Improving the performance of hand posture classification by perimeter sensor with sEMG

This paper reports a work for improving the performance of the hand posture classifier by using perimeter of the human forearm. Misclassification occurs due to four factors: residual deformation of the muscle after the muscle activation, sensor locations, error from the voltage regulator and the transition state during posture change. In order to reduce the effect of the factors, a sensor location which gives low residual muscle deformation was selected and one channel of sEMG was employed. The least square regression was used for removing the transition state. The proposed method was verified through simulation with pre-acquired data sets.