Source Camera Identification App

This project is a desktop application that identifies the device used to capture a digital image by analyzing image characteristics using multiple methods.

Overview

The application performs three independent analyses to estimate the most probable source device of a photo:

1. JPEG Compression analysis – examines quantization matrices and DCT artifacts.
2. PRNU (Photo response non-uniformity) – extracts and matches sensor noise patterns unique to each device.
3. CNN-Based noise residual analysis – uses a trained convolutional neural network to classify noise features.

The results from these three methods are fused to produce a combined prediction of the device using which the picture was taken.

Features

• Train models from image folders organized by device type.
• Perform device detection on images with metadata removed.
• View detailed results for each method and combined output.
• Multithreaded training and analysis with progress and log display.

Project Structure

picture_device/
├── cnn_model.py
├── database.py
├── fusion_analysis.py
├── gui.py
├── jpeg_analyzer.py
├── main.py
├── prnu_analyzer.py
├── utils.py
├── requirements.txt
└── models/ (generated after the training)

Usage

1. Install the dependencies:

   pip install -r requirements.txt
   

2. Run the application:

   python main.py
   

3. Training

   • Select a folder with images.
   • Press "Start training" button to generate reference models for each analysis method.

4. Detection
   • Go to the "Device detection" tab.
   • Select an image for testing.
   • Press "Analyze image" button to see the results.

Gallery

Additional Training

"About" Tab

Authors

Mansur Ozaman: https://github.com/ozxmn