Monitoring water migration in crispy snacks using NIR

In this project a prelimenary image preprocessing pipeline was developed to analyze Near-Infrared Spectroscopy Imaging data of components of crispy snacks, when imaged under different relative humidities. The preprocessing pipeline (in preprocessCrispySnacks.py) performs the following steps:
1. Image data reading for all sets of measurements (directories)
2. Image registration
3. Data reduction on the individual sets of measuerments individually by PCA
4. Data reduction on the concatenated set of all sets of measurements by PCA
5. Foreground/background segmentation by Gaussian Mixture modelling
6. Identification of inidividual containers from intersecting the foreground cluster
with manually segmented containers.
7. Reorganizing the pixel-level data by container, by labelling pixel-level data
with x,y coordinates and container label. Only keeping labelled pixels
8. Saving pixel-level data as csv-file
Which are further detailed in the comments within the Python script.

This repository also contains several directories with imaging data from different relative humidities (as indicated by the directory names *awXX*.
The data were part of a larger project and were reused for this project. The only novel element is the preprocessing script in this repository. The data are added for users to be able to test the preprocessing pipeline.