Skip to main content
main-content
Top

About this book

This book discusses blind investigation and recovery of digital evidence left behind on digital devices, primarily for the purpose of tracing cybercrime sources and criminals. It presents an overview of the challenges of digital image forensics, with a specific focus on two of the most common forensic problems. The first part of the book addresses image source investigation, which involves mapping an image back to its camera source to facilitate investigating and tracing the source of a crime. The second part of the book focuses on image-forgery detection, primarily focusing on “copy-move forgery” in digital images, and presenting effective solutions to copy-move forgery detection with an emphasis on additional related challenges such as blur-invariance, similar genuine object identification, etc. The book concludes with future research directions, including counter forensics. With the necessary mathematical information in every chapter, the book serves as a useful reference resource for researchers and professionals alike. In addition, it can also be used as a supplementary text for upper-undergraduate and graduate-level courses on “Digital Image Processing”, “Information Security”, “Machine Learning”, “Computer Vision” and “Multimedia Security and Forensics”.

Table of Contents

Frontmatter

Chapter 1. Introduction

Abstract
In this chapter, an introduction to the essence of digital image forensics has been given, including its differences with more widely known multimedia security techniques such as watermarking and steganography. Using illustrative examples, several image forensics techniques have been discussed. This chapter aims to make the reader interested in this budding research domain.
Aniket Roy, Rahul Dixit, Ruchira Naskar, Rajat Subhra Chakraborty

Chapter 2. Camera Source Identification

Abstract
Camera source identification is one of the fundamental problems in digital image forensics. In this chapter, at the outset, the reader is made familiar with the basic components of a modern digital camera, along with the processing, acquisition, and storage of digital images in camera sources. Then a detailed literature review has been done for efficient, blind camera source identification. One of the major and effective solutions to the problem is to extract appropriate features from the images, then train a classifier, and finally classify the test samples using that trained classifier. In this chapter, we have discussed a camera source identification methodology, based on extraction of the discrete cosine transform residual features, and subsequent random forest-based ensemble classification with AdaBoost. The classification accuracy was further improved by incorporating dimensionality reduction by principal component analysis. The experiments were performed on the Dresden Image Database, and the state-of-the-art techniques were compared in detail. Moreover, the proposed technique shows low overfitting trends when the constructed classifier for one image database is applied to a different image database.
Aniket Roy, Rahul Dixit, Ruchira Naskar, Rajat Subhra Chakraborty

Chapter 3. Copy-Move Forgery Detection in Digital Images—Survey and Accuracy Estimation Metrics

Abstract
Copy-move forgery is a common type of modification attack on images. In such type of manipulation, part of an image is copied and pasted on any other location of the same image. Hence, this form of manipulation is difficult to be identified by methods which look for statistical inconsistencies between different regions of an image. Existing methods are primarily motivated toward finding copy-moved area(s) in an image. In this chapter, we provide a detailed survey of existing techniques of copy-move forgery detection and demonstrate the application of three metrics for forgery detection, viz., detection accuracy, false positive rate, and false negative rate. On the basis of proposed parameters, we compare and analyze the success of different block-based copy-move forgery detection methods. These metrics help to choose the algorithm most suitable for the context.
Aniket Roy, Rahul Dixit, Ruchira Naskar, Rajat Subhra Chakraborty

Chapter 4. Copy-Move Forgery Detection Exploiting Statistical Image Features

Abstract
Majority of image manipulation identification methods attempt to find variation or dissimilarities in statistical image features. But such schemes do not identify in case of copy-move forgery type of image manipulation, because the forged region comes from the same image. In this chapter, we present a statistical image features-based copy-move forgery detection method. The mean value gives the pixel intensity of the entire image, and variance is used to find how each pixel intensity differs from its neighbors within a block. We evaluate and compare the presented algorithm with other existing methods through a comprehensive set of experiments. Our experimental results show that this technique achieves considerably higher detection accuracy as compared to the existing methods.
Aniket Roy, Rahul Dixit, Ruchira Naskar, Rajat Subhra Chakraborty

Chapter 5. Copy-Move Forgery Detection with Similar But Genuine Objects

Abstract
Copy-Move Forgery Detection (CMFD) is a classic image forensics problem. However, CMFD with Similar but Genuine Objects (SGO) has received relatively less attention in the current literature. Recently, it has been observed that state-of-the-art CMFD techniques are not quite effective in solving this important yet less explored problem variant. We have addressed this specific problem by using Rotated Local Binary Pattern (RLBP) texture features, followed by feature matching by Generalized 2-Nearest Neighborhood (g2NN) technique, agglomerative hierarchical clustering, and geometric transformation estimation through homography computation among the clusters. Experimental results verify that our CMFD framework not only outperforms the state-of-the-art CMFD schemes for tampered images having similar but genuine objects, but also commensurates with the accuracy of state-of-the-art schemes for other copy-move forgery types. The proposed method has also been found to be robust with respect to several post-processing operations, e.g., compression and filtering.
Aniket Roy, Rahul Dixit, Ruchira Naskar, Rajat Subhra Chakraborty

Chapter 6. Copy-Move Forgery Detection in Transform Domain

Abstract
In this chapter, we present a copy-move forgery detection technique which utilizes Undecimated Dyadic Wavelet Transform (DyWT) for its operation. Dyadic Wavelet Transform (DyWT) is advantageous because it does not involve signal down-sampling and hence provides a better approximation compared to other wavelet transforms. Subsequently, we propose a technique to reduce the number of false positives obtained through application of this technique, to further improve the efficiency of the proposed scheme.
Aniket Roy, Rahul Dixit, Ruchira Naskar, Rajat Subhra Chakraborty

Chapter 7. Conclusion and Future Research Directions

Abstract
This chapter gives a brief summary of the work presented in this book. It also provides several future research directions and open problems related to the exciting and relatively young field of digital image forensics.
Aniket Roy, Rahul Dixit, Ruchira Naskar, Rajat Subhra Chakraborty
Additional information