节点文献
续笔行为的数字化采集、确认及处理的实验研究
【作者】 王静；
【导师】 李艺；
【作者基本信息】 南京师范大学 ， 教育技术学， 2019， 硕士
【摘要】 续笔行为是指书写者在汉字书写过程中,有意识、有目的地对已完成的笔画进行再次延长、覆盖、连接等加工或修饰的书写行为。其在常规的硬笔—纸张介质中频繁出现且较容易识别,但在目前较为普及的智能化书写平台中未能准确识别,只是简单地将其判断为书写错误,要求书写者重新书写,从而使书写者丧失书写热情,导致其在汉字练习过程中只考虑书写是否正确,而不追求汉字美观。本文基于此,提出续笔识别与处理算法,实现基于触摸屏手写平台上续笔行为的识别与处理,使包含续笔关系的汉字按照书写者的真实意图通过正确性阶段判断。本文穷尽有可能发生的所有续笔情形,并根据续笔关系中涉及的笔画数量、笔段数量、笔段间的位置关系、连接方向以及单字中存在的续笔次数等多层面划分续笔行为,共统计出十一类续笔类型。根据续笔类型的划分标准,计算不同笔画的笔段间相对位置,将可能符合续笔判断标准的笔画进行续笔类型、续笔相关笔画序号、相关笔段序号、相叠区域的始末点坐标等信息的标记,借助标记将或许存在续笔关系的笔画走向码整合,与新标准笔画类型编码库比对,若存在,则表明该续笔行为识别正确,若不存在,则初步判断的续笔关系为伪续笔,至此续笔识别阶段结束。续笔行为识别成功仅是对续笔笔画进行标注,并不能使含续笔行为的汉字通过正确性阶段判断,接下来应将具有续笔关系的笔画进行处理,使其按照书写者的书写意图整合续笔笔画与原笔画,除笔画点集整合外,若续笔笔画的笔画类型发生变化,则还需更改原笔画的走向码,再将整合后的笔画点集作为原笔画的最终点集送到正确性阶段进行判断。若正确性阶段通过,则表明续笔行为处理成功,若正确性阶段仍判断书写错误,则表明续笔行为处理失败。为使该算法更成熟化地应用于数字化书写平台,对汉字书写、练习有一定指导性意义。本研究开展两轮续笔行为验证实验,分别为研究者内测3500常用字集、10名普通测试者测试45字集。两轮实验过后,具有续笔关系的汉字正确性阶段通过率达80%以上。因此该续笔判断与处理算法可用于智能化手写平台,但部分算法仍存在继续优化的空间,具体的应用策略也需进一步确定。
【Abstract】 The supplementary strokes refer to a writing behavior in which the finished strokes get extended,covered,connected or other modified again when writing Chinese characters.It often appears in regular hard pen-paper media and identifies easily,but in most cases,which is not accurately identified in the currently popular intelligent writing platform,and simply judged as a wrong writing act.The writer who loses the passion for writing need to erase and rewrite the Chinese character,they only consider how to write Chinese characters correctly,not pursue the beauty of Chinese characters.For judging the supplementary strokes in Chinese characters through the correctness stage,the paper proposes an algorithm identifying and processing the supplementary strokes on the digital handwriting platform.This paper exhausts all the possible situations of the supplementary strokes.Three categories,fourteen small types of the supplementary stroke can be formed on the number of strokes,the number of stroke segments,the positional relationship between the segments,the connection direction,and the number of occurrences in a word.Firstly,Calculating the positional relationship between stroke segments of different strokes and determining the type of the supplementary strokes according to the division standard of the supplementary strokes.Secondly,adding the mark information to the stroke segment information,for example the type of the supplementary stroke,related serial number of stroke,related serial number of stroke segment,and the coordinates of the start and end points about the overlapping area.Lastly,checking it with the new standard stroke type code after integrating the stroke direction code of the continuous stroke relationship.If it exits,the supplementary stroke can be correctly identified.On the contrary,the continuation relationship between strokes in the preliminary judgment is a pseudo-continuation stroke,and the recognition phase of the supplementary stroke ends.Identify successful the supplementary strokes which judge error by correctness.So the strokes with continuous stroke relationship should be processed to integrate the strokes and original strokes according to the writer’s writing intention.Besides the point set integration,the orientation code of the original stroke needs to be changed if the type of stroke changed.And then,the integrated stroke point set as the final point set of the original stroke is sent to the correctness stage.If the processed continuation stroke passes the correctness stage,it indicates that the processing of the supplementary stroke is successful,and if the correctness stage still judges the writing error,it indicates that the processing of the supplementary stroke fails.In order to make the algorithm more correctly on the digital writing platform,the paper carries out two rounds of the supplementary stroke recognition experiments,which are respectively tested by the researcher to test 3500 common word set and 10 ordinary testers to write 45 words set.Experimental results show the pass rate of the correctness of Chinese characters contain the supplementary strokes reaches over 80%.Therefore,the judgment and processing algorithm about the supplementary stroke can be used for intelligent handwriting platform,but some algorithms have space for continuous optimization.