Gestures Recognition Method Based on Electromyographic Signal

Different gestures were identified through analyzing and processing the electromyographic signal(EMGs) collected from the forearm. That in turn was used to control the upper limb rehabilitation equipment. The wavelet denoising was used after filtering the power frequency interference and the normalized processing. The high and low frequency coefficients were decomposed from signal through wavelet transform. The variance calculated from the frequency coefficients was used as a characteristic value. Through the neural networks classification, the recognition rates of seven kinds of gestures are over 99%. The seven kinds of gestures were wrist inward, wrist outward, fist stretch, fist clench, wrist up, wrist down and palm downward spiral.