A data-adaptive S-transform

The S-transform is a time frequency analysis technique combining properties of the short-time Fourier and wavelet transforms. It provides frequency-dependent resolution while maintaining g a direct relationship with the Fourier spectrum. However, the frequency resolution of S-transform in high-frequency is unsatisfactory. In this paper, we present a data-adaptive S-transform by optimizing the window width according to the measure of 'concentration'. The proposed method is tested on a set of synthetic signals. The result shows that the proposed algorithm achieves higher resolution and energy concentration than the original S-transform and short-time Fourier transform.