INTEGRATED SYSTEM FOR FORENSIC DOCUMENT ANALYSIS: MULTISCRIPT RECOGNITION AND HANDWRITTEN SIGNATURE VERIFICATION
Keywords:
Forensic Document Analysis, Multiscript Recognition, Handwritten Signature Verification, Machine Learning, Feature Extraction, Document ExaminationAbstract
BACKGROUND: Forensic document analysis plays a crucial role in verifying the authenticity of signatures and recognizing handwritten content. Traditional methods often struggle with diverse handwriting styles and various script forms, necessitating the development of integrated systems that enhance the accuracy and efficiency of document examination. This study introduces a novel approach to forensic document analysis by integrating multiscript recognition and handwritten signature verification into a unified system. METHODS: The integrated system employs a quantitative experimental design utilizing advanced machine learning algorithms for multiscript recognition and signature verification. The approach involves data collection from a diverse set of handwritten documents, including multiple scripts and signature variations. Feature extraction techniques are applied to identify key characteristics, such as slant, stroke width, and signature dimensions. The system uses supervised learning models, including Convolutional Neural Networks (CNNs) and Support Vector Machines (SVMs), to train on these features and improve the accuracy of both script recognition and signature verification. RESULTS: The integrated system demonstrated a high level of accuracy in recognizing multiple scripts and verifying handwritten signatures. Performance metrics indicate significant improvements in the correct classification of script forms and the detection of forged signatures compared to traditional methods. The system achieved an overall accuracy rate of 92% in multiscript recognition and 89% in signature verification. CONCLUSION: The integrated system effectively enhances forensic document analysis by combining multiscript recognition with signature verification. This approach provides a robust tool for forensic experts, improving the reliability of document examination and reducing the incidence of misidentifications. Future work will focus on expanding the system’s capabilities to include additional scripts and handwriting variations.
References
Schlapbach, A., & Bunke, H. (2006). Off-line writer identification using Gaussian mixture models. In International Conference on Pattern Recognition.
Bulacu, M., & Schomaker, L. (2007). Text-independent writer identification and verification using textural and allographic features. IEEE Transactions on Pattern Analysis and Machine Intelligence, 29.
Djeddi, C., & Souici-Meslati, L. (2010). A texture-based approach for Arabic writer identification and verification. In International Conference on Machine and Web Intelligence.
Djeddi, C., & Souici-Meslati, L. (2011). Artificial immune recognition system for Arabic writer identification. In International Symposium on Innovation in Information and Communication Technology.
Nosary, A., Heutte, L., & Paquet, T. (2004). Unsupervised writer adaption applied to handwritten text recognition. Pattern Recognition Letters, 37.
Garain, U., & Paquet, T. (2009). Off-line multi-script writer identification using AR coefficients. In International Conference on Document Analysis and Recognition.
Tolosana, R., Gonzalez-Sosa, E., Vera-Rodriguez, R., Fierrez, J., & Ortega-Garcia, J. (2023). Exploring Transformers for On-Line Handwritten Signature Verification. arXiv preprint arXiv:2307.01663.
Riba, P., Daza, L., Fornes, A., & Llados, J. (2019). Writer identification and verification in handwritten documents. Pattern Recognition Letters, 121, 50-59.
Zhu, L., Peng, X., Zhou, J., & Huang, Y. (2023). Handwritten Signature Recognition using Deep Learning. IEEE Transactions on Pattern Analysis and Machine Intelligence.
Khatun, I., Rahman, M. S., & Islam, M. T. (2022). Offline Signature Verification Using Deep Metric Learning. International Journal of Electrical and Computer Engineering, 12(4), 400-407.
