🧠 Local 2D Otsu Thresholding

Local 2D Otsu Thresholding is a Numba-accelerated implementation of a local 2D Otsu thresholding method for 3D images.

This algorithm adapts the 2D Otsu threshold within a sliding 3D window, enabling robust segmentation in datasets with spatially varying intensity distributions.

⚙️ Installation

Install the latest stable version from PyPI:

pip install otsu2D

🚀 Example Usage

import numpy as np
from otsu2D import getBinary

# Create sample 3D image
img = np.random.randint(0, 256, size=(10, 10, 10), dtype=np.uint8)

# Get threshold map and binary image
binary = getBinary(img, window_size=(3, 3, 3),  mean_window_size =(3,3,3))

🖥️ Command Line Usage

Run local 2D Otsu thresholding on a single 3D image stack:

otsu-2d \
  --image path/to/input_stack.tif \
  --window 3 3 3 \
  --mean_window 3 3 3 \
  --outdir outputs \
  --format tif

The output is saved as a binary 3D stack. When using --format tif, the result is saved as a single multi-page TIFF file.

🗂️ Batch Processing

To process all .tif and .tiff files inside a directory and its subdirectories:

otsu-2d-batch \
  --input_dir path/to/input_directory \
  --outdir path/to/output_directory \
  --window 3 3 3 \
  --mean_window 3 3 3 \
  --format tif

The subdirectory structure is preserved in the output directory.

📦 Dependencies

• NumPy
• Numba
• tifffile
• Matplotlib, optional, only for visualization with --show

Install the required dependencies with:

pip install otsu2D

To include optional plotting support:

pip install "otsu2D[plot]"

📜 License

This project is licensed under the MIT License. See the LICENSE file for details.

🤝 Contributing

Contributions are welcome! If you'd like to fix a bug, add a feature, or improve performance, please open a pull request or contact the maintainers.

💬 Contact

For questions, issues, or feedback, open an issue on GitHub.