ADAPTIVE FILTERING OF NONSTATIONARY SIGNALS APPLIED TO THE EOG.

An adaptive filter which specializes in the cancellation of noise in nonstationary signals is described and applied to the processing of the saccadic electrooculogram (EOG). The usual adaptive filter is optimally corrected by using an a priori knowledge of the noise variance. The output of the corrected adaptive filter (CAF) is a weighted average of the usual adaptive filter output and the raw data. Analysis of the experimental results demonstrates the adequacy of the CAF for a accurate estimation of the saccade duration and peak saccade velocity.