Improved single-trial signal extraction of low SNR events

Studies of Birch's outlier processing method reveal that addition of time-varying influence functions to the algorithm enhances the detection and extraction of low SNRs of highly variable signals from white and colored noise processes. Use of a time-invariant minimum-mean-square-error filter does not yield results comparable to the new process. Simulations reveal the process efficiency at levels above -7.5 db.