A blind denoising process with applications to underwater acoustic signals

In many applications, such as transient classification, underwater communication and positioning system, it could be of great interest to reduce the noise level of the received signal before processing. In most cases, the algorithms dedicated to this problem either required some a priori knowledge about the useful signal and/or noise characteristics, either made some not always verified assumptions about the disturbing term statistics. The purpose of this article is to present a new blind denoising approach, without any considerations about the signal of interest and the noise. Results obtained on real and synthetic data are proposed and discussed.