EE381K-14 Multidimensional Digital Signal Processing - Projects
Contents:
objective -
guidelines -
ideas -
format
Projects:
Spring 2008 -
Spring 2005 -
Spring 2003 -
Fall 2000 -
Fall 1998 -
Fall 1996
The primary outcome of this course will be a project, which all totaled
will count for half of the final grade.
The focus of your project should be the theory, algorithms, and design
of multidimensional signal processing systems.
The field of multidimensional signal processing is quite broad.
It includes the various fields of still image processing, such
as enhancement, restoration, halftoning, scanning, interpolation,
segmentation, and compression.
It also includes image sequence processing, such as
seismic data processing, video processing, and biomedical imaging.
It also includes spatial array signal processing, such as
radar and sonar signal processing, and multichannel signal processing,
such as surround sound spatial audio signal processing.
The project will be split into a literature survey and a research report.
The literature survey, which is due in the middle of the semester,
will summarize and evaluate at least three key journal papers
on a particular topic.
The research report, which is due at the end of the semester,
will describe the problem you have addressed, previous
approaches to the problem, the original research you conducted
over the semester, and an implementation of the key ideas.
A successful project will demonstrate a grasp on the literature on
the subject and familiarity with relevant software.
The research report for a successful project will likely be of
sufficient quality for submission to an IEEE conference.
The class is well-timed to submit results to the
IEEE Asilomar Conference
on Signals, Systems, and Computers because its submission deadline
is June 1st.
I've compiled a list of deadlines for
upcoming conferences to which you might consider submitting your final report.
These guidelines are taken from Prof. Edward A. Lee's guidelines for
projects in his
Specification
and Modeling of Reactive Real-Time Systems
course at the University of California at Berkeley:
- Original references should be cited (and even read and understood).
- The project should reflect a serious effort to go beyond the course
material, obtaining additional sources from the net, journals, or books.
- All prior work, published or not, public domain or proprietary,
should be fully credited ("my officemate, Joe Schmo, says ...",
"This section of code is modified from XXX gotten from YYY").
- Do not build software from scratch. Your project will not be
evaluated on the basis of how much effort you put into it, but rather
on how effective your work is. Go to the net or commercial software
and find something to build on.
- Learn to use the relevant computer languages, at least to the level of
proficiency required to make your point. Get the compiler, simulator,
design environment, and install it.
- If you are already engaged in relevant research, leverage it.
A good starting place for ideas for the project is the required
and supplemental books for the course.
Once you are considering a particular topic, the next step is
to read the relevant sections of textbooks and skim a few of the
papers referenced.
Two key Web sites for researching papers are:
- IEEE
Explore Online Publications Database has record of all IEEE conference
and journal papers since 1950, including text abstracts and
PDF versions of the complete paper.
- INSPEC database,
which lists titles and abstracts of technical papers from all
disciplines published since 1969.
The seminal work in a particular field will likely have occurred
in the 1960s or 1970s, or have its roots in key publications in
the 1960s or 1970s.
Projects based on applications follow:
- Optimal multidimensional sampling with application to light
fields (image-based rendering).
The following report is a good starting point:
C. Zhang and T. Chen,
"Generalized Plenoptic Sampling",
Carnegie Mellon Technical Report AMP01-06, Sep. 2001.
The report amounts to finding optimal multidimensional sampling
strategies. The technical report builds on work on multidimensional
resampling, e.g. Section B of the technical report will be covered
lectures 5-7 in the class.
The technical report builds on papers by Prof. Tsuhan Chen himself
as well as papers M-0 in the reader.
The application of the technical report is light fields.
In addition to Prof. Tsuhan Chen's work, Prof. Robert Gray
and Prof. Bernd Girod at Stanford University have been
reaching compression of light fields.
- Multidimensional codes for modulation in a communication system.
A coset code is an example.
We'll cover the concept of a coset in the lecture 7.
Multidimensional codes enable the transmission of a fractional
number of bits per symbol.
This would be a good project if you've already taken a graduate
course in Digital Communications and a graduate course in
Information Theory.
There has been fundamental work done in this area by Forney:
- G. D. Forney, Jr. and L.-F. Wei,
"Multidimensional
constellations. I. Introduction, figures of merit, and
generalized cross constellations",
IEEE Journal on Selected Areas in Communications,
vol. 7, no. 6, Aug. 1989, pp. 877-892.
- G. D. Forney, Jr.,
"Multidimensional constellations. II. Voronoi constellations",
IEEE Journal on Selected Areas in Communications,
vol. 7, no. 6, Aug. 1989, pp. 941-958.
- G. D. Forney, Jr., M. D. Trott, and Sae-Young Chung,
"Sphere-bound-achieving coset
codes and multilevel coset codes",
IEEE Trans. on Information Theory,
vol. 46, no. 3, May 2000, pp. 820-850.
- A study of the emerging joint ITU-ISO H.264/MPEG 4 Advanced Video
Coding standard for video communications,
algorithm improvements to improve implementation in software, and
integration of the new algorithms in public domain C code
for a standard-compliant video codec.
- A comparison of still image compression standards in terms of
subjective quality (not PSNR or some other pixel-based measure)
and at least one idea for improving subjective quality in
still image compression with an implementation.
The comparison should include JPEG 2000.
- M. Gormish and M. Marcellin,
"The JPEG 2000 Standard", invited presentation,
Data Compression Conference, 2000.
- N. Damera-Venkata, T. D. Kite, W. S. Geisler, B. L. Evans, and
A. C. Bovik,
"Image Quality Assessment Based on a Degradation Model",
IEEE Trans. on Image Processing, vol. 9, no. 4,
pp. 636-650, Apr. 2000.
- M. D. Adams and F. Kossentini, "Reversible Integer-to-Integer
Wavelet Transforms for Image Compression: Evaluation and
Analysis," IEEE Trans. on Image Processing,
vol. 9, no. 6, pp. 1010-1024, June 2000.
A project on algorithm development follows.
- A study of algorithms for active machine vision,
and development and implementation of a new algorithm.
A project with a CAD emphasis follows.
- A study of multidimensional dataflow models for
non-separable multidimensional systems and a prototype
implementation of a scheduler for a multidimensional dataflow model.
Active related research groups at UT Austin:
Last updated 02/16/08.