A performance of modern gesture control device with application in pattern classification

This research aimed to propose the performance testing of a modern gesture control device called MYO armband with an application in feature extraction based on multi-channels EMG. The EMG signal was collected from the forearm muscles during 6 gestures including hand close, hand open, hand flexor, double tap, and normal hand position. In this research, we applied the well-known feature extraction method called mean absolute value (MAV). The EMG features were represented in scatter diagrams to explain their behaviors. The well-known quantitative parameters used to evaluate the performance of EMG feature included davies-bouldin criterion (DB index) and scattering criterion. To present the quality of EMG signal, the signal to noise ratio (SNR), total harmonic distortion (THD), and power density spectrum (PSD) were used. The results showed that the EMG signal quality and EMG features extracted by MYO armband was robust and effective since the quantitative parameters were higher than the conventional EMG measurement system. The result was promising.