🕵️ KANTEI

A Python-based heuristic tool for detecting potential steganography in images (PNG / JPEG).

---

🐍 Python Compatibility

This project requires:

Python 3.9 or newer
(Recommended: Python 3.9 -- 3.13)

Python 3.14+ may work but is very recent and some third-party libraries may take time to fully support it.

Check your Python version

Windows:

py --version

or

python --version

Linux / macOS:

python3 --version

---

📦 Installation

1️⃣ Clone the repository

git clone https://github.com/zer0trace1/KANTEI.git
cd KANTEI

---

2️⃣ Create a virtual environment

Windows (PowerShell)

py -m venv [name]
.\[name]\Scripts\activate

Linux / macOS

python3 -m venv [name]
source [name]/bin/activate

---

3️⃣ Install dependencies

pip install -r requirements.txt

If you do not have requirements.txt, install manually:

pip install pillow numpy matplotlib

Optional (for future ML module):

pip install scikit-learn pandas

---

🚀 Running the Program

Depending on your OS and Python installation:

Windows

py stego-scanner.py image.png --plot

or

python stego-scanner.py image.png --plot

Linux / macOS

python3 stego-scanner.py image.png --plot

If multiple Python versions are installed, you can specify one explicitly:

py -3.12 stego-scanner.py image.png

---

🔍 Analyze an Image

python stego-scanner.py image.png --plot --report report.json

---

Available Options

Option	Description
--visualize	Generates bitplane images (PNG/BMP)
--plot	Generates metric graph (*_metrics.png)
--report file.json	Saves full report as JSON
--csv metrics.csv	Appends metrics to a cumulative CSV

---

🧪 Generate Test Images (LSB Injection)

python inject_lsb.py

By default, it generates:

• img1_infected_30.png
• img1_infected_100.png

---

📊 Generated Outputs

• *_metrics.png → Metric visualization graph
• *_ELA.png → JPEG ELA image
• *_bit0.png ... *_bit7.png → Bitplanes
• metrics.csv → Accumulated dataset of runs
• report.json → Full analysis report

---

🧠 Technical Overview

The system combines classical steganalysis techniques:

• LSB Chi-Square (statistical deviation detection)
• Simplified RS Analysis
• ELA (JPEG recompression difference)
• Structural file analysis (trailing data detection)

The tool produces a heuristic score (0--1) and a verdict:

• LOW
• MEDIUM
• SUSPICIOUS

---

⚠️ Limitations

• Heuristic-based detection (not mathematically definitive).
• Does not automatically extract hidden payloads.
• Advanced DCT-based detection not yet implemented.

---

🔮 Roadmap

• 🤖 Machine Learning classifier based on extracted features
• 📊 Large-scale dataset evaluation
• 🧠 Full RS and SPA implementation
• 🔍 Advanced DCT-based JPEG analysis
• 📦 Packaging as an installable CLI tool

---

⚖️ Legal Notice

This project is intended for educational and cybersecurity research purposes only.
Do not use it for illegal activities.