Software System for Biometric Human Authentication Based on Handwritten Signature Dynamics

In this work, we present a software system for biometric authentication based on the dynamics of handwritten signatures. Our proposed approach focuses on analyzing dynamic signature characteristics, such as pen tip coordinates, pressure, azimuth, and tilt, captured using a graphics tablet. We employed normalization and interpolation techniques for time series data processing and feature extraction. A key feature of our developed method is the application of fuzzy set theory to model complex and subjective features like pressure patterns and writing rhythm. This allows for consideration of individual handwriting characteristics and improves authentication accuracy. The system analyzes a wide range of characteristics, including signature length, relative proportions (height-to-width ratio), average pressure, relative pen-up duration, pressure unevenness, and the area enclosed by the time series graph. The developed system was experimentally tested on the MCYT_Signature_100 database. The results demonstrate the high effectiveness of the proposed method, especially with limited training data. The proposed solution can be integrated into various security systems and applied in areas such as banking, electronic voting, access control systems, and document authorship verification.