SVD-compress

Description

Final Project for MATH 545 at UMass Amherst.

Uses Singlar Value Decomposition on an Input Image and recomposes it with a different number of components in the diagonal

Warning: The script.sh will take a few hours to run.

Dependencies:

• numpy
• matplotlib
• pillow (Python Imaging Library)
• ffmpeg (for making animation from processed images)

Documentation

Process valley.jpg in steps of 1 upto 360 components:

python3 compress.py stock/valley.jpg 360 1

Make animation from these processed images: (only works if stepsize is 1)

ffmpeg -framerate 24 -i processed%00d.png -c:v libx264 -profile:v high -crf 2 -preset veryslow -pix_fmt yuv420p valley.mp4

Make csv data file from output image:

python3 sort.py valley

Make graphs from data:

python3 data_analysis.py data/valley.csv 0 361 20