Gaussian Mixture Models for Robust Unsupervised Scanning-Electron Microscopy Image Segmentation of North Sea Chalk

Here, we reproduce the research presented in:

Dramsch, J. S., Amour, F., & Lüthje, M. (2018, November). Gaussian Mixture Models for Robust Unsupervised Scanning-Electron Microscopy Image Segmentation of North Sea Chalk. In First EAGE/PESGB Workshop Machine Learning.

We use Gaussian Mixture Models to find the optimal separation of chalk (oolites) in backscatter scanning-electron microscopy images.

Then we apply morphological filtering to smoothe out the edges in the notebook SEM Segmentation.ipynb

We then perform granulometry on the perimeters of each grain in the notebook SEM Granulometry.ipynb