Related Experiment Video
Updated: Jun 18, 2026
08:15
Capturing Dynamic Finger Gesturing with High-resolution Surface Electromyography and Computer Vision
Published on: March 28, 2025
Using sample entropy for automated sign language recognition on sEMG and accelerometer data
Vasiliki E Kosmidou1, Leontios I Hadjileontiadis
1Department of Electrical & Computer Engineering, Faculty of Engineering, Aristotle University of Thessaloniki, University Campus, 541 24, Thessaloniki, Greece. vkosm@auth.gr
Medical & Biological Engineering & Computing
|November 28, 2009
Summary
This study introduces sample entropy (SampEn) for automated Greek Sign Language (GSL) recognition, achieving 92% accuracy. This method surpasses traditional time-frequency features (TFF) for reliable GSL gesture recognition.
Area of Science:
- Biomedical Engineering
- Computer Science
- Linguistics
Background:
- Sign language (SL) is a vital communication method for the deaf.
- Automated SL gesture recognition can bridge communication gaps with hearing individuals.
- Developing efficient recognition systems is crucial for expanding SL accessibility.
Purpose of the Study:
- To evaluate sample entropy (SampEn) for automated recognition of Greek Sign Language (GSL) isolated signs.
- To compare SampEn-based feature extraction with traditional time-frequency features (TFF).
- To assess the potential for fast and reliable automated GSL gesture recognition.
Main Methods:
- Collected data from five-channel surface electromyogram and 3D accelerometer from signers' dominant hands.
- Applied sample entropy (SampEn)-based analysis for feature extraction.
- Utilized time-frequency feature (TFF) analysis as a baseline comparison method.
- Tested recognition accuracy on a 60-word lexicon of isolated GSL signs.
Main Results:
- SampEn achieved a mean classification accuracy of 92%.
- Time-frequency features (TFF) yielded a mean classification accuracy of 66%.
- SampEn demonstrated superior performance compared to TFF for GSL recognition.
- SampEn also facilitated feature vector dimension reduction.
Conclusions:
- Sample entropy (SampEn) is a superior method for automated Greek Sign Language (GSL) gesture recognition compared to TFF.
- SampEn offers high recognition accuracy and enables feature vector dimension reduction.
- The findings support the development of fast and reliable automated GSL recognition systems.
