Intrinsic mode entropy: an enhanced classification means for automated Greek Sign Language gesture recognition

Area of Science:

• Biomedical Engineering
• Human-Computer Interaction
• Linguistics

Background:

• Sign languages are crucial for deaf communication.
• Automated gesture recognition can bridge communication gaps between deaf and hearing individuals.
• Greek Sign Language (GSL) recognition presents unique challenges and opportunities.

Purpose of the Study:

• To develop an automated system for recognizing Greek Sign Language (GSL) gestures.
• To evaluate the effectiveness of intrinsic mode entropy (IMEn) for gesture classification.
• To identify optimal parameters for IMEn calculation in GSL gesture recognition.

Main Methods:

• Collected data from 3D accelerometers and surface electromyograms (sEMG) from the forearm.
• Applied intrinsic mode entropy (IMEn) analysis to sensor data for various window lengths.
• Utilized Mahalanobis distance and discriminant analysis to optimize IMEn parameters and feature selection.

Main Results:

• Achieved 100% classification accuracy for ten GSL gestures using IMEn as the sole feature.
• Identified effective intrinsic mode function scales and window lengths for IMEn calculation.
• Demonstrated the robustness of IMEn in distinguishing GSL gestures.

Conclusions:

• Intrinsic mode entropy (IMEn) is a highly effective feature for automated GSL gesture recognition.
• The proposed method offers a promising foundation for developing advanced GSL recognition systems.
• This technology has the potential to significantly improve communication tools for the deaf community.