Double-Threshold sEMG Onset Detection with Median-MAD Baseline Estimation
34th Signal Processing and Communications Applications Conference, SIU 2026, İstanbul, Türkiye, 7 - 10 Temmuz 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/siu71813.2026.11636592
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Anahtar Kelimeler: baseline estimation, double-threshold detection, median absolute deviation, onset detection, robust statistics, surface electromyography
- İstanbul Medipol Üniversitesi Adresli: Evet
Özet
Surface electromyography (sEMG) onset detection is a fundamental task in biomechanical analysis, motor control research, and human-machine interfacing. Although threshold-based detection methods are widely used due to their simplicity and computational efficiency, their reliability critically depends on accurate statistical modelling of baseline EMG activity. Conventional estimators based on the mean and standard deviation are highly sensitive to outliers and non-Gaussian noise components frequently present in baseline EMG segments, resulting in unstable threshold placement and reduced detection accuracy. This study proposes a robust statistical baseline modelling approach for a classical double-threshold EMG onset detection framework. Baseline central tendency and dispersion are estimated using the median and median absolute deviation (MAD), which are inherently resistant to outliers. To ensure comparability with the standard deviation under Gaussian assumptions, the MAD is scaled using a consistency factor of 1.4826. The detection pipeline further incorporates Teager-Kaiser energy conditioning, RMS envelope extraction, and temporal validation constraints to improve detection stability. Comparative evaluation against a conventional mean-standard deviation baseline model demonstrates that the proposed robust estimation significantly reduces threshold sensitivity to transient artifacts and improves temporal onset detection accuracy. The proposed approach preserves the simplicity and real-time feasibility of classical amplitude thresholding while substantially enhancing robustness under realistic noise conditions.