Image segmentation with Python

This project is a simple script that process a picture to simplify its color palette and obtain an image with less shading.

It is only an experiment, not a ready-to-use software.

Installation

For those who want to use

Go to the release page and follow the instructions to download and to install the module with pip.

Then, you will be able to type in a terminal:

• python -m picture2avatar [args...] to run the command line tool
• picture2avatar to launch the gui

You will also be able to use the module in other projects:

import picture2avatar
help(picture2avatar)

For those who want to contribute

git clone ...
pipenv install --dev

Then, edit files, add features...

To run unit tests and generate coverage info:

pipenv run pytest --cov=picture2avatar --cov-report html

To package the module:

pipenv run python setup.py sdist bdist_wheel

Usage

Apply an algorithm to a picture and store the result in a file:

# Example
pipenv run python picture2avatar.py input_picture.jpg pre -o result.jpg -p 10
# Help
pipenv run python picture2avatar.py -h
# GUI
pipenv run python guipicture2avatar.pyw

• If no output file is given, the picture is displayed on the screen.
• If no parameter is given to the algorithm ("-p value") the default one is used.

List of available algorithms

• pre: Performs a threshold on each band of the picture
  • optional parameter: number of 1 in the mask (e.g. 2 gives 0xC0, 3 gives 0xE0), default value 1
• grow: Performs a region growth algorithm on the image
  • optional parameter: maximal difference between the color of two pixel of the same region
• clu: Performs a clustering on the colors of the image
  • optional parameter: number of colors to keep at the end
  • ⚠️ it can be very long to process
• greed: Use a greedy algorithm to keep only a given number of colors of the picture (faster than clustering)
  • optional parameter: number of colors to keep at the end