Real-time control of finger and wrist movements in a virtual hand using traditional features of sEMG and Bayesian classifier

In this study, we present a real-time system to control a virtual hand using traditional features of surface electromyography (sEMG). The sEMG signal was recorded while performing simple finger and wrist movements related to the day-to-day activities. Traditional features of sEMG: RMS (Root Mean Square), VAR (Variance) and WL (Waveform Length) were computed using the sliding window technique. These features were classified using two types of classifiers: k-Nearest Neighbor (k-NN) and Bayesian (Discriminant Analysis). These classified patterns were used to control the designed virtual hand. This proposed system for controlling virtual hand can provide a better training and visual feedback to people with disability and for amputees.