摘要
机器书写技术的快速发展带来了笔迹真伪鉴定的新挑战。识别机器模仿的笔迹需要从视觉和运动特征入手，构建合适的分类模型。本文系统综述了该领域的研究进展，重点分析了基于图像分析和运动建模的主要识别方法，讨论了其优势和局限性。在此基础上，提出了一种结合专家知识的机器笔迹检测系统框架，并讨论了该框架在实际应用中可能面临的挑战和限制。通过构建专家知识库和规则引擎，实现数据驱动和知识驱动的融合，以期获得更加全面可靠的判别结果。最后，文章探讨了未来的改进方向，包括扩大数据集、精细化运动认知建模、挖掘新型物理化学特征等，为实现可信人机书写环境提供参考。
关键词： 机器笔迹；笔迹鉴定；专家知识；图像分析；运动建模
Abstract
The rapid development of machine writing technology has brought new challenges to the authentication of handwriting authenticity. To recognize the handwriting imitated by the machine, it is necessary to construct a suitable classification model based on the visual and motion features. This paper systematically reviews the research progress in this field, focusing on the main recognition methods based on image analysis and motion modeling, and discusses their advantages and limitations. On this basis, a framework of machine handwriting detection system combining expert knowledge is proposed, and the possible challenges and limitations of this framework in practical application are discussed. By constructing expert knowledge base and rule engine, the integration of data driven and knowledge driven can be realized in order to obtain more comprehensive and reliable judgment results. Finally, the paper discusses the future improvement direction, including expanding the data set, refining the motor cognitive modeling, mining new physicochemical characteristics, etc., to provide reference for the realization of trusted human-machine writing environment.
Key words： Machine handwriting; Handwriting identification; Expert knowledge; Image analysis; Motion mode-ling
参考文献 References
[1] Li J, Song G, Zhang M .Occluded offline handwritten Chinese character recognition using deep convolutional generative adversarial network and improved GoogLeNet [J]. Neural Computing and Applications,2018, 32(9):1-15.
[2] Alafif, Tarik, Alharbi, et al. GEAD: generating and evaluating handwritten Eastern Arabic digits using generative adversarial networks[J]. International Journal of Information Technology,2021,14(1):1-9.
[3] ZC L, L W. Style-preserving English handwriting synthesis[J]. Pattern Recognition: The Journal of the Pattern Recognition Society,2007,40(7):2097-2109.
[4] Alonso E, Moysset B, Messina O R. Adversarial Generation of Handwritten Text Images Conditioned on Sequences. [J]. CoRR,2019, abs/1903.00277
[5] 高俊.基于速度的手写笔迹实时仿真算法研究[D].江西财经大学,2020.
[6] 司佳,陈思平,袁洲,等.基于图像识别与生成技术的人工智能技术应用[J].科技资讯,2023,21(22):47-50.
[7] Liu Y, Wang L, Cheng J, et al. Multi-focus image fusion: A Survey of the state of the art[J]. Information Fusion,2020, 6471-91.
[8] 徐秋菊.图的规范化拉普拉斯特征多项式和谱[D].江西师范大学,2019.
[9] Okawa M. Synergy of foreground–background images for feature extraction: Offline signature verification using Fisher vector with fused KAZE features[J]. Pattern Recognition, 2018,79480-489.
[10] Jagtap B A, Hegadi S R ,Santosh K .Feature Learning for Offline Handwritten Signature Verification Using Convolutional Neural Network[J]. International Journal of Technology and Human Interaction (IJTHI), 2019,15(4): 54-62.
[11] Zois N E, Alexandridis A, Economou G. Writer independent offline signature verification based on asymmetric pixel relations and unrelated training-testing datasets[J]. Expert Systems With Applications, 2019, 12514-32.
[12] Hafemann G L, Sabourin R, Oliveira S L. Offline Handwritten Signature Verification - Literature Review. [J]. CoRR, 2015, abs/1507.07909
[13] Boulid Y, Lazrek A, Moreau R, Anquetil E. Machine-printed and handwritten text discrimination for Arabic documents[C]//2017 14th IAPR International Conference on Document Analysis and Recognition (ICDAR). IEEE, 2017: 1463-1468.
[14] Diaz M, Ferrer A M, Impedovo D , et al.A Perspective Analysis of Handwritten Signature Technology[J].ACM Computing Surveys (CSUR),2019,51(6):1-39.
[15] 杜晓瑜,张燕.签名笔迹动态特征分析技术现状[J].中国科技信息,2024,(15):75-77.
[16] Wu Z, Fang P, Shen F. Online signature verification based on global feature of writing forces[M]//Advances in Biometrics. Springer, Berlin, Heidelberg, 2005: 827-833.
[17] Chen X, Champod C, Yang X, et al. Assessment of signature handwriting evidence via score-based likelihood ratio based on comparative measurement of relevant dynamic features[J]. Forensic Science International,2018, 282101-110.
[18] Ruben T, Ruben R V, Julian F , et al.BioTouchPass2: Touchscreen Password Biometrics Using Time-Aligned Recurrent Neural Networks[J]. IEEE Transactions on Information Forensics and Security,2020,152616-2628.
[19] 才让先木. 基于域泛化的跨数据集离线签名验证研究[D]. 四川大学, 2023.
[20] 张宇宽, 施少培, 王雅晨, 等. 笔迹鉴定标准比较研究[J]. 中国司法鉴定, 2024,(01): 99-106.