Basic hand action classification based on surface EMG using autoregressive reflection coefficient

Surface electromyography (sEMG) signals play a major role in pattern recognition and prosthesis control. In this paper, an efficient scheme is proposed to classify basic hand movements from the sEMG signals based on autocorrelation domain feature extraction. At first a given frame of raw sEMG signal is divided into short duration sub-frames and each frame is subjected to autoregressive (AR) modeling. Instead of using the entire frame at a time, sub-frame based analysis is expected to provide consistent estimates and capture short duration variations. Although AR parameters are widely used as features, they do not have bounded values and matrix inversion is required to compute them. As an alternate we propose to use the AR reflection coefficients which are bounded between 0 to 1 and provide better consistency, noise immunity and lower computational complexity. The reflection coefficients obtained from each sub-frame are finally averaged to construct the proposed feature vector. For the purpose of classification, the K-nearest neighborhood (KNN) classifier is applied in a hierarchical approach. The proposed method is tested on a publicly available sEMG dataset containing six different hand movements collected from three females and two males. It is observed that the proposed method offers consistently a very high accuracy in classifying the six hand movements.