TISA - Thin-section Image Segmenter & Analyzer

🚀 TISA (Thin-section Image Segmenter & Analyzer) is an advanced tool for automated segmentation and analysis of petrographic thin sections. It enables efficient identification and classification of mineral grains using state-of-the-art segmentation algorithms.

---

📸 Key Features

• 🖥️ Modern GUI: Built with PyQt5 for an intuitive user experience.
• 🧩 Automated Segmentation: Supports multiple segmentation techniques (K-means, Deep Learning).
• 🎨 Fixed Color Palette Mapping: Ensures consistent label visualization across images.
• 🛠️ Automatic Artifact Removal: Detects and removes thin-section edges and unwanted lighting effects.
• 📊 Data Export: Saves segmented images and statistical analyses in .csv format.

---

🏗️ Installation

1️⃣ Prerequisites

Ensure you have Python 3.8+ installed. Then, install dependencies:

pip install -r requirements.txt

2️⃣ Running the Application

To launch TISA’s graphical interface:

python -m tisa.main

---

📖 Documentation

For detailed usage instructions and advanced settings, check out the full documentation:

📚 TISA Documentation

---

⚙️ Segmentation Methods

TISA provides multiple segmentation algorithms:

• K-Means Clustering: Groups pixels based on color similarity.
• Deep Learning-Based Segmentation: Uses CNNs for precise mineral identification.
• Felzenszwalb Algorithm: Graph-based segmentation suited for texture variations.

Each method is adjustable through parameters such as:

Parameter	Description
mod_dim1	Number of filters in the first convolutional layer.
mod_dim2	Number of filters in the second convolutional layer.
train_epoch	Number of training iterations.
max_label_num	Maximum number of segmentation classes.

---

📂 Project Structure

TISA/
│── tisa/                       # Main source code
│   ├── core/
│       ├── analysis.py         # Calculate label percentages
│       ├── mosaic.py           # Build mosaic and split images
│       ├── segmentation/
│           ├── automated.py    # Automated segmentation
│           ├── models_tf.py    # From PetroSeg
│   ├── icons/                  # Icons
│   ├── utils/                  # Utilities
│   ├── gui.py                  # Graphical User Interface (PyQt)
│   ├── main.py                 # Application entry point
│── docs/                       # Documentation
│   ├── index.md                # Main documentation page
│   ├── guide.md                # User guide
│   ├── segmentation.md         # Segmentation explanation
│── requirements.txt            # Dependency list
│── README.md                   # Quick project overview

---

🏛️ Based on PetroSeg by Azzam et al.

TISA builds upon PetroSeg, originally developed by Azzam et al.

Reference: 📄 Azzam, F., Blaise, T., & Brigaud, B. (2024). Automated petrographic image analysis by super- vised and unsupervised machine learning methods. Sedimentologika, 2(2). PetroSeg GitHub Repo

---

📬 Contributing

If you’d like to contribute to TISA, feel free to:

• Report issues or suggest improvements via GitHub Issues.
• Contribute code by forking the repository and submitting Pull Requests.
• Share petrographic datasets to enhance segmentation capabilities.

---

📄 Citation

If you use this project, please cite as follows:

SOARES CORREIA, M. (2025). TISA v1.0.0 - Thin-section Image Segmenter and Analyzer. Université Paris-Saclay. https://doi.org/10.5281/zenodo.14996768

---

📄 License

TISA is distributed under the GNU GPL v3 License, allowing free use, modification, and redistribution.

---

📢 Final Words

TISA is designed to make thin-section analysis faster and more efficient. If you have any questions or suggestions, feel free to reach out! 🚀