Feature Extraction and Classification of Hand Movements Surface Electromyogram Signals Based on Multi-method Integration

Abstract

On the basis of analysing the characteristics of hand movement surface electrocardiogram electromyogram (sEMG) signals, we propose a feature extraction and classification method for hand movement sEMG signals based on a multi-method integration combining the wavelet, fractal and statistics methods. To start, the hand movement sEMG signals are de-noised by using the wavelet transform, the de-noised and reconstructed signals are decomposed, and the average high frequency coefficients in each scale space are calculated to constitute the feature vectors as the first part of the hand movement sEMG signals classification features. Next, according to the characteristics of hand movement sEMG signals and the classification needs, we analyse the multi-fractal spectrum of the de-noised and reconstructed signals at multiple scales and extract the relevant parameters of multi-fractal spectrum as the second part of the hand movement sEMG signals classification features. Then, according to the characteristics of hand movement sEMG signals, we extract the relevant statistical characteristics of sEMG signals as the third part of hand movement sEMG signals classification features. According to the extracted features, we use the Least Square Support Vector Machine and the Backpropagation neural network as classifiers to individually classify and combine the characteristics of hand movement sEMG signals and the experimental results. The final classification features are identified to accomplish the classification of hand movement sEMG signals. Finally, the advantages of the proposed method are illustrated by comparative analysis from multiple perspectives.