🔐 Crypto-Based Image Steganography Tool

A secure and flexible Python-based steganography system that embeds encrypted data into images using cryptographic randomness and LSB encoding.

🌟 Project Overview

This project presents a cryptographically-enhanced image steganography system that securely hides a message image inside a cover image, with minimal visual distortion. Unlike traditional LSB (Least Significant Bit) methods, this tool leverages symmetric encryption and pseudo-randomized bit placement, ensuring that only users with the correct key can retrieve the hidden message.

---

✨ Features

• 🔒 Cryptographic Embedding
  Uses a user-provided key to:

  • Encrypt the message using Fernet symmetric encryption.
  • Seed a pseudo-random generator to securely select pixel locations for embedding.

• 🎚️ Bit-Depth Flexibility
  Supports 1 to 8 bits per pixel channel for embedding — choose between stealth or higher capacity.

• 🖼️ Multi-Format Image Support
  Supports common image formats like JPEG, PNG, and BMP via OpenCV, assuming the files are readable.

• 🧪 Robust Against Attacks
  Without the correct key:

  • Pixel positions remain unpredictable.
  • Decrypted content is inaccessible.

• 💻 User-Friendly CLI
  Simple prompts to encode and decode messages, with minimal user configuration.

---

🔧 Methodology

1. Encoding Process

• Load cover and message images.
• Encrypt message bytes using Fernet.
• Convert encrypted bytes to binary.
• Pseudo-randomly select pixel positions using a key-seeded NumPy RNG.
• Embed bits using LSB across chosen indices.

2. Decoding Process

• Load stego image.
• Pseudo-randomly retrieve embedding indices using the same key.
• Extract bits and decrypt them to reconstruct the original message image.

---

🛡️ Security Architecture

• Encryption: Fernet ensures both confidentiality and integrity.
• Key-based Embedding: The embedding process is tied to the key, making reverse-engineering infeasible.
• Layered Defense:
  • Obscured pixel positions
  • Encrypted message content
  • No key = no message

---

🖥️ Usage

🔐 Encode

$ python steg_tool.py
Give me the cover image path: cover.jpg
Give me the message image path: secret.png
Do you want to start encoding the message? (y/n): y

🔓 Decode

Do you want to decode the message? (y/n): y

The stego image will be saved as stego.png, and the recovered message as decoded_message.png.

---

🧪 Experimental Results

• Visual Quality: Stego images were nearly indistinguishable from original cover images even at 4-bit embedding.
• Recovery: Decoded messages matched the original with high fidelity.
• Security: Unauthorized decoding attempts without the key fail completely, validating cryptographic security.

---

📁 Sample Output

Original Cover	Stego Image

Original Message	Decoded Message

---

🧠 Tech Stack

• Python 3
• OpenCV (cv2)
• NumPy
• Cryptography (Fernet)

---

👨‍💻 Developers

• Satwik Raj Wadhwa (230937)
• Gaurav Kumar (230792)
• Vinay Chavan (231155)
• Pritam Priyadarshi (230793)
  EE655, IIT Kanpur — Guided by Prof. Koteswar Jerripothula

---

📜 License

This project is part of a university course and is intended for academic use only. Contact the developers for further usage permissions.