The IMFogram: a New Time-Frequency Representation Method for Nonstationary Signals Analysis

In signal processing the time-frequency analysis of nonlinear and nonstationary processes, as well as the determination of the unknown number of active sub-signals of a blind-source composite signal are, in general, challenging tasks. Standard techniques, like short-time Fourier transform, and wavelet transform are limited in addressing the problem.An alternative approach is based on the simple but clever idea of first decomposing a signal into simple oscillatory components, using methods like Empirical Mode Decomposition-based techniques or Iterative Filtering methods, and then analyzing them separately in the time-frequency plane.

In this talk, we introduce a brand new method, called IMFogram, which allows us to produce accurate time-frequency representations of such simple oscillatory components. We provide proof that the IMFogram converges in the limit to the Spectrogram of a nonstationary signal.Furthermore, we compare its performance with Spectrogram, Continuous wavelets, and Synchrosqueezing transform methods, showing the ability of this newly proposed method to produce crisper and more detailed time-frequency representations of nonstationary signals.We conclude our presentation with some applications, including recent results in the study and analysis of Schumann resonances via China Seismo-Electromagnetic Satellite (CSES) program data sets.