EMG pattern classification using spectral estimation and neural network

In this paper, we propose a method of pattern recognition of EMG signals of hand gesture using spectral estimation and neural network. Proposed system is composed of the Yule-Walker algorithm and the LVQ. The use of the Yule-Walker algorithm is to estimates the power spectral density (PSD) of the signal. The spectral estimate returned is the magnitude squared frequency response of AR model. A fine tuning step will then be incorporated to improve the accuracy of the classification by way of the LVQ. We describe in detail the experiment conducted to verify the usefulness of the proposed method for EMG pattern classification of hand gesture.