A Signal-to-Noise Investigation of Nonlinear Electromyographic Processors

Abstract
A general class of nonlinear electromyographic (EMG) processors has been evaluated for their signal-to-noise ratio (SNR) properties. These filters incorporate as the detector element either a power-law (xn) or root-law (x1/n) transformation (detector) followed by a linear low-pass smoothing filter. ``Linearized'' versions of these processors in which the smoothing filter is followed by an element with the inverse detector characteristics were also evaluated.