Related Experiment Video
Updated: Jul 5, 2025
08:15
Capturing Dynamic Finger Gesturing with High-resolution Surface Electromyography and Computer Vision
Published on: March 28, 2025
494
Simple recognition of hand gestures using single-channel EMG signals
Mina Pourmokhtari1, Borhan Beigzadeh1
1Biomechatronics and Cognitive Engineering Research Lab, School of Mechanical Engineering, Iran University of Science and Technology, Tehran, Iran.
Summary
This study classified finger movements using electromyography (EMG) signals and k-nearest neighbors (KNN). The combination of Mean Absolute Value (MAV), Maximum (Max), and Minimum (Min) features achieved the highest accuracy for prosthetic control.
Area of Science:
- Biomedical Engineering
- Signal Processing
- Rehabilitation Technology
Background:
- Electromyography (EMG) signals measure skeletal muscle electrical activity.
- Surface EMG (SEMG) is a non-invasive technique for controlling upper limb prosthetics.
- Biosignals, like EMG, facilitate neural interfacing with computers.
Purpose of the Study:
- To classify five distinct finger movements using SEMG signals.
- To evaluate the effectiveness of the k-nearest neighbors (KNN) algorithm for gesture classification.
- To identify optimal time-domain features for accurate SEMG-based gesture recognition.
Main Methods:
- Collected SEMG data from four channels during five distinct finger movements.
- Extracted time-domain features: Maximum (Max), Minimum (Min), Mean Absolute Value (MAV), Root Mean Square (RMS), and Simple Square Integral (SSI).
- Employed the k-nearest neighbors (KNN) classifier to categorize the finger movements based on extracted features.
Main Results:
- The combination of MAV, Max, and Min features yielded the highest classification accuracy.
- Classification accuracies across the four channels using (MAV, Max, Min) were 91.0%, 89.9%, 89.8%, and 96.0%.
- The (MAV, Max, Min) feature set proved most effective for classifying finger gestures.
Conclusions:
- The proposed feature set and KNN classifier demonstrate high accuracy for SEMG-based finger movement classification.
- This approach shows potential for developing advanced, intuitive control systems for upper arm prostheses.
- Further research can explore additional features and advanced machine learning models for improved prosthetic control.
