Suppression of interference terms in Wigner distribution by median filtering

The Wigner distributions is known to have interfering cross-terms when applied to a multicomponent signal. Linear filtering techniques have been applied to suppress the interference but result in smearing of the time-frequency distribution. A nonlinear filtering technique is proposed that suppresses the interference without smearing the distribution. The technique consists of the median filtering of the cumulative distribution function and its performance is studied with computer simulated data. Good results were obtained even for a long data window.<>