Hand activity detection from electromyogram using instantaneous frequency

In this paper, we analyze the electromyogram signal for detection of activity in human hand. The ensemble empirical mode decomposition (EEMD) algorithm is used to represent the electromyogram signal in terms of intrinsic mode functions (IMFs). Then instantaneous frequency is estimated on the EEMD generated IMFs. Unlike Fourier frequency, instantaneous frequency is a function of time making it useful to analyze signals that have time-varying spectral properties, such as the electromyogram signals. We propose a novel mean instantaneous frequency parameter and compare its performance with conventional parameters namely mean absolute value and mean frequency for detection of muscle activity in human hand. Results for actual electromyogram signals obtained from an open source database and those recorded by us using the miBEAT setup are presented.