pretends to be a possible source of information about poros-
ity, ultrasonic frequencies above 1 MHz will be typical in
this case to generate signi®cant levels of grain noise. We
should then consider certain hypothesis, so as to make the
material model simple enough, and afterwards verify its
validity in experimental form. As said in Section 1, we
will assume a two-phase material, composed by a solid
phase and pores. Starting from this hypothesis and neglect-
ing multiple scattering, the registers of structural noise can
be assumed as the superposition of the echoes generated in
the scatters (Fig. 2). Additionally, it must be taken into
account that the ultrasonic pulse modi®es as it penetrates
the material. In the simplest case, it suffers attenuation,
though normally this attenuation differs in each one of the
pulse frequency components so as to deform the pulse. The
said deformation can also be related with porosity. In this
approximation, two types of work were carried out that will
now be described.
The ®rst work consisted in the extraction of statistical