Classification of surface EMG signals using optimal wavelet packet method based on Davies-Bouldin criterion

Area of Science:

• Biomedical Engineering
• Signal Processing
• Machine Learning

Background:

• Surface electromyographic (sEMG) signals are crucial for controlling prosthetic devices.
• Accurate classification of sEMG signals is essential for intuitive prosthesis control.
• Existing methods face challenges in feature dimensionality and classification accuracy.

Purpose of the Study:

• To develop an optimal wavelet packet (OWP) method for sEMG signal classification.
• To improve the accuracy of discriminating between different prosthesis movements.
• To reduce feature dimensionality for efficient signal processing.

Main Methods:

• An optimal wavelet packet decomposition based on the Davies-Bouldin criterion was applied to sEMG signals.
• Principle components analysis (PCA) was used for feature dimensionality reduction.
• A neural network classifier was employed to discriminate four types of prosthesis movements.

Main Results:

• The proposed OWP method achieved a mean classification accuracy of 93.75%.
• This accuracy significantly outperformed the energy of wavelet packet coefficients method (86.25%) and the fuzzy wavelet packet method (87.5%).
• The combination of OWP, PCA, and neural network demonstrated superior performance.

Conclusions:

• The OWP method provides an effective approach for sEMG signal classification.
• This technique enhances the accuracy and efficiency of prosthesis movement discrimination.
• The findings suggest potential for improved human-machine interfaces in prosthetics.