Survey of the stability of uniqueness of muscle synergy patterns in handwritten signature over time

Biometric characteristics of the human body can play a decisive role in the accuracy of automatic signature verification systems due to their stability over time and resistance to variability in different conditions. In this study, the accuracy of an automatic handwritten signature verification system is checked during nine months. In this system, the electromyography (EMG) signals from the hand muscles of people during signing are recorded at different times up to nine months, and after the pre-processing of the signals, muscle synergy patterns are extracted by the none-negative matrix factorization (NMF) method. Finally, the patterns extracted by the SVM classifier are classified into two classes: genuine and forgery signatures.

Keywords:

pattern recognition , EMG , SVM Classifier , Signature verification