Image to nutils topology

This reposity provides a tool to convert .png to Nutils readable topology data structure (i.e. analysis-suitable geometry).

Instructions for usage

• Name you image as <image_name>_<pixel_size>.png
• Save it in subfolder images/
• Create a .raw and .json wrapper from a png file
  • bash img2json.sh <image_name>_<pixel_size> (for example bash img2json.sh walle 35)
• Go to examples subfolder and run example_2d script
  • python3 example_2d.py fname=<image_name>_<image_size> (for example python3 example_2d.py fname=walle_35)

Topology-preservation

The tool contains a feature called topology preservation -- an alogrithm based on the paper. It can be activated with topopreserve=True. For example python3 example_2d.py fname=walle_35 topopreserve=True. Note that the sharpness of the reconstructed boundariese can be varied with maxrefine argument. For example python3 example_2d.py fname=walle_35 maxrefine=3.

Before topology preservation:

After topology preservation:

Example images:

Pacman 300px X 300px

Airfoil 100px X 100px

Rocket 100px X 100px

WALL-E 70px X 70px

Carotid aortary 32px X 32 px

Required (external) packages

• Nutils
• Scikit-mage