Image Processing Reference
In-Depth Information
Appendix F * : MATLAB ® Exercises
with COMSOL ® Data
As a supplement to the materials in this topic, the MATLAB ® and COMSOL ®
data files used to create most of the images presented are provided to the reader
for demonstration, experimentation, and further development. This COMSOL
data was generated using the method described in Chapter 5. The MATLAB
code incorporates many of the theory and techniques discussed throughout
this topic. The main program has been broken down into smaller programs to
help focus on different aspects of the code. The first program is called ewald
and is used to create 2-D Ewald diagrams from the COMSOL data files The
second is the program called born which is used to generate first Born approxi-
mation reconstructions using the provided COMSOL data files The third is
the program called Cepstrum , which is used as the next step in the imaging
process of applying a cepstrum filter to the reconstructed image files An addi-
tional version of this program is Cepstrum2 , which is the same as Cepstrum
but adds an additional reconstruction where the incident wave is subtracted in
cepstrum space for improved performance. The final program is called PDFT,
which is used to demonstrate the application of the Prior Discrete Fourier
Transform (PDFT) as described in Chapter 7 and Appendix C.
The following suggested exercises are included to help the reader use and
understand these codes and to see how they relate to the information pre-
sented in the text. In addition to the suggested scenarios that could be tried,
some direction is given for the assessment and interpretation of the results.
These exercises were developed to be used as exercises in a college course, but
are also helpful to the general reader in understanding the concepts discussed
in the topic, and to understand the MATLAB code and how it works.
ewald exercises
1. Use ewald in MATLAB to generate Ewald circles for various targets.
Do this several times varying the target, the permittivity, the number
of sources, and the number of receivers. In all cases, where does the
majority of useful data seem to reside? What does this suggest? If the
opposite were true, what might this suggest? What might do a bet-
ter job of filling the space, increasing the number of sources or the
number of receivers? How would the images change if the incident
frequency were increased or decreased? Why?
* Supplementary materials including MATLAB code for exercises are available on the topic's