Abstract
The purpose of this paper is to develop a new signal subspace approach in time-frequency domain for the enhancement of the nonstationary signal in the presence of additive noise. A time-frequency filter based on Wigner distribution (WD) is proposed to solve this problem via a three - stage scheme so that the nonstationary signal and noise are represented in the time-frequency domain. Then the signal subspace in the time-frequency domain is detected by a proposed iterative algorithm. The optimal signal subspace in the time-frequency domain is also discussed in this study. Finally, the time-frequency projection filters are developed using the orthogonal projection on the eigenspace of the signal subspace in the time-frequency domain. While enhanced by the time-frequency projection filters, most of the noise outside the signal subspace in the time-frequency domain is removed.