摘要
中国书法文化历史悠久,其中硬笔书法兼具艺术与实用价值。为应对电子设备普及造成的硬笔书写能力下降问题,本文提出一种融合字体风格、笔画顺序和笔力的多特征力反馈硬笔书法教学模式。方法上,首先提出了一种基于对比学习的Dense-CycleGAN模型,用于生成不同书写风格的硬笔书法字体库。其次,利用匈牙利算法对汉字的笔画顺序进行标准化处理。最后,基于力反馈设备采集的书写数据,建立了笔画宽度到书写力度的映射模型。在五种风格字体上的实验结果表明,本文提出模型在生成字体的结构相似性指数达到了0.587的均值,优于传统CycleGAN;笔顺规范算法的整体相似度动态时间规整均值为0.044,余弦相似度均值为0.998,精度较高。用户评估实验中,书写引导性评分为4.5/5,教学辅助性为4.1/5,验证了该模式的教学实用性与推广潜力。该书写模式真实再现了硬笔书法书写过程,实现了兼顾字体风格、笔顺、笔力特征的硬笔书法教学,为硬笔书法的教育提供了一种新型融合策略。
Chinese calligraphy has a long and rich history,with hard-pen calligraphy bearing both artistic and practical significance.To address the decline in hard-pen handwriting ability caused by the widespread use of electronic devices,this paper proposes a multi-feature hard-pen calligraphy teaching mode based on force feedback,which integrates font style,stroke order,and writing pressure.Specifically,a Dense-CycleGAN model based on contrastive learning is developed to generate calligraphy font libraries in different styles.The stroke order of Chinese characters is standardized using the Hungarian algorithm.Furthermore,a mapping model from stroke width to writing pressure is constructed based on data collected via force feedback devices.Experimental results on five font styles show that the proposed model achieves an average Structural Similarity Index Measure(SSIM)of 0.587 in character generation,outperforming the traditional CycleGAN.The stroke order standardization yields a Dynamic Time Warping(DTW)score of 0.044 and an average cosine similarity of 0.998,indicating high accuracy.In user evaluation experiments,the writing guidance and teaching assistance received scores of 4.5/5 and 4.1/5,respectively,validating the practicality and applicability of the proposed mode.This writing mode faithfully reproduces the hard-pen calligraphy process and enables instruction that comprehensively considers font style,stroke order,and writing pressure,offering a novel integrated strategy for calligraphy education.
作者
张会欣
赵启荣
熊敏
Zhang Huixin;Zhao Qirong;Xiong Min(School of Big Data&Software Engineering,Chongqing University,Chongqing 400000,China)
出处
《电子测量技术》
北大核心
2025年第24期110-120,共11页
Electronic Measurement Technology
基金
重庆市语言文字工作委员会重点项目(yyk22110)资助。
关键词
力反馈
人机交互
硬笔书法
多特征融合
force feedback
human-computer interaction
hard-pen calligraphy
multi-features fusion
