基于Web的联机手写汉字识别仿真系统设计  

Design of a Web-based Online Handwritten Chinese Character Recognition Simulation System

作　　者：曲丽娜[1] QU Li-na(School of Information Engineering at Jilin Engineering Normal University,Changchun Jilin 130052,China)

机构地区：[1]吉林工程技术师范学院信息工程学院,吉林长春130052

出　　处：《吉林工程技术师范学院学报》2018年第10期105-107,共3页Journal of Jilin Engineering Normal University

基　　金：吉林省教育厅"十二五"科学技术研究项目(2014571)

摘　　要：联机手写汉字识别的模拟仿真系统是基于卷积神经网络方法开发的。利用TensorFlow实验平台,以中国科学院采集的CASIA-OLHWDB数据集进行训练获得分类数据。使用基于Python的Flask框架搭建Web服务器,采用Ajax技术实现数据在Web端的实时自动刷新。仿真结果表明,本系统能够较精确识别联机手写汉字,达到预计目的。An online handwritten Chinese character recognition simulation system was developed on the basis of the algorithm of convolution neural network,in which classified data were obtained by drilling with the CASIA-OLHWDB dataset collected by Chinese Academy of Science on TensorFlow platform and refreshed real-time and automatically via Ajax on the Web servers built upon the Python-based Flask framework.The result shows that the system is able to recognized online handwritten Chinese characters accurately enough to meet the requirements.

关 键 词：卷积神经网络 联机手写汉字识别 TensorFlow框架 Flask框架 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]