Extraction of the Evoked Potentials Based on the Fourth Order Correlations and an Improved Threshold Function

摘要

According to the characteristics of restraining the additive Gaussian-noise and boosting up the faint signal of the higher-order correlations, this paper designed an adaptive filter which is based on a selected fourth order correlation slice (FOCS)of the input signal. When filtered by this filter, the noise of the noisy evoked potentials (Eps) signal can be eliminated distinctly.Contraposed the disadvantages of traditional threshold function.This paper also proposed an improved threshold function. This function not only has the smoothing property of the soft threshold function, but also has the keeping characteristics property of the hard threshold function. We extracted Eps by combining the adaptive filter with the improved threshold function, and the simulations results proved the method was quick and accurate.