EFFECT OF OBSERVATION SYMBOL GRANULARITY ON SIGNATURE CLASSIFICATION ACCURACY
DOI:
https://doi.org/10.29121/shodhkosh.v3.i1.2022.6031Abstract [English]
Observation symbol granularity plays a crucial role in the accuracy of Hidden Markov Model (HMM)-based classification systems. In the context of online signature verification, the number of quantized observation symbols directly impacts the model’s capacity to capture subtle variations in user signatures. This paper investigates how varying symbol counts—from 200 to 750—influences classification accuracy, Equal Error Rate (EER), and convergence behavior. Using the SVC 2004 dataset and Hybrid Wavelet Transform (HWT)-derived pressure features, we analyze system performance across five symbol scaling intervals. Results indicate that a moderate symbol granularity (300–400) achieves optimal EER with efficient convergence and lower overfitting risks. These findings inform model tuning for signature-based biometric authentication systems.
References
Jain, A. K., Flynn, P., & Ross, A. A. (2011). Introduction to Biometrics. Springer. DOI: https://doi.org/10.1007/978-0-387-77326-1
Impedovo, D., & Pirlo, G. (2008). Automatic Signature Verification: State of the Art. IEEE Trans. SMC. DOI: https://doi.org/10.1109/TSMCC.2008.923866
Galbally, J., Marcel, S., & Fierrez, J. (2015). Biometric Antispoofing Methods: A Survey. IEEE TIFS. DOI: https://doi.org/10.1109/ACCESS.2014.2381273
Rabiner, L. R. (1989). A Tutorial on Hidden Markov Models. IEEE Proceedings. DOI: https://doi.org/10.1109/5.18626
Kholmatov, A., & Yanikoglu, B. (2005). Identity Authentication Using Online Signatures. Pattern Recognition Letters. DOI: https://doi.org/10.1016/j.patrec.2005.04.017
Ferrer, M. A., Galbally, J., & Alonso-Fernandez, F. (2020). Exploiting Explainable AI in Signature Verification. Pattern Recognition Letters.
Rattani, A., & Derakhshani, R. (2019). A Survey of Online Signature Verification. IEEE Access.
Yilmaz, O., et al. (2021). Pressure Analysis for Writer Identification on Tablets. Computers & Security.
Hassanat, A., & Jassim, S. (2022). Efficient HMM Estimation in Biometric Sequences. Expert Systems with Applications.
Rantzsch, H., et al. (2020). Deep Learning Signature Verification via Siamese Networks. Pattern Recognition Letters.
SVC 2004 Dataset: http://www.cse.ust.hk/svc2004/
Bharadi, V., & Chavan, M. (2015). Pressure-Driven Feature Selection in Signatures. ICCUBEA.
Kekre, H. B., & Bharadi, V. A. (2014). Hybrid Wavelets Using Orthogonal Transforms. Confluence.
Bishop, C. M. (2006). Pattern Recognition and Machine Learning. Springer.
Bilmes, J. (1998). A Gentle Tutorial on EM for HMMs. UC Berkeley.
Duda, R. O., Hart, P. E., & Stork, D. G. (2001). Pattern Classification. Wiley.
Marzinotto, S., et al. (2011). Evaluation of Signature Biometrics on Mobile Devices. IEEE BTAS.
Zhang, Z., et al. (2022). Adaptive Symbol Clustering for HMM-Based Handwriting Verification. IEEE Transactions on Biometrics.
Gupta, P., & Gupta, S. (2020). Comparative HMM and CNN-Based Signature Verification. IJCA.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2022 Dr. Vinayak A. Bharadi, Dr. Manoj Chavan
This work is licensed under a Creative Commons Attribution 4.0 International License.
With the licence CC-BY, authors retain the copyright, allowing anyone to download, reuse, re-print, modify, distribute, and/or copy their contribution. The work must be properly attributed to its author.
It is not necessary to ask for further permission from the author or journal board.
This journal provides immediate open access to its content on the principle that making research freely available to the public supports a greater global exchange of knowledge.
