A Novel sEMG Denoising Method Based on Multi-Level Improved Wavelet Transform

Inhalt:
Surface EMG plays an important role in muscle health assessment and motion recognition. However, the collected EMG signals are often polluted by different noises. The superposition of these different components leads to the deviation in the understanding of the nature of EMG signals. Therefore, this paper proposes a Multi-Level Improved Wavelet Transform method to remove other noises from raw sEMG signals. The method firstly analyzed the components of the original signal. Then, decomposed the source signal by improved wavelet transform with multi-level threshold, and removed the noise components by threshold judgment. Finally, the signal was reconstructed to obtain the pure sEMG signal. Experimental results show that this method is superior to traditional methods such as wavelet transform, digital filter, and moving average filter. It lays a foundation for signal feature extraction, recognition, and wearable device control.