# Forensic Scanner Identification Using Machine Learning

**Authors:** Ruiting Shao, Edward J. Delp  
**arXiv:** [2002.02079](https://arxiv.org/abs/2002.02079)  
**Date:** 2020-02-06  
**Summary page:** https://www.summarizepaper.com/en/paper/2002.02079v1/

## Key points

- Development of robust techniques for digital image authentication, source identification, and tamper detection is necessary due to advanced image editing tools
- Authors Ruiting Shao and Edward J. Delp introduce a machine learning-based system for analyzing scanner devices in forensics
- The system utilizes deep-learning algorithms to autonomously extract intrinsic features from scanned images
- Through rigorous experimentation, the system achieves high accuracy in identifying the source scanner of an image
- Equipped with a reliability map that identifies manipulated regions within a scanned image
- This innovative approach enhances forensic investigations and highlights the crucial role of machine learning in advancing digital forensics methodologies

## Summary

In the realm of forensic image analysis, the development of robust techniques for digital image authentication, source identification, and tamper detection has become necessary due to the advent of advanced image editing tools. In their paper titled "Forensic Scanner Identification Using Machine Learning," authors Ruiting Shao and Edward J. Delp introduce a machine learning-based system specifically designed for analyzing scanner devices in forensics. By utilizing deep-learning algorithms, this system can autonomously extract intrinsic features from a wide range of scanned images. Through rigorous experimentation, the authors demonstrate that the system achieves high accuracy in identifying the source scanner of an image. Additionally, it is equipped with a reliability map that identifies manipulated regions within a scanned image. This innovative approach not only enhances forensic investigations but also highlights the crucial role of machine learning in advancing digital forensics methodologies.