A Precise Muscle Activity Onset/Offset Detection via EMG Signal

In this study, we propose a muscle activity detection algorithm which reflects the activation state. Since muscle in outer part is related to extraction of fingers and inner part controls contraction of fingers, we use two channels of EMG from inner and outer part of wrist for muscle activity detection. By using two channels, we detect a contraction state which we defined as close state. For the muscle activity detection, binarized EMG, and polarized EMG is calculated, and detected through post-processor. For the performance evaluation, we compared the proposed algorithm to conventional method. The proposed algorithm showed 20ms lower estimate error than conventional method.