基于幻象多尺度贝叶斯网络的在线手写签名认证
Shenyang Jianzhu University, School of Computer Science and Engineering, Shenyang 110168
| 摘 要： | 针对在线手写签名认证在小样本条件下存在特征表达受限、参数学习不稳定以及分类决策可靠性不足等问题，提出了一种融合轻量化表征、多尺度特征建模与概率判别机制的认证方法。该方法面向在线签名序列中噪声干扰、特征冗余以及动态信息利用不足等情况，首先对原始签名序列进行平滑处理，并结合特征重要性评估与主成分分析完成特征筛选与降维，构建同时包含全局统计属性和局部动态变化信息的输入表征。在特征提取阶段，采用Ghost特征映射对输入信息进行初始表征，并结合InceptionNext-TF模块与DASE模块开展多尺度深层特征建模，以刻画签名样本在不同尺度下的变化特征。在分类阶段，引入变分贝叶斯全连接层对输出权重进行分布化建模，并结合基于贝叶斯优化的自适应参数搜索策略，对相关超参数进行调整。基于MCYT-100和SVC-2004 Task2公开数据集的实验结果表明，在10-shot条件下，等错误率分别为1.46%和3.05%。 |
| 关 键 词： | InceptionNext-TF; 变分贝叶斯全连接层; 基于贝叶斯优化的自适应参数搜索策略 |
| DOI： | 10.57237/j.cst.2026.02.003 |
View:
Downloads:
收稿日期: 2026-03-25
接受日期: 2026-04-17
在线出版日期: 2026-04-27
Ghost Multi-scale Bayesian Network for Online Handwritten Signature Verification
沈阳建筑大学, 计算机科学与工程学院, 辽宁 110168,
| Abstract: | For the problems of limited feature representation, unstable parameter learning, and insufficient reliability of classification decisions in online handwritten signature verification under small-sample conditions, a verification method integrating lightweight representation, multi-scale feature modeling, and probabilistic discrimination is proposed. Aiming at noise interference, feature redundancy, and insufficient utilization of dynamic information in online signature sequences, the method first performs smoothing on the original signature sequences, and then combines feature importance evaluation with Principal Component Analysis to conduct feature selection and dimensionality reduction, thereby constructing an input representation that contains both global statistical attributes and local dynamic variation information. In the feature extraction stage, Ghost feature mapping is adopted for the initial representation of input information, and the InceptionNext-TF module together with the DASE module is used for multi-scale deep feature modeling to characterize variation patterns of signature samples at different scales. In the classification stage, a variational Bayesian fully connected layer is introduced to model the output weights in a distributional manner, and a Bayesian optimization-based adaptive parameter search strategy is further employed to adjust the relevant key hyperparameters. Experimental results on the public MCYT-100 and SVC-2004 Task2 datasets show that, under the 10-shot setting, the Equal Error Rates are 1.46% and 3.05%, respectively. |
| Keywords: | InceptionNext-TF; Variational Bayesian Fully Connected Layer; Bayesian Optimization-based Adaptive Parameter Search |
基金
| 1. | 国家自然科学基金 (62073227) |
| 2. | 辽宁省科技厅项目 (2023JH2/101300212) |
参考文献
| [1] | Tolosana R, Vera-Rodriguez R, Fierrez J, et al. Biometric signature verification using recurrent neural networks [C] // 2017 14th IAPR international conference on document analysis and recognition (ICDAR). IEEE, 2017, 1: 652-657. https://doi.org/10.1109/ICDAR.2017.112 |
| [2] | Ghosh R. A Recurrent Neural Network based deep learning model for offline signature verification and recognition system [J]. Expert Systems with Applications, 2021, 168: 114249. https://doi.org/10.1016/j.eswa.2020.114249 |
| [3] | Vorugunti C S, Mukherjee P, Pulabaigari V. Osvnet: Convolutional siamese network for writer independent online signature verification [C] // 2019 international conference on document analysis and recognition (ICDAR). IEEE, 2019: 1470-1475. https://doi.org/10.1109/ICDAR.2019.00236 |
| [4] | Ahrabian K, BabaAli B. Usage of autoencoders and Siamese networks for online handwritten signature verification [J]. Neural computing and applications, 2019, 31(12): 9321-9334. https://doi.org/10.1007/s00521-018-3844-z |
| [5] | Shavit H, Jatelnicki F, Mor-Puigventós P, et al. From xception to NEXcepTion: New design decisions and neural architecture search [J]. arXiv preprint arXiv: 2212.08448, 2022. https://doi.org/10.48550/arXiv.2212.08448 |
| [6] | Longjam T, Kisku D R, Gupta P. Writer independent handwritten signature verification on multi-scripted signatures using hybrid CNN-BiLSTM: A novel approach [J]. Expert Systems with Applications, 2023, 214: 119111. https://doi.org/10.1016/j.eswa.2022.119111 |
| [7] | Khoh W H, Ong T S, Pang Y H, et al. Score level fusion approach in dynamic signature verification based on hybrid wavelet‐Fourier transform [J]. Security and Communication Networks, 2014, 7(7): 1067-1078. https://doi.org/10.1002/sec.829 |
| [8] | Yang L, Cheng Y, Wang X, et al. Online handwritten signature verification using feature weighting algorithm relief: L. Yang et al [J]. Soft Computing, 2018, 22(23): 7811-7823. https://doi.org/10.1007/s00500-018-3477-2 |
| [9] | Vorugunti C S, Gorthi R K S, Pulabaigari V. Online signature verification by few-shot separable convolution based deep learning [C] // 2019 international conference on document analysis and recognition (ICDAR). IEEE, 2019: 1125-1130. https://doi.org/10.1109/ICDAR.2019.00182 |
| [10] | Vorugunti C S, Pulabaigari V, Gorthi R K S S, et al. OSVFuseNet: Online Signature Verification by feature fusion and depth-wise separable convolution based deep learning [J]. Neurocomputing, 2020, 409: 157-172. https://doi.org/10.1016/j.neucom.2020.05.072 |
| [11] | Tan H, He L, Huang Z C, et al. Online signature verification based on dynamic features from gene expression programming [J]. Multimedia Tools and Applications, 2024, 83(5): 15195-15221. https://doi.org/10.1007/s11042-021-11063-z |
| [12] | Zheng J, Chen Z, Huang L, et al. DWSCNN online signature verification algorithm based on CAE-MV feature dimensionality reduction [J]. IEEE Access, 2024, 12: 22144-22157. https://doi.org/10.1109/ACCESS.2024.3355449 |
| [13] | Doroz R, Kudlacik P, Porwik P. Online signature verification modeled by stability oriented reference signatures [J]. Information Sciences, 2018, 460: 151-171. https://doi.org/10.1016/j.ins.2018.05.049 |
| [14] | Tang L, Kang W, Fang Y. Information divergence-based matching strategy for online signature verification [J]. IEEE Transactions on Information Forensics and Security, 2017, 13(4): 861-873. https://doi.org/10.1109/TIFS.2017.2769023 |
| [15] | Diaz M, Ferrer M A, Quintana J J. Anthropomorphic features for on-line signatures [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 41(12): 2807-2819. https://doi.org/10.1109/TPAMI.2018.2869163 |
