Handwritten signatures verification based on arm and hand muscles synergy

Abstract

Intra-personal variability, which measures the difference between people's signatures, may be affected by various signing challenges. Considering the variety of physical conditions, it is almost impossible for persons to write their exact handwritten signature in the same way in several attempts. Also, the authentication system requires less complexity to respond quickly to real-time applications. This study attempts to confirm handwritten signatures by using hand muscle synergy as a biometric characteristic. To design the signature verification system, surface electromyography (EMG) signals from eight (arm and hand) muscles of the volunteers were recorded by surface EMG pads during the signing. Muscle synergy was extracted from EMG signals after preprocessing using the non-negative matrix factorization (NMF) method. Genuine and forgery data are then classified by the K-means classifier. The system achieves an equal error rate (EER) of 2.75 to identify the extracted data related to the genuine and forged signatures. © 2022