基于神经网络的纸笔手写规范汉字量化评价研究  

Research on quantitative evaluation of standard Chinese characters written by pen and paper based on neural network

作　　者：王敏 马万 祝闯 史善飞 舒江波[2] 卢帅成 WANG Min;MA Wan;ZHU Chuang;SHI Shanfei;SHU Jiangbo;LU Shuaicheng(Jiangsu Province Support Software Engineering R&D Center for Modern Information Technology Application in Enterprise,Suzhou 215104,Jiangsu,China;National Engineering Research Center of Educational Big Data,Central China Normal University,Wuhan 430079,China)

机构地区：[1]江苏省现代企业信息化应用支撑软件工程技术研发中心,江苏苏州215104 [2]华中师范大学教育大数据应用技术国家工程研究中心,武汉430079

出　　处：《华中师范大学学报（自然科学版）》2023年第6期813-820,共8页Journal of Central China Normal University：Natural Sciences

基　　金：教育部人文社会科学研究规划基金项目(22YJAZH093);科技创新2030——“新一代人工智能”重大项目(2022ZD0117105);苏州市职业大学专业(群)服务产业发展专项课题(SZD2022ZX005)。

摘　　要：汉字书写教育是我国素质教育的重要内容之一.当前,我国中小学生规范汉字书写教育存在两大问题:1)缺乏有效指导,规范汉字书写要领以定性描述为主,细节要点的量化特征没有体现,学生无法准确把握书写细节;2)缺乏有效的评价,大部分教师或家长对学生的书写只能给出整体的定性评价,无法给出精准的细节评价.针对上述问题,该文开展日常自然书写场景下纸笔手写规范汉字的量化评价研究,主要包括三个方面的工作:1)利用手写汉字的骨架图像和预评价结果构建基于卷积神经网络的分类评价模型,实现对手写汉字书写规范性的自动分类评价;2)基于手写汉字笔段属性数据实现对书写细节的量化评价,并给出书写调整的建议;3)利用手写汉字各项书写规则的书写规范性评分,构建基于多元线性回归的书写评分模型,实现对手写汉字整字书写质量的评价.Chinese character writing education is an important part of our country's quality education.Currently,there are two major problems in the education of standard Chinese character writing for primary and secondary school students in China.Firstly,there is a lack of effective guidance,with the emphasis being mainly on qualitative descriptions of standard Chinese character writing essentials,and the quantitative characteristics of details are not reflected,making it difficult for students to accurately grasp the writing details;secondly,there is a lack of effective evaluation,with most teachers or parents only able to give an overall qualitative evaluation of students'writing,and unable to provide precise detailed evaluation.To address these issues,this study focuses on the quantitative evaluation of standard Chinese character handwriting using pen and paper in natural writing scenarios,including three main areas of work:1)using the skeleton image of handwritten Chinese characters and pre-evaluation results to construct a classification evaluation model based on convolutional neural networks,achieving automatic classification evaluation of the standardization of handwritten Chinese characters;2)using the stroke attribute data of handwritten Chinese characters to achieve quantitative evaluation of writing details and providing suggestions for writing adjustments;3)using the standardization score of various writing rules of handwritten Chinese characters to construct a writing scoring model based on multiple linear regression,achieving an evaluation of the overall quality of handwritten Chinese characters.

关 键 词：规范汉字 书写规则 量化评价 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]