On the use of Recurrence

The characterization of the signals in terms of linear/ nonlinear and deterministic/stochastic nature may be appropriated for knowledge extraction and machine learning algorithms. Some real-world signals exhibit signal modality characteristics that are strongly related with the physical mechanisms involved in the production of them. As an example, we can mention how the resonant sounds are more predictable than the vibrating sounds, or how the determinism of ultrasonic scans decreases as the porosity of scattering materials increases. In these situations the use signal modality derived parameters may have an easier interpretation than frequency or statistical derived parameters. In this work, we have chosen the Recurrence Plots and the Recurrence Quantification Analysis as a feasible alternative to extract signal modality information. We have successfully applied this technique in a couple of situations: characterization of maritime mammals calls, and measuring the porosity of materials by means of ultrasonic