PoroREV

This repository estimates porosity REV (Representative Elementary Volume) from a segmented 3D image using random cube sampling.

The script supports:

• .raw volumes (with dimensions from input.txt)
• .tif / .tiff 3D stacks (dimensions read from file)

---

Installation

1. Python 3.9+ recommended.
2. Install dependencies:

• pip:

pip install numpy matplotlib seaborn tifffile imageio

• conda:

conda create -n pororev python=3.10 -y
conda activate pororev
conda install -c conda-forge numpy matplotlib seaborn tifffile imageio -y

---

Inputs

• PoroREV.py: main script
• input.txt: user-editable configuration file

---

How to Run

• Default config file:

python3 PoroREV.py

• Custom config file path:

python3 PoroREV.py --config input.txt

Configuration (input.txt)

1) Phase labels

• pore: pore label value in the segmented image
• solid: solid label value in the segmented image

Important behavior:

• The current code maps voxels equal to solid to 0 (solid).
• All other voxel values are mapped to 1 (pore).
• pore and solid must be different values.

2) Input file

• filename: input image path (.raw, .tif, .tiff)
  • Relative paths are resolved relative to the config file location.

3) RAW file settings (used only for .raw)

• filesize_x, filesize_y, filesize_z: raw dimensions in voxels
• dtype: raw voxel type (examples: uint8, uint16, int16, float32)

Note:

• For .tif/.tiff, raw size fields and dtype are ignored.

4) REV sampling settings

• min_side: smallest sampled cube side length (voxel)
• step: cube side increment between samples
• samples: random cubes per side length
• seed: random seed for reproducibility

5) Output settings

• save_fig: output figure path (.png, .pdf, etc.)
• save_csv: output table path

6) Optional REV detection by CI

• rev_ci_threshold (optional): CI half-width threshold
• rev_consecutive: required consecutive side lengths below threshold

If rev_ci_threshold is commented out or missing, automatic REV-size detection is skipped.

Outputs

1. Figure (save_fig):

• Mean porosity vs. cube size
• Shaded 95% confidence interval

2. CSV (save_csv) with columns:

• cube_size_voxel
• mean_porosity
• std_porosity
• ci95_halfwidth
• cv