| dc.description.abstract | Ultrasonic imaging for a multi-layer medium is a common challenge in seismology,
medical diagnosis, and non-destructive testing. One application for multi-layer imaging
is ultrasonic immersion test where the material under test and transducer array
are immersed in water. The main imaging challenge in immersion test (or in any
multi-layer medium) is that since the sound wave propagates with different speeds
in different layers, the homogeneity assumption is not valid. Thereby calculating the
sound travel time for the backscattered received signal is complicated. In this dissertation,
we propose a new approach to model the array received signals in order to
image the material under test.
In the first approach, we propose a distributed reflector modeling approach to
characterize the interface between water and the solid as well as any crack inside the
solid test sample. This approach relies on incoherently distributed reflector modeling.
A distributed reflector can be modeled as infinitely many point sources located close
to each other. We use distributed reflector modeling in order to estimate the shape of
the reflectors. To do so, we present our data model in a two-dimensional coordinate
system, and then develop a covariance fitting based approach to parametric estimation
of the shape of the interface between the two media and that of a crack inside the
test material. Numerical computer simulations show the accuracy of the proposed
approach. However the proposed approach is a parametric localization method which
needs the repetition of the ultrasonic test.
In the second approach we present a data model to use for image reconstruction
of a multi-layer medium without need to repeat the ultrasonic experiment. In this
approach, we also use the spatially distributed source to model the interfaces between
the layers of a multi-layer medium. Then, based on the Huygens principle, we develop
a new array spatial signature for all the points inside a multi-layer medium. This
new array spatial signature can be used in existing imaging techniques including the
conventional beamforming technique, the MUSIC method, and the Capon algorithm
in order to image a multi-layer medium. These aforementioned three algorithms are
traditionally applied for a homogeneous medium where the sound velocity is constant
in the material under test. Numerical simulations as well as experimental data show
that the distributed reflector modeling outperforms other approaches such as rooted
mean square velocity.
In the third approach, to reduce the execution time for the imaging process, we
develop a Fourier-based imaging technique to estimate the scattering coefficient of
the points inside the second layer of a two-layer medium in order to obtain an image
of the region of interest. First, we use an approximation of the proposed data model
for the array backscattered signals due to the scattering of the point scatterers inside
the second layer of the material under test. Seeking the similarity with the definition
of Fourier transform, we propose a Fourier-based imaging algorithm, for imaging the
second layer of the material under test. In this proposed algorithm, the execution time
is considerably reduced compared to the three aforementioned imaging algorithms.
This proposed algorithm can be used in an online imaging process. | en |
