Cross-term Suppression in Wigner Distribution

Wigner Distribution, one of many methods to compute time frequency representation, satisfies large number of mathematical properties and gives optimal energy concentration in time frequency plane. However, Wigner distribution suffers from severe cross-term interference problem, which limits its scope for practical applications. Different modified versions of Wigner distribution have been developed to overcome its cross-term interference problem. Most of these techniques suppresses cross-term on the expense of quality of auto-terms. Schemes that compleltely remove cross-terms without affecting the resolution of Wigner distribution are computationally very expensive. This research proposes a computationally efficient solution to cross-term interference problem based on image processing and fractional filtering. Signal components are located in time frequency plane using image processing. Fractional filters are then applied to separate signal components. Wigner distribution of separated signal components is computed and added up to obtain cross-term free crisp time frequency representation. Performance of the proposed time frequency representation is evaluated both on synthetic and real life bat signals. One of the most significant application of Wigner distribution is instantaneous frequency estimation of signals. However, Wigner distribution based instantaneous frequency can only be applied to mono-component signal because of its crossterm problem. In this study we extend Wigner distribution based instantaneous frequency estimation scheme to multi-component signals having non-overlapping time frequency representation.