Influence of the Scale Function on Wavelet Transformation of the Surface Electromyographic Signal
Abstract
The scale function in wavelet transformation (WT) determines wavelet dilation and optimises the processing of a given signal. Here, the objective was to determine the influence of the scale function on the WT of 160 surface electromyograms using second-degree polynomial (WT(poly)) and exponential (WT(exp)) scale functions. For each WT, a mean frequency (MNF) was calculated from the original wavelet spectrum and from the cubic spline interpolated wavelet spectrum, and these were compared with the MNF obtained from a fast Fourier transform (FFT). The total intensity (Tp) for each WT was compared with the root mean square (RMS). The MNFs computed from the original wavelet spectra were significantly (P < 0.05) lower and higher when computed from the reconstructed wavelet spectra than those from the FFT. The Tp computed from WT(poly) showed significantly higher agreement with the RMS than the Tp from WT(exp). Finally, the WT(poly) may serve as a reference in electromyography.