A Feature Stability and Discriminability Based Weighting Method for sEMG Gesture Recognition under Varying Wearing Positions

Le résumé fourni par la source

The effect of wearing position variation on surface electromyography gesture recognition is investigated in this study. Experimental results show that changes in electrode position can alter the distribution of sEMG signal features and reduce recognition performance. To address this problem, a feature weighting method based on cross-position stability and inter-class discriminability is proposed. Time-domain features are extracted from sEMG signals and statistical features from inertial measurement unit (IMU) signals to construct multimodal feature vectors. The results indicate that features such as RMS and WL exhibit higher stability across positions, while ZC and SSC are more sensitive to position changes. Among these, RMS, WL, and IEMG consistently show high weights, whereas ZC and SSC exhibit lower and more variable weights, particularly between middle and distal positions. The proposed method improves the robustness of gesture recognition under varying wearing positions.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.