Series of code files related to surface roughness chracterisation using surface generation on ImageJ, CT scans and machine learning. The overall workflow of the project:
- Dozens of rough surface files (both .stl and .tif) are generated using the ImageJ macro 'Generate_Surface_1.3 (Cleaning Option).ijm' with a binary-style surface.
- These are scanned using a digital twin of the Nikon XTH 225 CT scanner with code from gvxr on Python 'gvxr_basics.py'.
- These scans are used within U-Net convolution neural networks to train a machine to improve the resolution of these scanned images, improving the quality of the output from the scanner and learning more about the object being scanned.