关 键 词：汉盲转换系统 BIGRU-CRF 分词连写 深度学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]