The invention relates to a wave-propagation-based sounding method, in which: an incident wave is emitted through a medium (1) including diffusers (2) capable of reflecting the wave; and, subsequently, signals representing a reflected wave reverberated by the medium from the incident wave are captured, said captured signals being the sum of a single scattering component and a multiple scattering component. The captured signals are processed by separating the multiple scattering component from the single scattering component by filtering at least one frequency transfer matrix representing responses between transducers from the transducer assembly.