Copy-Move Forgery Localization Using Convolutional Neural Networks and CFA Features

Copy-Move Forgery Localization Using Convolutional Neural Networks and CFA Features

Lu Liu (Institute of Information Science, Beijing Jiaotong University, Beijing, China), Yao Zhao (Institute of Information Science, Beijing Jiaotong University, Beijing, China), Rongrong Ni (Institute of Information Science, Beijing Jiaotong University, Beijing, China) and Qi Tian (Department of Computer Science, University of Texas at San Antonio, San Antonio, US)
Copyright: © 2018 |Pages: 16
DOI: 10.4018/IJDCF.2018100110

Abstract

This article describes how images could be forged using different techniques, and the most common forgery is copy-move forgery, in which a part of an image is duplicated and placed elsewhere in the same image. This article describes a convolutional neural network (CNN)-based method to accurately localize the tampered regions, which combines color filter array (CFA) features. The CFA interpolation algorithm introduces the correlation and consistency among the pixels, which can be easily destroyed by most image processing operations. The proposed CNN method can effectively distinguish the traces caused by copy-move forgeries and some post-processing operations. Additionally, it can utilize the classification result to guide the feature extraction, which can enhance the robustness of the learned features. This article, per the authors, tests the proposed method in several experiments. The results demonstrate the efficiency of the method on different forgeries and quantifies its robustness and sensitivity.
Article Preview

1. Introduction

As society develops, digital image plays a more and more important role in our daily life. Recently, digital image serves as carrier of information and medium of communication. A digital image not only can be used as evidences in court, but also can be used by doctors to diagnose the patients. In addition, digital images frequently occur in newspaper, magazine and television, which enable people to receive and understand information vividly.

With the development of image processing techniques, there are many tools, which can decorate an image easily without leaving obvious traces to the human eyes. Nowadays, people usually upload images that have been beautified to the Internet. It has brought a lot of benefits but has also brought some problems. Because not every image modification is performed for beauty, vicious tampering has forged realistic fake images. Once these manipulated images are maliciously used to mislead the public about the truth, it will be no doubt to seriously threaten stability and development of the society.

Copy-move forgery is one of the most common forgeries, where some parts of an image are copied and pasted to other parts of the same image. In order to harmonize the tampered regions with its surroundings, some other image processing operations, such as rotation, scaling, JPEG compression and blurring could be involved together with copy-move manipulation. Due to its effectiveness and simplicity, even a non-professional can deliberately tamper a digital image to obtain personal benefit or achieve personal intent. An example of copy-move forgery is shown in Figure 1, where a region was copied and pasted from the top right corner to the top left corner. This photo is tampered for beauty and variety. Even so, this is still not allowed in the photography competition. Actually, most images were tampered and were intended to hide an object in the image, or to emphasize a particular object. In consequence, it is urgent and important to research some methods to deal with the problem of authenticity and origin of digital images.

Figure 1.

An example of copy-move forgery

To authenticate a digital image, many techniques have been developed. In general, these techniques can be divided into two types, active techniques and passive techniques. Active techniques usually use watermarks (Kundur & Hatzinakos, 1999) or digital signatures (Lu & Liao, 2003) as the authentication information to distinguish the truth from the false images. In these methods, particular data needs to be embedded before publication of the digital images. It is unrealistic in practical applications. In fact, not all images are embedded with watermarks or digital signatures. Based on this, it is more likely that suspicious images are not embedded with watermarks. In contrast, passive techniques determine the authenticity and integrality of the digital images only by detecting the manipulation traces left in images, without the support of any additional pre-processing operation. Therefore, more attention has been devoted to passive techniques recently.

Until now, a number of passive techniques have been developed to detect the copy-move forgery. The proposed methods can be divided into two categories: key point-based methods or block-based methods. In the literature (Christlein, Riess, & Jordan, 2003), the author proposed a common processing pipeline about copy-move forgery detection.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 11: 4 Issues (2019): 2 Released, 2 Forthcoming
Volume 10: 4 Issues (2018)
Volume 9: 4 Issues (2017)
Volume 8: 4 Issues (2016)
Volume 7: 4 Issues (2015)
Volume 6: 4 Issues (2014)
Volume 5: 4 Issues (2013)
Volume 4: 4 Issues (2012)
Volume 3: 4 Issues (2011)
Volume 2: 4 Issues (2010)
Volume 1: 4 Issues (2009)
View Complete Journal Contents Listing