Abstract
A novel tool of bio signal processing is proposed to identify human muscle action through sEMG. The tool is based on Integration of continuous wavelet transforms, wavelet time entropy and wavelet frequency entropy to identify muscle actions through sEMG. The experiments are carried out on triceps, biceps and flexor digitorum superficial (FDS) muscles. sEMG signals are measured at different intensities of FDS muscle contractions in order to verify the consistency of results. By taking the average entropies and based on lowest average wavelet entropy, it is found in calibrated experiment that complex Shannon wavelet family is the best candidate to identify the muscle activities among: Derivative of Gaussians wavelet family, Derivative of Complex Gaussians wavelet family, Complex Morlet family, Symlets, Coiflets and Daubechies wavelet families. Moreover, the results are consistent over the time-variant signal. The results presented in this paper have futuristic engineering implication in biomedical engineering and bio-robotic applications. © 2011 Springer-Verlag.