Denoise-cli

A command-line utility for running a multichannel denoising algorithm (from image-recovery).

How to use:

You can check the necessary input parameters at any time by running:

denoise-cli --help

Basically, you need to supply:

• -i an [i]nput image,
• -o the directory where you want the [o]utput images to be,
• -s a [s]tarting value for λ,
• -e an [e]nding value for λ,
• -t how many values of λ should be used (s[t]eps),
• -m the [m]aximum amount of iterations to run for each value of λ,
• -c the [c]onvergence threshold for exiting the algorithm.

Optionally you may supply the verbosity level of the output:

• -v for WARN,
• -vv for INFO,
• -vvv for DEBUG,
• -vvvv for TRACE,

You can do that like so:

denoise-cli -vv -i angry_birb_noisy.png -o . -s 0.001 -e 0.08 -t 20 -m 1000 -c 10e-10

• This will produce 20 images, the first using λ = 0.001 and the last using λ = 0.08, with the images in between using intermediary values of λ spread geometrically.

Have fun! ✨

Example:

Running:

./denoise-cli -i birb_noisy.png -o . -s 0.001 -e 0.08 -t 5 -m 1000 -c 10e-10

Results in:

λ = 0.0010000000	λ = 0.0029906976	λ = 0.0089442719	λ = 0.0267496122	λ = 0.0800000000

Input image source: birb_noisy.png by Markus S. Juvonen under License CC-BY-NC 4.0. Gaussian noise was added to the original image using GIMP.

Resulting images found here. Copyright by Markus S. Juvonen under License CC-BY-NC 4.0. Gaussian noise was added to the original image using GIMP; Noisy image was denoised using the denoising algorithm from the image-recovery library.

Copyright

This code is licensed under the GNU Affero General Public License version 3 or later. See LICENSE or https://www.gnu.org/licenses/agpl-3.0.en.html.

Acknowledgements

This is a CLI wrapper for the denoising algorithm in the library image-recovery: code by Lílian Ferreira de Freitas, mathematics by Emilia L. K. Blåsten