Basic Hand Gestures Classification Based on Surface Electromyography.

Area of Science:

• Biomedical Engineering
• Machine Learning
• Human-Computer Interaction

Background:

• Surface electromyography (sEMG) is a valuable tool for analyzing muscle activity.
• Accurate hand gesture classification is crucial for advanced prosthetics and human-computer interfaces.
• Optimizing machine learning models is essential for improving classification performance.

Purpose of the Study:

• To develop an innovative classification system for hand gestures.
• To enhance the performance of Support Vector Machine (SVM) classifiers for sEMG data.
• To investigate the efficacy of the Cuckoo Search algorithm for optimizing SVM parameters.

Main Methods:

• Utilized 2-channel surface electromyography (sEMG) signals for hand gesture data acquisition.
• Implemented a Support Vector Machine (SVM) classifier for gesture recognition.
• Employed the Cuckoo Search swarm algorithm for kernel function and parameter optimization of the SVM.
• Compared the proposed optimized SVM system against standard SVM classifiers with various kernel functions.

Main Results:

• The proposed system, featuring an SVM optimized by the Cuckoo Search algorithm, demonstrated superior performance.
• Achieved a high average classification rate of 98.12% for hand gesture classification.
• The optimized SVM approach outperformed standard SVM classifiers in accuracy.

Conclusions:

• The developed sEMG-based hand gesture classification system offers high accuracy.
• The Cuckoo Search algorithm effectively optimizes SVM parameters for improved gesture recognition.
• This innovative approach holds significant potential for applications in prosthetics and human-computer interaction.