Human facial neural activities and gesture recognition for machine-interfacing applications

Area of Science:

• Biomedical Engineering
• Neuroscience
• Human-Computer Interaction

Background:

• Human-machine interface (HMI) technology often uses neural activity for machine control.
• Facial electromyography (EMG)-based HMI systems have typically used a limited set of facial gestures.
• A need exists for more versatile facial gesture recognition in HMI applications.

Purpose of the Study:

• To develop a multipurpose interface for recognizing human facial gestures via neural and muscle activity.
• To identify the most accurate facial gestures for applications requiring up to eleven control commands.
• To enhance the flexibility and command capacity of facial EMG-based HMI systems.

Main Methods:

• Recorded eleven facial gesture electromyography (EMG) signals from ten volunteers.
• Filtered detected EMGs and extracted root mean square features.
• Trained and classified various gesture combinations using a Fuzzy c-means classifier.

Main Results:

• Evaluated multiple combinations of facial gestures for recognition accuracy.
• Identified gesture combinations with the highest accuracy within each group.
• Achieved an average recognition accuracy greater than 90% for selected combinations.

Conclusions:

• The proposed method effectively recognizes human facial gestures using EMG signals.
• Selected gesture combinations demonstrate high accuracy (>90%) for command control.
• The system offers a flexible and accurate solution for advanced HMI applications.