目的 针对包装业务中签名防伪验证性能存在的问题,提出并验证在纯力学驱动条件下具有高判别能力的力学特征,提升包装身份认证与责任追溯的安全性和可靠性。方法 在梳理现有研究中常用力学特征的基础上,结合简化动力学模型,构建包括法向力、主辅方向分力比、整段书写过程的力积分及力微分等4类候选特征,并采用基于排序的特征选择方法进行筛选。通过自研数字触笔采集签名数据,可同时记录三维力、方位角和笔轴角。结果 与动态时间规整等方法进行对比后发现,基于力学特征的方法在等错误率和分类准确率等指标上显著优于基线方法,验证了所提出特征集的有效性。结论 所筛选的动力学特征为力学驱动的签名验证系统提供了性能提升依据,并可为后续动力学模型的改进和包装场景下的系统优化提供参考依据。
Abstract
To address the performance issues of signature anti-counterfeiting verification in packaging, the work aims to propose and validate a mechanical feature with high discriminative capability under purely mechanical driving conditions, thereby enhancing the security and reliability of packaging authentication and responsibility tracing. On the basis of reviewing commonly used mechanical features in existing research, four types of candidate features were constructed by integrating a simplified dynamic model, including normal force, the ratio of primary to auxiliary directional force components, force integral, and force differential over the entire writing process. A ranking-based feature selection method was employed for screening. Signature data was collected with a self-developed digital stylus pen, capable of simultaneously recording three-dimensional force, azimuth angle, and barrel rotation angle. A comparative analysis with methods like dynamic time warping revealed that the mechanical feature-based method significantly outperformed the baseline methods in metrics such as equal error rate and classification accuracy, validating the effectiveness of the proposed feature set. The selected dynamic features provide a basis for enhancing the performance of mechanically-driven signature verification systems and offer valuable insights for future improvements to dynamic models and system optimization in packaging scenarios.
关键词
签名验证 /
特征提取 /
动力学模型
Key words
signature verification /
feature extraction /
dynamical model
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] IMPEDOVO D, PIRLO G.Automatic Signature Verification: The State of the Art[J]. IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), 2008, 38(5): 609-635.
[2] 张佳禾. 基于深度学习的多环境离线签名识别研究[D]. 西安: 西安理工大学, 2023: 1-2.
ZHANG J H.Research on Multi-Environment Off-Line Signature Recognition Based on Deep Learning[D]. Xi'an: Xi'an University of Technology, 2023: 1-2.
[3] 焦慧敏, 王党校, 张玉茹, 等. 基于书写摩擦力的签名识别方法[J]. 自动化学报, 2011, 37(7): 883-890.
JIAO H M, WANG D X, ZHANG Y R, et al.Signature Verification Using Handwriting Friction Force[J]. Acta Automatica Sinica, 2011, 37(7): 883-890.
[4] SRIHARI S N, CHA S H, ARORA H, et al.Individuality of Handwriting[J]. Journal of Forensic Sciences, 2002, 47(4): 1-17.
[5] PLAMONDON R, MAARSE F J.An Evaluation of Motor Models of Handwriting[J]. IEEE Transactions on Systems, Man, and Cybernetics, 1989, 19(5): 1060-1072.
[6] 刘若男. 面向动态特征的签名笔迹身份识别[D]. 沈阳: 沈阳工业大学, 2024: 15-16.
LIU R N.Dynamic Feature-Oriented Signature Handwriting Identification[D]. Shenyang: Shenyang University of Technology, 2024: 15-16.
[7] FIERREZ J, ORTEGA-GARCIA J.On-Line Signature Verification[M]// Handbook of Biometrics. Boston: Springer US, 2007: 189-209.
[8] JIAO H M, WANG D X, ZHANG Y R.Different Role of Friction and Normal Force for Force-Based Signature Verification[C]// 2009 2nd International Congress on Image and Signal Processing. Tianjin: IEEE, 2009: 1-5.
[9] LEE L L, BERGER T, AVICZER E.Reliable Online Human Signature Verification Systems[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1996, 18(6): 643-647.
[10] RICHIARDI J, KETABDAR H, DRYGAJLO A.Local and Global Feature Selection for On-Line Signature Verification[C]// Eighth International Conference on Document Analysis and Recognition (ICDAR'05). Seoul: IEEE, 2006: 625-629.
[11] TSIS M Y, LAN L S.Online Recognition of Chinese Handwritten Characters Based on the Point Sistribution Model[C]// In International Computer Symposium, 2004: 505-510.
[12] CHEN H, AGAZZI O E, SUEN C Y.Piecewise Linear Modulation Model of Handwriting[C]// Proceedings of the Fourth International Conference on Document Analysis and Recognition. Ulm: IEEE, 1997: 363-367.
[13] PLAMONDON R.A Kinematic Theory of Rapid Human Movements: Part Ⅲ. Kinetic Outcomes[J]. Biological Cybernetics, 1998, 78(2): 133-145.
[14] PLAMONDON R.A Kinematic Theory of Rapid Human Movements: Part Ⅱ. Movement Time and Control[J]. Biological Cybernetics, 1995, 72(4): 309-320.
[15] PLAMONDON R, FENG C H, WOCH A.A Kinematic Theory of Rapid Human Movement. Part Ⅳ: A Formal Mathematical Proof and New Insights[J]. Biological Cybernetics, 2003, 89(2): 126-138.
[16] HOLLERBACH J M.An Oscillation Theory of Handwriting[J]. Biological Cybernetics, 1981, 39: 139-156.
[17] DENIER J, THURING J.The Guiding of Human Writing Movements[J]. Kybernetik, 1965, 4(2): 145-148.
[18] JONAS R, KETABDAR H, DRYGAJLO A.Local and Global Feature Selection for On-Line Signature Verification[C]// Eighth International Conference on Document Analysis and Recognition. IEEE, 2005.
[19] WANG D X, ZHANG Y R, YAO C, et al.Toward Force-Based Signature Verification: A Pen-Type Sensor and Preliminary Validation[J]. IEEE Transactions on Instrumentation and Measurement, 2010, 59(4): 752-762.
基金
国家级大学生创新创业训练计划（22150725009,22150725011）;数字印刷装备北京市重点实验室平台建设项目（KYCPT202508）
PDF(1058 KB)
渝公网安备 50010702501716号 渝ICP备15012534号-2
