Time-Frequency Methods for Multicomponents Signals

Instantaneous frequency (IF) is a very important parameter in a large number of applications. Generally, the IF is a non-linear function of time. For such cases the analysis of time-frequency content provides an efficient solution. This paper is a comparative study of the performances in IF estimation of the two time-frequency based methods. Limitations of the two methods are presented and some possible improvements. The first estimation method uses the complex argument distribution (CTD) and the second one uses the ridges extraction method from the time-frequency distribution based on mathematical morphology operators (TF-MO). Multicomponent signals with non-linear IF corrupted by Gaussian white noise are considered as numerical examples.