Copy-move forgery detection using multiresolution local binary patterns

doi:10.1016/j.forsciint.2013.04.023

Forensic Science International

Volume 231, Issues 1–3, 10 September 2013, Pages 61-72

https://doi.org/10.1016/j.forsciint.2013.04.023 Get rights and content

Abstract

Copy-move forgery is one of the most popular tampering artifacts in digital images. In this paper, we present an efficient method for copy-move forgery detection using Multiresolution Local Binary Patterns (MLBP). The proposed method is robust to geometric distortions and illumination variations of duplicated regions. Furthermore, the proposed block-based method recovers parameters of the geometric transformations. First, the image is divided into overlapping blocks and feature vectors for each block are extracted using LBP operators. The feature vectors are sorted based on lexicographical order. Duplicated image blocks are determined in the block matching step using k-d tree for more time reduction. Finally, in order to both determine the parameters of geometric transformations and remove the possible false matches, RANSAC (RANdom SAmple Consensus) algorithm is used. Experimental results show that the proposed approach is able to precisely detect duplicated regions even after distortions such as rotation, scaling, JPEG compression, blurring and noise adding.

Introduction

Digital images play an important role in our daily lives due to their flexibility. But they are susceptible to malicious activities. According to Ref. [1], even in 1989, almost 10% of all color photographs published in United States has been altered and forged. This is mainly due to advent of low-cost hardware and sophisticated photo-editing software which alleviate digital images manipulation. Professional forgers with advanced technology often leave no trail on forged images. This trend decreases credibility of digital images which are presented in scientific journals, news items, medical record, and financial documents. Therefore, it is necessary to create new tools and techniques to discover the authenticity and integrity of digital images.

Several methods have recently been proposed to detect image forgeries: watermarking-based schemes [2], [3], perceptual hash-based schemes [4], and digital image forensic-based schemes [5], [6]. In the first group, the watermark information and ownership identification is embedded imperceptibly into the digital image. With the assumption that tampering will alter a watermark, an image can be authenticated by verifying the extracted watermark. However, this method has some significant limitations: (1) the watermark must be inserted at the time of recording where is often unavailable in practical applications, (2) this method can only be used for watermarked images and (3) the information embedding operation degrades image quality. In the second method, the perceptual hash function generates a hash value from the original image and it is stored by the authenticator. In the authentication process, a new hash value is computed from a suspected image and compared with the stored one to detect a forgery. Unlike watermarking methods, the hash-based schemes do not change the image quality but generation hash value during taking a photo is not always possible in practical applications. In contrast to the first two groups, digital image forensics techniques work in the absence of any digital watermark or signature. In the third category, the intrinsic features focusing on inconsistency and similarities in the image are extracted. Core of these techniques is appropriate features extraction which depends on the type of image forgery.

Copy-move forgery is one of the most popular tampering artifacts where a part of an image is copied and pasted to another part of the same image, with the aim of adding or hiding an object. Since duplicated regions come from the same image, they have similar properties like noise, color and texture. On the other hand, a forger can easily modify the geometry of the copied part, add noise, or compress the resulting image. So, a copy-move forgery detection technique should be robust to rotation, scaling, blurring, noise and compression.

In this paper, we propose a copy-move forgery system based on Multiresolution Local Binary Patterns (MLBP). This method is not only able to precisely detect the altered area, but also robust against rotation, scaling, JPEG compression, blurring and noise adding. In this approach, RANSAC (RANdom SAmple Consensus) algorithm is used to remove the false matches and estimate the geometric transformation parameters with high reliability. The rest of the paper is organized as follows. In Section 2, the related works on copy-move forgery detection are introduced. Section 3 briefly introduces LBP and its extensions. In Section 4, the proposed algorithm is presented. In Section 5, several experiments are performed for forensic analysis. Discussion and comparison with other methods are presented in Section 6. Finally, the conclusion is drawn in Section 7.

Section snippets

Related works

Many techniques have been proposed to address the problem of copy-move forgery detection. Block-based matching is one of the main methods for this purpose. First, the image is divided into overlapping blocks and then a feature extraction process is applied to represent the image blocks. In this step, the biggest challenge is to determine the features which yield the same or very similar values for duplicated blocks, even under various modifications. Since the similar regions would have similar

Local binary patterns (LBP) and its extensions

Among the feature descriptors, Local Binary Patterns (LBP) is one of the most famous and powerful ones. It has gained increasing attention in many image analyses applications due to its low computational complexity, invariance to monotonic gray-scale changes and texture description ability [20]. In practice, the LBP operator combines characteristics of statistical and structural texture analysis: it describes the texture with micro-primitives, often called textons, and their statistical

The proposed algorithm

In this section, details of the proposed method for duplicated and distorted region detection are presented. There are three main steps in our algorithm: feature extraction, block matching, and estimation of the geometric transformation parameters and remove the false matches. Fig. 3 shows the block diagram for our copy-move forgery detection scheme.

Experimental results

In this section, we first introduce experimental setup and evaluation metrics. Then, in the next section results of our algorithm are compared with two other methods which are based on SIFT [17] and DCT [8] descriptors. The former method is a feature-based technique, while the later technique is block-based.

Discussion and comparison with other methods

Experiments on the test images undergone post-processing operations such as scaling, JPEG compression, Gaussian blurring and AWGN show that the MLBP features can withstand these operations and allow forgery identification. Although duplicated regions with rotation through limited angles can accurately be detected.

The results in the Table 1 indicate that size of copied region and also type and number of MLBP operators has very small effect on average detection accuracy when no perturbation is

Conclusion

In this paper an efficient forensic method based on Multiresolution Local Binary Patterns (MLBP) for detecting copy-move forgery in digital images was proposed. The method does not need digital watermarks, signatures information or any metadata. The proposed method not only detects duplicated regions but determines the geometric transformations applied to the forged regions. Experimental results demonstrated that the proposed approach could even detect duplicated regions with common

References (31)

Y. Huang
Improved DCT-based detection of copy-move forgery in images
Forensic Sci. Int.
(2011)
Y. Cao et al.
A robust detection algorithm for copy-move forgery in digital images
Forensic Sci. Int.
(2012)
B. Mahdian et al.
Detection of copy-move forgery using a method based on blur moment invariants
Forensic Sci. Int.
(2007)
S. Bravo-Solorio et al.
Automated detection and localisation of duplicated regions affected by reflection, rotation and scaling in image forensics
Signal Process
(2011)
H. Shao
Image region duplication detection based on circular window expansion and phase correlation
Forensic Sci. Int.
(2012)
T. Ojala et al.
A comparative study of texture measures with classification based on featured distribution
Pattern Recog.
(1996)
C. Amsberry, Alterations of photos raise host of legal, ethical issues, The Wall Street Journal, Jan...
I.J. Cox et al.
Digital Watermarking
(2001)
I.J. Cox et al.
A review of watermarking principles and practices, In: digital signal processing for multimedia systems
(1999)
P. Blythe et al.
Secure digital camera

A. Rocha et al.

Vision of the unseen: Current trends and challenges in digital image and video forensics

ACM Comput. Surv (CSUR)

(2011)

H. Farid

A survey of image forgery detection

IEEE Signal Process. Mag.

(2009)

J. Fridrich et al.

Detection of copy-move forgery in digital images

A.C. Popescu et al.

Exposing digital forgeries by detecting duplicated image regions, Technical Report TR2004-515

(2004)

S. Bayram et al.

An efficient and robust method for detecting copy-move forgery

Cited by (121)

The detection of copy move forgery image methodologies
2023, Measurement: Sensors
In every area that uses digital photos, image security is a major concern. Photos of suspects, crime scene photos, biometric photographs, and other images have long been used in forensic investigations and public safety. With the evolution of digital imaging, the digital images used in this area have increased significantly. As much as digital image processing has facilitated the development of numerous new approaches in forensic research, it has also made image manipulation simple. Digital image validity has become a problem with the public availability of various snipping image manipulation software. As a result, the need for image tampering detection software has become essential. Digital photographs have become widely used in past years for a variety of applications. The material will be disseminated through print media, the website, as well as scholarly publications. It is utilized as powerful evidence in numerous offenses as well as documentation for something like a range of functions The introduction to the photo processing and editing tools has made it simple and accessible to create or change photographs. Are of the most frequent kind of picture forgery is a copy-move forgery, in which a portion like a photographic is duplicated as well as pasted further in a photo to conceal at some point crucial or display an error scenario. This study examines many sorts of digital image forgeries as well as the detection of forgery tools. A review of available approaches for faked images was conducted.
A new Copy-Move forgery detection method using LIOP
2022, Journal of Visual Communication and Image Representation
Citation Excerpt :
The copy-move forgery was identified by comparing the similarities of feature vectors detected by applying discrete cosine transformation (DCT) to each assigned block. The multi-resolution local binary patterns operators have been thoroughly tested for copy moving forgeries detecting by the researchers in [42]. The presented analysis shows that wide variety of conditions are only partially resistant to geometrical assaults.
The most prevalent type of digital image falsification occurs when a portion of a image is copied and pasted onto another section of the same image. Falsification of the image made in this way is called copy-move forgery (CMF). This study presents a new and effective approach for copy-move forgery detection (CMFD) using the Local Intensity Order Pattern (LIOP) to overcome the restrictions of existing CMFD techniques. The input image is first converted to a YCbCr color space and then split into Y, Cb, and Cr color channels. The LIOP features are then extracted from each color channel and all the features are combined. The feature vectors are ordered lexicographically and related features are detected by comparing the LIOP features. Although the LIOP feature has rarely been used in CMFD prior to this study, the success rate of the proposed method is high. In addition, since the channels are not correlated to each other in the YCbCr color space, each color channel is considered as a gray image, and the success rate is increased by combining the features extracted from each of the color channels. The proposed approach was assessed using the CoMoFoD and GRIP datasets. Experimental findings demonstrated that the suggested method was successful and displayed robustness in post-processing attacks.
Copy-move forgery detection using binary discriminant features
2022, Journal of King Saud University - Computer and Information Sciences
Citation Excerpt :
Most of the keypoint based methods rely on SIFT and SURF (Bay et al., 2006) features which although invariant to transformations, are computationally expensive and are not good in detecting flat duplicate regions. Several other local features are also used in copy-move forgery detection like Local Binary Pattern (LBP) (Davarzani et al., 2013), Binary Robust Invariant Scalable Keypoints (BRISK) (Kumar et al., 2015), DAISY (Sekhar and Shaji, 2015) etc. In spite of the lower computational complexity of feature extraction and matching, these methods have a poor performance in detecting forgery especially when the forged regions are smooth.
Digital images have numerous applications in various fields such as cyber forensics, courts, newspapers, magazines, medical field and many more. Therefore, the trustfulness of the contents of an image is very essential. Image manipulation has become very much simple and common due to the development of high resolution digital cameras, powerful computers and numerous image editing software. Hence, falsification of visual content of images is not at all limited to specialists and detection of forged images have become a highly challenging task. Of the different kinds of image forgery, the proposed method deals with one particular kind known as copy-move forgery. The method is an integration of the conventional block-based and keypoint based methods. The method first identifies the forged keypoint locations and then localizes the forged region using a region extraction technique. To improve the accuracy of detection, a Binary Discriminative Feature descriptor (BDF) is used for feature detection and matching. The suspected regions are identified by replacing the matching feature points with corresponding superpixel blocks. The neighbouring blocks are then merged with the suspected regions based on color similarity, using Color Histogram Matching and a final morphological close operation extracts the detected forged regions. Experimental results show that the proposed method has better detection accuracy, in terms of precision, recall and F1 Score, when compared with the state-of-the-art methods of copy-move forgery detection, under conditions of plain copy-move forgery as well as post-processing conditions like brightness changes, contrast adjustments, color reduction and blurring.
Comparative analysis of feature extraction and fusion for blind authentication of digital images using chroma channels
2021, Signal Processing: Image Communication
Blind authentication is one of the challenging techniques which has attracted considerable attention with the increasing security hacks on digital images. This paper evaluates different models based on feature extraction to detect digitally-altered images. Various feature-extraction methods have been investigated and compared including LBP, MSLBP, CSLBP, SLBP, WLD, LPQ, MSLPQ and DCT. Moreover, we propose a number of ways to combine a variety of features utilizing single and multi-scale representations of images. For classification, SVM is employed with different kernels to identify forged and authentic images. To evaluate the effectiveness of the investigated models, several experiments have been conducted using $k$ -fold cross-validation and computed performance measures for two benchmark image tampering datasets (CASIA v1.0 and CASIA v2.0). Additionally, we have conducted statistical analysis for the top-ten models and the results confirmed that the best models for CASIA v1.0 and CASIA v2.0 are MSLBP-DCT and MSLPQ, respectively. Further improvements have been achieved by integrating features from the three color channels (Y, Cb and Cr) with and without feature reduction using PCA and LLP. In this case, the results show that MSLPQ-DCT achieved a better accuracy of 98.56% on CASIA v1.0 with 1020 features and MSLPQ achieved a slightly better accuracy of 97.4% on CASIA v2.0 with 1536 features.
Fingerprint liveness detection using an improved CNN with the spatial pyramid pooling structure
2021, Advances in Computers
While fingerprint identification systems have been widely applied to daily life, how to protect them from presentation attacks has become a hot topic in the field of biometric verification. A feasible strategy of fingerprint recognition, called Fingerprint Liveness Detection (FLD), has attracted a lot of attention from researchers. Convolutional Neural Network (CNN) technology has been widely adopted by researchers in FLD, since it can automatically learn high-level semantic features from a large number of labeled samples. However, the main problem of CNN is that it only takes image of a fixed-size/scale as input to the model. Therefore, two common techniques, crop and warp operation, have been devised. Due to the lack of some discriminative spatial information, the final detection performance of the model is not ideal. In this chapter, a novel FLD method based on an improved CNN with Spatial Pyramid Pooling (SPP) is proposed to overcome the limitation of image size/scale. Moreover, in order to resolve the problem of insufficient images, pre-train model parameters are adopted using the auxiliary ImageNet 2012 dataset, which serve as the initialization of our model parameter. Afterwards, we fine-tune our model parameters, using training samples from LivDet 2011 and LivDet 2013 datasets. Lastly, we conducted an experiment with LivDet 2011 and LivDet 2013 datasets to test our model. The result shows that the performance of classification of our method is superior to other methods.
Fuzzy methaheuristic model for copy-move forgery detection on images
2024, Multimedia Tools and Applications

View all citing articles on Scopus

View full text

Copy-move forgery detection using multiresolution local binary patterns

Abstract

Introduction

Section snippets

Related works

Local binary patterns (LBP) and its extensions

The proposed algorithm

Experimental results

Discussion and comparison with other methods

Conclusion

Forensic Sci. Int.

Forensic Sci. Int.

Forensic Sci. Int.

Signal Process

Forensic Sci. Int.

Pattern Recog.

Digital Watermarking

A review of watermarking principles and practices, In: digital signal processing for multimedia systems

Secure digital camera

Vision of the unseen: Current trends and challenges in digital image and video forensics

ACM Comput. Surv (CSUR)

A survey of image forgery detection

IEEE Signal Process. Mag.

Detection of copy-move forgery in digital images

Exposing digital forgeries by detecting duplicated image regions, Technical Report TR2004-515

An efficient and robust method for detecting copy-move forgery