Capture of the Voluntary Motor Intention from the Electromyography Signal

Abstract

The objective of this work is to automatically identify basic hand movements: Opening, Closing, Bending, Extension, Pronation and Supination, including the Resting condition. Feature extraction was implemented making use of three approaches: time, frequency and time-frequency domains, obtaining the characteristics Mean Absolute Value (MAV), Root Mean Square (RMS), Wave Length (WL), Autoregressive Coefficients (AR) and Discrete Wavelet Transform (DWT). Principal Component Analysis (PCA) was applied for dimensionality reduction and classification was performed using Linear Discriminant Analysis (LDA). As a result it was possible to identify the movements with success rates that reached 92% with the hybrid vectors conformed by the coefficients MAV, RMS and AR.