Fitting ellipsoids to 3D grains in RVE

This repository discusses how we can fit a 3D ellipsoid to an arbitrary 3D shape for RVE evaluation. The idea of this method is mostly based on this paper, so I would recommend reading it first (especially Appendix B). Here I extend that idea into fitting a 3D ellipsoid to an arbitrary 3D shape.

Source data

The RVE data used in this evaluation is first analyzed from MTEX and generated from DREAM3D, which is stored in fit-ellipsoid/data for reference.

Methodology

The report of the method is stored in fit-ellipsoid/report/Fitting an ellipsoid to an arbitrary 3D shape.pdf, with detail explanation on the mathematical motivation of the method.

Running the code

First, open Git Bash (on Windows, or just terminal on Linux) in a location you want to store the code and run (only for the first time):

git clone https://github.com/dinhtrung21/fit-ellipsoid.git

Then, on Git Bash we open VSCode to the folder:

code fit-ellipsoid

From this point we will work on VSCode terminal. Make sure that you have install Python for VSCode terminal, this can be easily checked by the command python3. Then, we install the needed packages using this line:

pip3 install –r requirements.txt

Finally, the automatic evaluation process can be initiated by running evaluate.py from the command line as following:

python3 evaluate.py

After the first time, you can run the code straight from the terminal on Windows (PowerShell) without using Git Bash.

Results

The results are stored in fit-ellipsoid/result for reference.

Development

• To adjust the RVE parameters or experimental constants, see evaluate.py.
• If you are interested in the algorithms of the fitting method, see ellipsoid.py.
• If you are interested in the algorithms of the Hellinger distance, see util.py.