<title>Comparison of time-frequency-based techniques for estimating instantaneous frequency parameters of nonstationary processes</title>

Abstract

The aim of this work is to contrast techniques used to estimate two instantaneous frequency parameters of the surface electromyographic (EMG) signal, the instantaneous median frequency and the instantaneous mean frequency, based on their estimation error. Three methods are compared: Cohen class and Cohen-Posch class time- frequency representations are used to compute both the above- mentioned instantaneous frequency parameters, and a cross-time- frequency based technique is adopted to derive the instantaneous mean frequency. The results demonstrate that the algorithm based on Cohen-Posch class transformations leads to a standard deviation of the instantaneous frequency parameters that is smaller than that obtained using Cohen class representations. However, the cross- time-frequency estimation procedure for instantaneous mean frequency produced the smallest standard deviation compared to the other techniques. The algorithms based on Cohen class and Cohen- Posch class transformations often provided a lower bias than the cross-time-frequency based technique. This advantage was particularly evident when the instantaneous mean frequency varies non-linearly within the epochs used to derive the cross-time- frequency representation of the surface EMG signal.© (1999) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.