Wavelet-Based Detrending for EMG Noise Removal

Myoelectric Signals (MES) have a long traditionwith regard to prostheses control. Due to the signals' nature, MES are prone to interference and noise. Various methods existfor preprocessing these signals before classification algorithmsto derive control information are applied. While these methodshelp to improve the source signals, parameters must be carefullyselected and implemented on a case-to-case basis. After presentingseveral noise removal methods and drawbacks, we introduce anovel approach by applying wavelet detrending to the signal.The approach brought forward yields an excellent signal-to-noiseratio and provides in some cases a complete removal of noiseinterference. Weak signals and muscle fatigue do not impactthe results. Besides serving as input for various classificationmethods, the detrended signal can also be directly used forimplementing robust control methods like Cookie Crusher orthreshold algorithms. A basic Cookie Crusher control modelwas chosen to verify the approach in comparison to traditionalamplitude level schemes. Results show that detrended signal datacan be utilized for reliable prosthesis control even for usersexhibiting low amplitude MES.