Abstract:

Although non-stable signals have some excellent local time-frequency properties, it is difficult to analyze their time-frequency properties by using some conventional signal processing methods.In order to solve these problems, a new timefrequency analysis approach based on generalized S-transform is introduced to non-stable signals processing in the 1D and 2D space. At the same time, in order to avoid the time complexities and computing difficulties in the direct 2D S-transform，the images are put into a series of 1D signals by choosing different scalable local time windows. Then, the algorithm simulation and time-frequency analysis are achieved through both basic S-transform and generalized S-transform that has an adjustable time-frequency resolution. At last, the results demonstrate that the generalized S-transform has more flexible time-frequency focusing property than S-transform.

Key words: S-transform,  , generalized , S-transform,  , time-frequency , analysis,  , local , timefrequency , spectrum,  , image- , processing