Wigner-Ville Distribution with a Mask Filter

Abstract

Summary The Wigner-Ville distribution is a powerful high resolution time-frequency algorithm used for nonstationary signals. Yet, it suffers from cross-term interference that occurs between the components of a signal. In spite of success in using smoothing windows to smooth out the interference, in both time and frequency directions, loss in resolution is inevitable. In this research, we present an adaptive Wigner-Ville distribution algorithm that modifies the Wigner- Ville method by applying a masking filter to its kernel. The algorithm results in eliminating crossterms and spurious energy while preserving the high energy precision portrayed by the Wigner-Ville method. The masking filter adapts to the shape of a reference data set and is implemented accordingly to filter the amplitude spectrum of the original data set. The filtering algorithm is an iterative process that is used repeatedly to enhance the sharpness of the spectrum. The method exploits the ability of the smoothed-pseudo Wigner-Ville in eliminating cross-terms by applying it as a masking filter on the high resolution Wigner-Ville distribution. We implement two existing mask filters and propose a new mask filter. Our results show that the new algorithm can robustly eliminate cross-terms and preserve highly localised auto-terms, which results in a high-resolution time-frequency spectrum.