Skip to main content

Investigating Manipulations with Advanced EditTrace and Cloning Detection

Learn how to use Advanced EditTrace and Cloning Detection to identify subtle manipulations, including content-aware edits and cloning.

D
Written by Dekel Faruhi

Introduction

Welcome to Proofig AI. In this tutorial, we’ll explore Advanced EditTrace and Cloning Detection—a powerful tool designed to detect subtle content-aware edits and cloning, helping researchers maintain the integrity of their work.
​

This video is an alternative presentation of the complete instructions below.

How It Works

Proofig AI uses AI-driven similarity lines to flag areas of interest within an image. These lines highlight regions with matching patterns, textures, or features:

  • Clusters of Two or More Lines: These clusters suggest stronger evidence of manipulation and are the primary focus for investigation.

  • Single-Line Flags: These flags may indicate smaller or subtler manipulations. While false positives can occur in regions with high natural similarity, the system’s high sensitivity ensures that even minor changes are flagged for review.

Investigating Dense Clusters

In this microscopy example, Proofig AI detected three cloning instances. Each instance is flagged by dense clusters of similarity lines, indicating areas with high matching patterns.
​

Alteration AI confidence score of 89%; the ends of each line cluster are boxed in a matching color.

To refine your analysis, you can toggle similarity lines on or off by pressing 'Hide Lines' or 'Show Lines'.
​


This allows you to view the image with or without overlays, helping you focus on specific flagged regions.

For further confirmation, apply color maps like Jet or HSV. These tools enhance textures and patterns, making it easier to validate whether flagged regions are indeed identical.

The same image under a color map with similarity lines hidden; the flagged regions remain boxed.

Analyzing Sparse Similarity Lines

In this second example, Proofig AI flagged three cloning instances with varying levels of similarity:

  • Two Clusters of Three Lines Each: These represent clear and deliberate manipulations.
    ​

    Alteration AI confidence score of 89%; each cluster's two ends are boxed in a matching color.

  • One Single-Line Flag: This represents a smaller manipulation. Even with fewer lines, single-line flags can still reveal genuine edits, especially when the manipulation is subtle.
    ​


​
Using color maps can confirm these flagged areas by revealing matching textures and patterns, ensuring no potential issue is overlooked.

Under a color map, each flagged pair is boxed in a matching color: orange, green or blue.

Conclusion

Advanced EditTrace and Cloning Detection equips researchers with powerful tools to uncover even the most seamless content-aware edits and cloning manipulations. By combining similarity lines with forensic tools like color maps, you can confidently validate the integrity of your scientific images.

Thank you for using Proofig AI.

Related Articles
Did this answer your question?