A Novel Local Transform Inverse S-Transform Algorithm for Statistical Filter

S-transform (ST) is a useful tool for time-frequency filter. However, the conventional inverse S-transform (IST) algorithm suffers from time or frequency leakage. In this paper, we proposed a novel local transform inverse S-Transform (LTIST) algorithm for statistical filter. First, the matrix S-transform (MST) and MIST are derived. Then the proposed LTIST approach applies to denoising. The statistical property of stochastic noise in the MIST is discussed. The results show that the proposed MIST algorithm has better time-frequency localization in statistical filtering than the conventional methods. Illustrative examples verify the effectiveness of the proposed algorithm.