International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 11 Issue: 05 | May 2024
www.irjet.net
p-ISSN: 2395-0072
Comprehensive Analysis of Deep Learning Approach for detecting Forgery in Digital Images Clifa Mascarenhas1, Nadine Dias2 1Student, Department of Information Technology and Engineering, Goa College of Engineering, Farmagudi, Goa,
India
2Assistant Professor, Department of Information Technology and Engineering, Goa College of Engineering,
Farmagudi, Goa, India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Image forgery is a growing concern in today's
manipulated images, strategic data preprocessing, transfer learning from pre-trained models, and rigorous training. The paper emphasizes the importance of maintaining image authenticity and proposes future directions for improving detection and localization of digital forgeries.
digital age, where images can be easily manipulated and altered using various software tools. The proposed project aims to tackle the pressing issue of image forgery in the digital era by utilizing advanced deep learning techniques for detection. By curating a dataset containing both authentic and manipulated images, the project seeks to expose the model to various forgery scenarios. Through the use of lightweight deep learning architectures, strategic data preprocessing, transfer learning from pre-trained models, and rigorous training, the project endeavours to develop a reliable system for automated forgery detection. This comprehensive approach is designed to enhance image forensics and safeguard the credibility and integrity of visual content in the face of increasing threats posed by image manipulation.
The rest of the paper is organized as follows: Section 2 describes the Related work done in this domain. The method proposed which includes the use of transfer learning from pre-trained models and Procedure is given in Section 3. Conclusions are discussed in Section 4.
2. RELATED WORK In the pursuit of tackling image forgery, researchers have explored various methodologies, each contributing to the evolving landscape of digital forensics. Traditional approaches often relied on handcrafted features and rulebased algorithms, but recent advancements have witnessed a paradigm shift towards data-driven techniques, particularly deep learning. Here, we delve into a brief review of related work, highlighting key developments in image forgery detection:
Key Words: image forgery, lightweight deep learning architectures, transfer learning, image forensics, image manipulation
1.INTRODUCTION Image forgery, the manipulation of digital images, is a growing concern in today's digital age. With the increasing use of digital images in various fields such as forensic investigation, criminal investigation, intelligence systems, medical imaging, insurance claims, and journalism, the credibility and integrity of visual content are at stake. The ease of image manipulation using advanced software tools has made it challenging to distinguish between authentic and manipulated images.
2.1 Traditional Methods Traditional image forgery detection methods often focused on extracting handcrafted features such as texture, color, and shape. Techniques like error level analysis (ELA), wavelet transforms, and statistical measures were commonly employed. While these methods showcased effectiveness, they struggled to adapt to the complexities of modern forgery techniques.
To address this challenge, researchers have turned to deep learning techniques to develop automated image forgery detection systems. Deep learning models, such as convolutional neural networks (CNNs), have shown promising results in detecting image forgery with high accuracy. These models can learn complex features from large datasets of authentic and manipulated images, enabling them to distinguish between genuine and manipulated images.
Traditional image forgery detection methods have historically relied on extracting handcrafted features from images, including texture, color, and shape, to identify potential anomalies indicative of tampering. These methods aimed to leverage intrinsic characteristics of digital images to distinguish between authentic and manipulated content. Techniques such as error level analysis (ELA), wavelet transforms, and statistical measures were commonly employed in this regard.
This paper presents a novel approach to image forgery detection using deep learning techniques. The proposed approach involves a curated dataset of authentic and
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 619