Forearm Electromyogram-Based Biometrics Using Bag-of-Words Classifiers

The paper analyzes the efficiency of using electromyogram (EMG) signals acquired on the forearm as biometric information. Extensive tests have been conducted on a publicly available multi-channel multi-session dataset collected from 8 electrodes placed on the forearms of 43 persons while performing a set of 16 distinct gestures. FFT-based features extracted from 6 non-overlapping frequency bands were combined with a bag-of-words classifier and evaluated for both authentication and identification applications. The effect of using three distinct encoding strategies is assessed in terms of the Area Under Curve (AUC), Equal Error Rate (EER), and Cumulative Match Characteristics (CMC). The experimental results indicate the robustness of the approach against the encoding strategy, the codebook dimension, and the length of the gesture-based password.