Novel Approach of Offline Signature Verification Using Online Signature Database and Pre-Trained Deep Convolution Neural Network: SqueezeNet
| International Journal of Electronics and Communication Engineering |
| © 2025 by SSRG - IJECE Journal |
| Volume 12 Issue 12 |
| Year of Publication : 2025 |
| Authors : Bhimraj Prasai Chetry, Gunajyoti Das, Biswajit Kar |
10.14445/23488549/IJECE-V12I12P117
How to Cite?
Bhimraj Prasai Chetry, Gunajyoti Das, Biswajit Kar, "Novel Approach of Offline Signature Verification Using Online Signature Database and Pre-Trained Deep Convolution Neural Network: SqueezeNet," SSRG International Journal of Electronics and Communication Engineering, vol. 12, no. 12, pp. 201-217, 2025. Crossref, https://doi.org/10.14445/23488549/IJECE-V12I12P117
Abstract:
Signature is a vital behavioral biometric trait. It has been used for secure authentication for centuries. An innovative framework that converts dynamic online signature data from the SVC 2004 database into offline grayscale images has been proposed here. And finally, offline signature verification is done using a pre-trained lightweight CNN, SqueezeNet. Essential signer-specific patterns are preserved in the process of online-to-offline conversion. Before feeding the SqueezeNet input, the signature images undergo a preprocessing step that includes grayscale-to-RGB conversion and resizing. Subsequently, transfer learning is used to distinguish between genuine and forged signatures. By adopting this strategy, the model can be efficiently deployed in resource-constrained environments without sacrificing accuracy. It uniquely integrates online and offline signature verification. It also provides extensive threshold-based evaluation using various fundamental classification metrics, biometric-specific performance metrics, and ROC curve analysis. User-specific and Global Youden Thresholds, User specific and global EER threshold, Equal Error Rate (EER), and analysis of False Acceptance Rate (FAR) and False Rejection Rate (FRR) versus threshold are included in this study. Global ROC provides a Global Youden Threshold. And the average FAR and FRR curves vs threshold gives the global EER threshold. Offline signature verification under raw threshold, user specific Youden threshold, and user-specific EER thresholds is performed here. Excellent accuracy, flexibility, and robustness are seen here. The idea of offline signature verification derived from online data, when combined with compact CNN architectures, SqueezeNet, can bridge the gap between online and offline signature verification systems. This work contributes toward scalable, cross-domain biometric verification solutions and opens up ways towards unified signature recognition systems. This first-of-its-kind, cross-domain framework delivers a scalable, accurate, and resilient signature verification solution for both random and skilled forgeries. The proposed Offline Signature Verification system using online signature database achieves the best average testing accuracy of 99.81% (With User Specific Youden Thresholding) for random forgeries and 94.81% (With User Specific Youden Thresholding) for skilled forgeries across all 40 users. Here, the testing accuracy of random forgeries ranges from (92.50-100.00) %, and skilled forgeries ranges from (72.50-100.00) %. Hence, the proposed system yielded very good accuracy in comparison to existing state-of-the-art results, offering a practical solution to real-world applications.
Keywords:
Behavioral Biometric, CNN, Offline Signature Verification, SqueezeNet, SVC 2004 Database.
References:
[1] Moises Diaz et al., “A Perspective Analysis of Handwritten Signature Technology,” ACM Computing Surveys, vol. 51, no. 6, pp. 1-39, 2019.
[CrossRef] [Google Scholar] [Publisher Link]
[2] Rajesh Kumar, J.D. Sharma, and Bhabatosh Chanda, “Writer-Independent Off-line Signature Verification using Surroundedness Feature,” Pattern Recognition Letters, vol. 33, no. 3, pp. 301-308, 2012.
[CrossRef] [Google Scholar] [Publisher Link]
[3] Amir Soleimani, Babak N. Araabi, and Kazim Fouladi, “Deep Multitask Metric Learning for Offline Signature Verification,” Pattern Recognition Letters, vol. 80, pp. 84-90, 2016.
[CrossRef] [Google Scholar] [Publisher Link]
[4] C.R. Divyashri, and V. Nischitha, “Beyond the Pen: Deep Learning Advances in Offline Signature-Based Writer Identification and Verification,” International Journal on Science and Technology, vol. 16, no. 2, pp. 1-9, 2025.
[CrossRef] [Google Scholar] [Publisher Link]
[5] Forrest N. Iandola et al., “SqueezeNet: AlexNet-level Accuracy with 50x Fewer Parameters and <0.5MB Model Size,” arXiv Preprint, pp. 1-13, 2016.
[CrossRef] [Google Scholar] [Publisher Link]
[6] Yosepin Petra Purbanugraha, Adian Fatchur Rochim, and Iwan Setiawan, “Improvement Accuracy Identification and Learning Speed of Offline Signatures Based on SqueezeNet with ADAM Backpropagation,” 2022 9th International Conference on Information Technology, Computer, and Electrical Engineering (ICITACEE), Semarang, Indonesia, pp. 248-253, 2022.
[CrossRef] [Google Scholar] [Publisher Link]
[7] Luiz G. Hafemann, Robert Sabourin, and Luiz S. Oliveira, “Offline Handwritten Signature Verification — Literature Review,” 2017 Seventh International Conference on Image Processing Theory, Tools and Applications (IPTA), Montreal, QC, Canada, pp. 1-8, 2017.
[CrossRef] [Google Scholar] [Publisher Link]
[8] Luiz G. Hafemann, Robert Sabourin, and Luiz S. Oliveira, “Learning Features for Offline Handwritten Signature Verification using Deep Convolutional Neural Networks,” Pattern Recognition, vol. 70, pp. 163-176, 2017.
[CrossRef] [Google Scholar] [Publisher Link]
[9] Xianmu Cairang et al., “Learning Generalisable Representations for Offline Signature Verification,” 2022 International Joint Conference on Neural Networks (IJCNN), Padua, Italy, pp. 1-7, 2022.
[CrossRef] [Google Scholar] [Publisher Link]
[10] Yongliang Zhang et al., “Cross Layer Weakly Supervised Data Augmentation Network for Offline Signature Verification,” 2024 International Joint Conference on Neural Networks (IJCNN), Yokohama, Japan, pp. 1-8, 2024.
[CrossRef] [Google Scholar] [Publisher Link]
[11] Muhammad Azi Saputra, and Ida Nurhaida, “Signature Originality Verification Using A Deep Learning Approach,” Electronic Journal of Education Social Economics and Technology, vol. 5, no. 1, pp. 19-29, 2024.
[CrossRef] [Google Scholar] [Publisher Link]
[12] Neha Sharma, Sheifali Gupta, and Puneet Mehta, “A Comprehensive Study on Offline Signature Verification,” Journal of Physics: Conference Series: International Virtual Conference on Intelligent Robotics, Mechatronics and Automation Systems 2021, Chennai, India, vol. 1969, pp. 1-17, 2021.
[CrossRef] [Google Scholar] [Publisher Link]
[13] Tom Fawcett, “An Introduction to ROC Analysis,” Pattern Recognition Letters, vol. 27, no. 8, pp. 861-874, 2006.
[CrossRef] [Google Scholar] [Publisher Link]
[14] Jack Hogan, and Niall M. Adams, “On Averaging ROC Curves,” Transactions on Machine Learning Research, pp. 1-12, 2023.
[Google Scholar] [Publisher Link]
[15] S. Rashidi, A. Fallah, and F. Towhidkhah, “Estimation of the Youden Index and its Associated Cutoff Point,” Biometrical Journal: Journal of Mathematical Methods in Biosciences, vol. 47, no. 4, pp. 458-472, 2005.
[CrossRef] [Google Scholar] [Publisher Link]
[16] S. Rashidi, A. Fallah, and F. Towhidkhah, “Feature Extraction based DCT on Dynamic Signature Verification,” Scientia Iranica, vol. 19, no. 6, pp. 1810-1819, 2012.
[CrossRef] [Google Scholar] [Publisher Link]
[17] Ibtisam Ghazi Nsaif et al., “A Review of Online Signature Recognition System,” Fusion: Practice and Applications, vol. 18, no. 1, pp. 130-144, 2025.
[CrossRef] [Google Scholar] [Publisher Link]
[18] Donato Impedovo, and Giuseppe Pirlo, “Automatic Signature Verification: The State of the Art,” IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), vol. 38, no. 5, pp. 609-635, 2008.
[CrossRef] [Google Scholar] [Publisher Link]
[19] Jiaxin Lu et al., “Research on Authentic Signature Identification Method Integrating Dynamic and Static Features,” Applied Sciences, vol. 12, no. 19, pp. 1-17, 2022.
[CrossRef] [Google Scholar] [Publisher Link]
[20] Aman Singla, and Ajay Mittal, “Exploring Offline Signature Verification Techniques: A Survey Based on Methods and Future Directions,” Multimedia Tools and Applications, vol. 84, pp. 2835-2875, 2025.
[CrossRef] [Google Scholar] [Publisher Link]
[21] Mustafa Berkay Yilmaz et al., “Offline Signature Verification using Classifier Combination of HOG and LBP Features,” 2011 International Joint Conference on Biometrics (IJCB), Washington, DC, USA, pp. 1-7, 2011.
[CrossRef] [Google Scholar] [Publisher Link]
[22] Srikanta Pal et al., “Performance of an Off-Line Signature Verification Method Based on Texture Features on a Large Indic-Script Signature Dataset,” 2016 12th IAPR Workshop on Document Analysis Systems (DAS), Santorini, Greece, pp. 72-77, 2016.
[CrossRef] [Google Scholar] [Publisher Link]
[23] S. Singh, S. Chandra, and Agya Ram Verma, “Enhancing Offline Signature Verification via Transfer Learning and Deep Neural Networks,” Augmented Human Research, vol. 9, 2024.
[CrossRef] [Google Scholar] [Publisher Link]
[24] S.D. Bhavani, R.K. Bharathi, and R.J. Anil Kumar, “Offline Signature Verification using Pre-Trained Deep Convolutional Neural Network,” AIP Conference Proceedings: Emerging Trends in Signal Processing, Instrumentation, Power, Control, and Automation Systems, Subang Jaya, Malaysia, vol. 2966, no. 1, 2024.
[CrossRef] [Google Scholar] [Publisher Link]
[25] Bahar Ciftçi, and Ramazan Tekin, “Deep Learning Based Offline Handwritten Signature Recognition,” Bitlis Eren Üniversitesi Fen Bilimleri Dergisi, vol. 13, no. 3, pp. 871-884, 2024.
[CrossRef] [Google Scholar] [Publisher Link]
[26] Sara Tehsin et al., “Enhancing Signature Verification Using Triplet Siamese Similarity Networks in Digital Documents,” Mathematics, vol. 12, no. 17, pp. 1-15, 2024.
[CrossRef] [Google Scholar] [Publisher Link]
[27] Victor K.S.L. Melo et al., “Deep Learning Approach to Generate Offline Handwritten Signatures Based on Online Samples,” IET Biometrics, vol. 8, no. 3, pp. 215-220, 2019.
[CrossRef] [Google Scholar] [Publisher Link]
[28] Huan Li, Ping Wei, and Ping Hu, “Static-Dynamic Interaction Networks for Offline Signature Verification,” Proceedings of the AAAI Conference on Artificial Intelligence, vol. 35, no. 3, pp. 1893-1901, 2021.
[CrossRef] [Google Scholar] [Publisher Link]
[29] Dit-Yan Yeung et al., “SVC2004: First International Signature Verification Competition,” Biometric Authentication, pp. 16-22, 2004.
[CrossRef] [Google Scholar] [Publisher Link]
[30] Dawid Najda, and Khalid Saeed, “Impact of Augmentation Methods in Online Signature Verification,” Innovations in Systems and Software Engineering, vol. 20, pp. 477-483, 2024.
[CrossRef] [Google Scholar] [Publisher Link]
[31] Suvarna Joshi, and Abhay Kumar, “Feature Extraction Using DWT with Application to Offline Signature Identification,” Proceedings of the Fourth International Conference on Signal and Image Processing 2012 (ICSIP 2012), pp. 285-294, 2013.
[CrossRef] [Google Scholar] [Publisher Link]
[32] Saeeda Naz, Kiran Bibi, and Riaz Ahmad, “DeepSignature: Fine-Tuned Transfer Learning Based Signature Verification System,” Multimedia Tools and Applications, vol. 81, no. 26, pp. 38113-38122, 2022.
[CrossRef] [Google Scholar] [Publisher Link]
[33] Min Hao et al., “SqueezeNet: An Improved Lightweight Neural Network for Sheep Facial Recognition,” Applied Sciences, vol. 14, no. 4, pp. 1-13, 2024.
[CrossRef] [Google Scholar] [Publisher Link]
[34] Yash Gupta et al., “Handwritten Signature Verification Using Transfer Learning and Data Augmentation,” Proceedings of International Conference on Intelligent Cyber-Physical Systems, Jaipur, India, pp. 233-245, 2022.
[CrossRef] [Google Scholar] [Publisher Link]
[35] Shih-Yin Ooi, Andrew Beng-Jin Teoh, and Thian-Songa Ong, “Offline Signature Verification through Biometric Strengthening,” 2007 IEEE Workshop on Automatic Identification Advanced Technologies, Alghero, Italy, pp. 226-231, 2007.
[CrossRef] [Google Scholar] [Publisher Link]
[36] Transfer Learning and Fine-Tuning, TensorFlow. [Online]. Available: https://www.tensorflow.org/tutorials/images/transfer_learning
[37] Oona Rainio, Jarmo Teuho, and Riku Klén, “Evaluation Metrics and Statistical Tests for Machine Learning,” Scientific Reports, vol. 14, pp. 1-14, 2024.
[CrossRef] [Google Scholar] [Publisher Link]