2. Summary
● Motivation: Image analysis is a growing field of interest, especially given the
large file sizes that images are allowed to be
● Problem: Other operations can be implemented in UNIX-Based Image Analyst
(UIA), while others can be enhanced
● Solution: Implement new operations in UIA while improving/debugging
previous ones
4. Literature Sources (continued)
"The Hadamard Transform." Neuron. Web. 8 Oct. 2015. <http://neuron-ai.tuke.sk/hudecm/science/7/7.html>.
"What Is the Walsh-Hadamard Transform and What Is It Good For?" Stack Exchange. 12 Mar. 2012. Web. 7
Sept. 2015. <http://dsp.stackexchange.com/questions/1693/what-is-the-walsh-hadamard-transform-and-
what-is-it-good-for>.
5. Work Performed
• Walsh-Hadamard transformation (WHT) written
• Integrated WHT into UIA by the principle of separability
• Enhanced error checking
• Wrote Markdown file documenting UIA’s purpose and how each operation
works
• Make my transform perform the same as MATLAB’s
• Error analysis of my WHT implementation vs. MATLAB’s
6. Work Performed (Continued)
• Error analysis on segments of the image to better identify any sources of error
• Additional test cases generated and verified to be working
• Additional comments in the code written to assist first-time readers
• Wrote final report and prepared final presentation
7. Results – Basic Operation
Program run
with wanted
command
and
arguments
Argument
checking for
validity
Convert input
image(s) to
Portable
Graymap
Format
(PGM)
Operation run
with new
input images
Images
converted
back to
original type,
temp files
removed, free
unneeded
memory, etc.
9. Results – Image 1
From left to right: Original image, UIA output, MATLAB output
Mean and RMSE of error values: 0.000, 0.000 for the whole
image and each segment (4 segments chosen)
10. Results – Image 2
From left to right: Original image, UIA output, MATLAB output
Mean and RMSE of error values: 0.000, 0.000 for the whole
image and each segment (4 segments chosen)
11. Results – Image 3
From left to right: Original image, UIA output, MATLAB output
Mean and RMSE of error values: 0.000, 0.000 for the whole
image and each segment (3 segments chosen)
12. Conclusions
• Increased awareness of image manipulation methods and how to implement
them in a programming language
• Experience with professional documentation and debugging
• Experience with working on existing code
13. Pros/Cons
• Pros
• Added a new transformation to UIA
• Modernized and enhanced its pre-existing code
• Cons
• Code will soon become obsolete with new releases of UNIX and
ImageMagick
• Overhead with learning undocumented code
14. Post-Mortem
• Would have researched other image manipulation techniques more so they
could be better understood for improvement purposes
• Would have implemented additional operations for comparison purposes
15. Future Work
• Additional operations so UIA becomes a more complete library
• Make UIA error-free
• Open-sourcing the code so it can be accessible by the masses
• More statistical analysis between different methods for comparison purposes