DeepData

CLI Tool for hiding files in images using Steganography

deepdata is a command-line utility written in C# that allows hiding any file inside an image (JPG, PNG, BMP) and extracting it later.
It supports multiple steganography methods including LSB, QIM, and JPEG DCT coefficient manipulation.

---

Features

• Hides any files inside PNG, BMP, or JPEG/JPG images
• Extracts hidden files from stego-images
• Supports multiple steganography methods:
  • LSB (Least Significant Bit) for lossless formats
  • QIM (Quantization Index Modulation) for enhanced capacity
  • DCT (Discrete Cosine Transform) for JPEG images
• Format-aware auto method selection
• Can be set up easily (channels strength etc.)

---

Installation

Manual

Clone and build using .NET:

  git clone https://github.com/dehobitto/DeepData.git
  cd DeepData/DeepData.CLI
  dotnet build

Recommended

Download prebuilt executable file from repository releases and add to PATH variable

---

Usage

If manually installed should use dotnet run, if the way I recommend - deepdata. I'll use the second one further:

Using auto-detected method

  deepdata embed photo.jpg secret.txt --output stego.jpg

  deepdata extract stego.jpg --output recovered.<extension>

  deepdata capacity image.jpg

Advanced Usage

  deepdata embed photo.jpg secret.txt --method qim --settings "channels=Y,Cb delta=4"

  deepdata extract stego.jpg --output recovered.<extension> --method lsb --settings "channels=R,G strength=2"

  deepdata capacity diagram.bmp --method lsb --settings "channels=R,G strength=2"

Help and Version

    deepdata --help
    deepdata --version

---

Overview

Libraries Used

• ImageSharp

  • Cross-platform image manipulation library
  • Used for lossless format handling (PNG, BMP)
  • Provides efficient pixel manipulation

• BitMiracle.LibJpeg.Classic

  • Direct access to JPEG DCT coefficients
  • Enables precise coefficient manipulation
  • Maintains image quality during embedding

Main Components

DeepData.CLI

• Program.cs: Entry point and command-line interface
• Commands/: Command implementations
  • command inherits from BaseCommand.cs implements _ICommand.cs
• Utils/: Helper utilities and common functions

DeepData.Steganograph

• Methods/: Steganography implementations
  • method inherits from StegoMethod implements IStegoMethod
    • Lsb.cs
    • Qim.cs
    • Dct.cs
• Utils/: Shared utilities
• Settings/: Configuration handling

Application Structure

DeepData/
├── DeepData.CLI/               # Command-line interface
│   ├── Commands/               # Command implementations
│   ├── Utils/                  # CLI utilities
│   └── Program.cs              # Entry point
├── DeepData.Steganograph/      # Core steganography logic
│   ├── Methods/                # Steganography methods
│   ├── Interfaces/             # Core abstractions
│   ├── Models/                 # Data structures
│   ├── Utils/                  # Shared utilities
│   └── Extensions/             # Extension methods

Component Interaction

1. CLI Layer (DeepData.CLI)

   • Parses user commands and arguments
   • Validates input parameters
   • Handles image format detection
   • Routes to appropriate steganography methods

2. Core Logic Layer (DeepData.Steganograph)

   • Implements steganography algorithms
   • Manages data embedding/extraction
   • Provides progress reporting

3. Method Layer (Methods/)

   • LSB: Bit-level manipulation for lossless formats
   • QIM: Enhanced capacity through quantization
   • DCT: JPEG-specific coefficient manipulation

---

Program Flow

Main Flow

Description

Start:

• User → CLI: Command (embed/extract)
• CLI: Parse arguments, validate input files, read source image, determine method
• CLI → Method: Create method instance
• Method: Check capacity, prepare data
• If Embed:
  • Read data file, embed data, generate stego-image
• Else Extract:
  • Extract data, validate extracted data, generate output
• Method → CLI: Return result
• CLI: Save output file, show progress
• CLI → User: Return success

---

Lossless Methods (LSB & QIM)

Description

Start:

• Data → Method: Input data bits
• Image → Method: Input image pixels
• If LSB:
  • Replace N least significant bits with data bits
• Else QIM:
  • Quantize value to nearest representative
• Method → Image: Modified image
• Image → Method: Input stego-image
• If LSB:
  • Extract N least significant bits
• Else QIM:
  • Determine which representative is closest
• Method → Data: Reconstructed data bits

---

Lossy Method

Description

Start:

• Data → DCT: Input data bits
• Image → DCT: Input JPEG image
• DCT: Decompress to DCT coefficients
• DCT: Process 8×8 blocks
• If Embedding:
  • Check capacity
  • Modify DCT with QIM
  • Skip zero coefficients
  • → Image: Compress to JPEG
• Else Extraction:
  • Read header
  • Extract from DCT
  • Skip zero coefficients
  • → Data: Save as file

Examples

Lsb showcase

We will see clearly how the lsb parameters affects the image.

Original	Strength = 2	Strength = 6	Strength = 6 R,G

Dct showcase

We will see clearly how the dct parameters affect the image.

Original	Delta = 10

Delta = 20	Delta = 40 Cb,Cr