Image Forgery Detection Techniques: Latest Trends and Key Challenges

Document Type

Article

Publication Title

IEEE Access

Abstract

The improvement and accessibility of high-resolution cameras have significantly increased image capturing by various media. Different editing tools are available that are frequently used to improve image quality, resulting in the alteration of images. So, determining the authenticity or integrity of the original image is a challenging task in any domain. Currently, an image or video is a critical source of legal data for digital forensics. Hence, the study begins with the primary objective of determining whether the digital evidence (image or video) associated with a legal case has been altered. Active forgery detection methods, such as digital watermarking and digital signatures, and passive forgery detection techniques, including copy-move, splicing, and retouching, are used to verify digital evidence. Recently, neural network (NN) based forgery detection has garnered notable attention due to its efficacy in detecting forgery on images. In order to assess the benefits and drawbacks of those models, this work starts by looking at different categories of forgery detection and how those classifications are implemented. We summarised and compared the different architectures that researchers had suggested, taking into account their respective specialised viewpoints, and then we analysed them according to quality. Most used methodologies in forgery detection are elaborated by mentioning the advantages and disadvantages of each technique. Additionally, the work examines the deployment of neural networks and machine learning (ML) techniques within forensic science to detect image forgery. We also underlined the present challenges and potential research directions that might help researchers fill the knowledge gaps.

First Page

169452

Last Page

169466

DOI

10.1109/ACCESS.2024.3498340

Publication Date

1-1-2024

This document is currently not available here.

Share

COinS