Electromagnetismlike Mechanism Descriptor with Fourier Transform for
a Passive Copy-move Forgery Detection in Digital Image Forensics
Sajjad Dadkhah
1
, Mario K
¨
oppen
1
, Hamid A. Jalab
2
, Somayeh Sadeghi
2
, Azizah Abdul Manaf
3
and Diaa Uliyan
4
1
Kyushu Institute of Technology, 680-4 Kawazu, Iizuka, 820-8502, Fukuoka, Japan
2
Faculty of Computer Science and Information Technology, University of Malaya, 50603 Kuala Lumpur, Malaysia
3
Advanced Informatics School, Universiti Teknologi Malaysia, 54100 Kuala Lumpur, Malaysia
4
Faculty of Information Technology, Middle East University, Amman, Jordan
dsajjad2@liveutm.onmicrosoft.com, mkoeppen@ieee.org, hamidjalab@um.edu.my,
{ssomayeh, diaa uliyan}@siswa.um.edu.my, azizaham.kl@utm.my
Keywords:
Authentication, Digital Forensics, Forgery Detection, Image Analysis, Image Processing.
Abstract:
Copy-move forgery is a special type of forgery that involves duplicating one region of an image by covering
it with a copy of another region from the same image. This study develops a simple and powerful descriptor
called Electromagnetismlike mechanism descriptor (EMag), for locating tampered areas in copy-move for-
gery on the basis of Fourier transform within a reasonable amount of time. EMag is based on the collective
attraction-repulsion mechanism, which considers each images pixel as an electrical charge. The main compo-
nent of EMag is the degree of the attraction-repulsion force between the current pixel and its neighbours. In
the proposed algorithm, the image is divided into similar non-overlapping blocks, and then the final force for
each block is evaluated and used to construct the tampered image features vector. The experimental results
demonstrate the efficiency of the proposed algorithm in terms of detection time and detection accuracy. The
detection rate of the proposed algorithm is improved by reduction of false positive rate (FPR) and increment
of true positive rate (TPR).
1 INTRODUCTION
Over the recent years, the improvement of the com-
puter knowledge and digital imagery equipment ex-
panded the use of digital images into numerous areas,
such as TV, journalism, medical imaging and etc (Fri-
drich, 1999). Moreover, the professional image pro-
cessing tools such as Photoshop, which enable the
easy manipulation of the digital images, have become
widely available for free. The existence of such po-
werful tools raise suspicions on the integrity of the
digital images (Jing and Shao, 2012; Dadkhah et al.,
2014; Ardizzone et al., 2015).
Digital image forgery is accordingly defined as a
tool that mainly solve the integrity problem of the di-
gital images. Image forgery is the science of changing
parts of an image to create a fake image for illegal pur-
poses. Image forgery is divided into three groups: (i)
Image splicing creates a fake image by cutting a part
of the image and pasting it to another image (Chen
et al., 2007). (ii) Image retouching is common in
magazine photo editing and does not visibly change
the image which is the least corrupting type of digital
image forgery (Li and Wang, 2012). (iii) Copy-move
forgery which is the most significant digital forgery,
involves duplicating one region of an image by cove-
ring it with a copy of another region from the same
image (Piva, 2013).
Copy-move forgery is widely utilized for illegal
purposes, in order to conceal or emphasize certain
image details by cloning an area of an image to a dif-
ferent area. This type of forgery has become notori-
ous, as its detection requires greater technical skills.
It is because the source and destination of the forged
image are same (Fridrich et al., 2003; Kirchner et al.,
2015). This paper focuses on this category of for-
gery. Furthermore, various detection methods have
been proposed to detect copy-move forgery attacks.
Different digital image forgery methods are investi-
gated in this study. The proposed algorithm in this
research develops a simple and powerful descriptor
for locating tampered regions in copy-move forgery
612
Dadkhah, S., Köppen, M., Jalab, H., Sadeghi, S., Manaf, A. and Uliyan, D.
Electromagnetismlike Mechanism Descriptor with Fourier Transform for a Passive Copy-move Forgery Detection in Digital Image Forensics.
DOI: 10.5220/0006232206120619
In Proceedings of the 6th International Conference on Pattern Recognition Applications and Methods (ICPRAM 2017), pages 612-619
ISBN: 978-989-758-222-6
Copyright
c
2017 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
on the basis of Fourier transform within a reasonable
amount of time.
The proposed EMag descriptors utilizes an
attraction-repulsion mechanism to move the sample
points towards the optimality. EMag has both been
successfully applied to the solution of different sorts
of engineering problems such as resource constraint
project scheduling problems (Turabieh and Abdul-
lah, 2011), image processing (Jalab and Abdullah,
2013),(Cuevas et al., 2012) and neural network trai-
ning (Jalab and Shaker, 2014).
Since the Electromagnetism-like Mechanism is a
new heuristic algorithm for global optimization and
there are fewer investigations about this algorithm
until now, the authors of this research are motivated
to utilize the EMag algorithm as an image feature
descriptor to be used for image copy-move detection.
The remainder of this paper is organized as follows:
Section 2 presents related work on copy-move forgery
detection. Section 3 explains the details of the pro-
posed method. Section 4 presents the Performance
analysis and experimental results. The research con-
clusions are presented in Section 5.
2 RELATED WORK
As mentioned in previous section, the source and des-
tination of the forged image in copy-move forgery is
same, such that detecting the forgery by using the na-
ked eye is almost impossible (Fridrich et al., 2003).
Some operations (e.g., rotation, JPEG compression,
resizing, and noise) are usually applied to the original
part before pasting. These post- processing transfor-
mations make the detection process more difficult, for
instance JPEG compression transmitted digital ima-
ges to compressed format. A forgery detector should
be robust to all manipulations and applicable to com-
pressed images (Wu and Chang, 2002). Thus, the de-
tection of forgery using methods that search for in-
compatibilities inside the digital image is impossible
(Farid and Lyu, 2003). Several copy-move forgery
detection techniques have been proposed to solve this
issue.
A copy-move detection method is proposed by
(Ardizzone and Mazzola, 2009) that utilize bit-plane
analysis. In this method, the image is analyzed in the
bit-plane domain. Blocks of bits are encrypted using
the ASCII code for each bit-plane, and the direction
of the strings is examined instead of the original bit-
plane. The cycle is classified, and similar groups of
bits are removed as doubtful areas, which are then
passed on to the next plane for processing. The output
of the previous planes shows where the image is chan-
ged. The execution time in this method is reasonable,
which is an advantage. However, bit-plane analysis
does not work with JPEG images because JPEG com-
pression and bit-plane representation are not related.
Bit-plane analysis is also ineffective when the pasted
area is rotated or scaled.
Authors in (Lin et al., 2009) proposed a method
that uses a radix sort. This approach works in a man-
ner that is similar to other block generation methods.
Radix sorting is utilized instead of lexicographic sor-
ting to improve time complexity and to enhance the
resistance to various noises and compressions (e.g.,
JPEG compression and Gaussian noise). However, ra-
dix sorting does not deal with rotation arbitrary angles
and cannot successfully detect small copied regions.
Ryu (Ryu et al., 2010) proposed a detection sy-
stem based on Zernike moments. Zernike moments
are used to extract the feature vectors of an image
block. Then, the features are sorted lexicographi-
cally and adjacent vectors are located. This method
works well in terms of robustness to noise and rota-
tion because Zernike moments are algebraically in-
variant against rotation, noise, and information con-
tent. However, it is weak against other transformati-
ons, such as scaling or JPEG compression.
Hu (Hu et al., 2011) proposed a detection met-
hod based on the discrete cosine transform (DCT).
His method is the improved version of Fridrich algo-
rithm, which is based on DCT. In this method, featu-
res of each image block are compressed together and
determine whether the number of matched blocks in
a certain area is more than a specific threshold. For
improving matching accuracy, lexicographical sorting
algorithm based on distance is proposed. This method
is not robust to rotation but it is robust to noise and
blurring.
Liu (Liu et al., 2011) proposed a passive image au-
thentication method that can detect duplicated areas
under rotation, which uses round blocks and Hu mo-
ments to determine forged areas. The image is de-
composed using a Gaussian pyramid to create sub-
images. Low frequency sub-images are chosen to
overcome possible distortion caused by noise conta-
mination and JPEG compression. Subsequently, each
sub-image is divided into several non-overlapping
round blocks. Hu moment features are extracted from
these blocks and used to match features. Finally, the
forged regions are located by comparing the shift vec-
tors and copy-region areas. The method mainly is
not robust to resizing or cropping images before the
image is pasted to another area.
Hou (Hou et al., 2012) proposed a copy-move de-
tection algorithm by using phase correlation within an
image. The advantage of this method is its low com-
Electromagnetismlike Mechanism Descriptor with Fourier Transform for a Passive Copy-move Forgery Detection in Digital Image Forensics
613
putational complexity. It is also able to detect small
tampered areas because it uses a larger overlap ratio.
Although, phase-correlation based methods are able
to detect small areas. However, if the image contains
multiple forged regions, it cannot detect copy-move
areas.
Mishra (Mishra et al., 2013) presented a tamper
detection method based on speeded up robust fea-
tures (SURF) and hierarchical agglomerative cluste-
ring (HAC). SURF is used to speed up the process
of keypoint extraction while using HAC to group up
keypoints. Authors used Haar wavelets to compute
descriptors which are robust to illumination changes.
This method is robust to noise and JPEG compression
but the result of their proposed algorithm is not satis-
factory in terms of the true positive detection rate.
Silva (Silva et al., 2015) proposed a digital image
tamper detection algorithm based on multi-scale ana-
lysis and voting processes of the image. In their met-
hod, interest points are extracted from the image ba-
sed on geometric constraints, then a multi-scale repre-
sentation are created, and formed the groups tested by
utilizing a robust descriptor. Their proposed method
is robust to rotation and scale but it is not good enough
for JPEG compression attack. Robust copy-move for-
gery detection is essential; the detector should be ro-
bust to post-processing operations and some types of
transformation. Most existing methods cannot deal
with all these manipulations and are often computati-
onally expensive.
3 PROPOSED METHOD
In this paper an efficient approach for detecting copy-
move forgery in digital images by proposing a new
EMag descriptor for locating copy-move tampered re-
gions is presented. The proposed Electromagnetism
Mechanism Descriptor with Fourier Transform is des-
cribed in the following stages: EMag Block feature
extraction, Block feature matching and tamper locali-
zation.
3.1 EMag Block Feature Extraction
The proposed EMag algorithm for Copy-move de-
tection is illustrated in Figure 1 . As illustrated in
Figure 1, the general procedure of the proposed algo-
rithm are as follows:
1. Dividing the digital image into blocks of pixels
with appropriate size.
2. Extraction of certain feature of each block by pro-
posed EMag algorithm and Fourier transforma-
tion.
3. Block feature matching and tamper localization.
The similar blocks are connected to localize the
copied region when tampering is detected.
The proposed algorithm in this paper has explo-
red a special feature inside digital images which is
influenced by Fourier transformation. The proposed
EMag algorithm extract the Electromagnetism des-
criptors within each blocks. Finally, the final force for
each block is calculated and used to construct the tam-
pered image features vector. However, the details of
the proposed EMag Block feature extraction are des-
cribed in the following steps:
Step 1. Preprocessing. The original image I is con-
verted into grayscale by equation (1).
I = 0.299R + 0.587G + 0.114B (1)
where R, G, and B are three channels of the
input color image and I is its luminance com-
ponent (Lin et al., 2009).
The grayscale image is then divided into
similar-sized non-overlapping blocks (B × B
pixels), which are smaller than the sizes of
the detected duplicated regions. The default
block size is 21×21 pixels.
The proposed method is analysed to deter-
mine the best block size value for attaining
the highest TPR and lowest FPR scores. Spe-
cifically, the block size value affects the num-
ber of matched points. A proper block size
Figure 1: The general procedure of the proposed copy-move forgery detection.
ICPRAM 2017 - 6th International Conference on Pattern Recognition Applications and Methods
614
value is therefore required to reduce the num-
ber of false matches. The goal is to maxi-
mise the TPR value while suppressing FPR.
The best FPR is the lowest value which me-
ans only a few percentage of the original ima-
ges are incorrectly recognized as forged ones,
while the best TPR is the highest value which
means all the forged images are correctly re-
cognized as forged.
Several block size values are tested to gauge
their influences on the identification of the
forged and original images, the best block
size value is empirically found to be 21×21,
and the best FPR is 6.2% and the best TPR is
95.4%. The number of blocks in the image is
calculated by [M/B]×[N/B] where M, N are
the image pixels.
Step 2. Dividing into sub-blocks for EMag Process.
Each block size B × B is divided into non-
overlapping sub-blocks of size 3 × 3.
Step 3. Electromagnetism descriptors. For each
block size B × B pixels, the total electrical
force Fi is calculated. The total electrical
force Fi is calculated for each 3×3 sub-block.
In the EMag implementation, the charge of pixels
q represents the value of image pixel. However, F
i
is calculated based on the electromagnetism theory
which states that the force exerted on a point charge
via other charges is inversely proportional to the dis-
tance between the charges and directly proportional
to the product of their charges. The overall resultant
attraction-repulsion force for each image block deter-
mines the actual feature of forgery image. The final
force vector for each image block is evaluated under
the Coulombs law by equation (2).
F
i
=
n
j6=i
(
x
i
×x
j
)
k
x
i
x
j
2
k
i f f (x
j
) < f (x
i
)
(
x
i
×x
j
)
k
x
j
x
i
2
k
i f f (x
j
) f (x
i
)
(2)
where i=1,2,.,n (n=9), × is multiplication, and x
j
,
x
i
are the value of the center pixel in the image sub-
block and its surrounding pixels, respectively. In this
formula, f (x
j
) < f (x
i
) represents attraction and f (x
j
)
> f (x
i
) corresponds to repulsion.
Step 4. Feature extraction by Fourier transform. A
two-dimensional discrete Fourier transform
is applied to the extracted EMag features,
which are translation invariant. Fourier trans-
form is utilized to formulate a function with
an intensity signal across the image. This
function is disjointed into a sum of orthogo-
nal functions. The two-dimensional discrete
Fourier transform of f(m, n) is given by equa-
tion (3).
F(k, l) =
m=
n=
f (m, n)e
jkm
e
jln
(3)
where f(m,n) is the image in the spatial dom-
ain, and the exponential term is the basis
function corresponding to each point F(k,l) in
the Fourier space. k and l are the frequency
variables. F(k,l) is frequency domain repre-
sentation of f(m, n). F(k,l) is a complex va-
lued function that is periodic both in k and l
with a period of 2 π and period range of π
k, l π (Rosenfeld and Kak, 2014).
Fourier transform is applied to each block of the
image B
i
to perform a correlation, which can help in
identifying similar correlation values in an image and
locate the location of the matched blocks.
3.2 Block Feature Matching and
Tamper Localization
The procedure of the proposed Block feature mat-
ching and tamper localization algorithm is described
in the following steps:
Step 1. Correlation computation for each blocks.
Correlation C
i
is computed between every
two blocks which is defined as the convolu-
tion of the individual blocks to locate the fea-
tures within the image.
Step 2. Sorting correlation values. For faster tamper
localization procedure all correlation values
C
B
i has to be sorted. All correlations are sor-
ted with a k-d tree (Bentley, 1975) and saved
in a matrix.
Step 3. Block matching. All blocks are compared
after the sorting to determine their simila-
rity on the basis of the block-matching thres-
hold. The matching process works by com-
puting where the maximum correlation value
exceeds the threshold and then determining
whether the two blocks are similar.
Step 4. Eliminating False Blocks. When two blocks
are identified as similar, they are not neces-
sarily matched blocks. In some images (e.g.,
sky or nature), several blocks are similar to
each other. Numerous similar blocks should
be present at a specific distance to ensure that
Electromagnetismlike Mechanism Descriptor with Fourier Transform for a Passive Copy-move Forgery Detection in Digital Image Forensics
615
similar blocks are copied and pasted. Thus,
the Euclidean distance is used to calculate be-
tween two similar blocks and identify the du-
plicated blocks. the Euclidean distance is cal-
culated by equation (4).
Pdist(p
1
, p
2
) =
q
(x
2
x
1
)
2
+ (y
2
y
1
)
2
(4)
where x
1
, x
2
, y
1
and y
2
are the coordinates of
the matched points.
Step 5. Tamper localization. The inverse of a proce-
dure is conducted on the transformed image
to retrieve the original image with a tam-
pered region map. The inverse of a two-
dimensional Fourier transform is computed
by equations (5)-(7). Figure 2 demonstra-
tes the procedure of proposed block matching
and tamper localization.
Figure 2: Block matching procedure.
By using equations (5)-(7), the spatial domain image
is first transformed into an intermediate image using
the N one-dimensional Fourier transform, which is
then transformed into the final image. The final image
reveals the tampered region on the basis of the loca-
tion of copied and pasted regions.
f (m, n) =
1
4π
2
Z
π
w
1
=π
Z
π
w
2
=π
F(w
1
, w
2
)
e
j
w
1
me
j
w
2
ndw
1
dw
2
(5)
where
F(k, l) =
1
m
m1
j=0
p(k, j)e
i2π
i j
m
(6)
and
p(k, j) =
1
m
m1
i=0
f (i, j)e
i2π
k j
m
(7)
Where w
1
and w
2
are frequency variables, and
F(w
1
, w
2
)is frequency-domain representation of f(m,
n) (Rosenfeld and Kak, 2014).
4 EXPERIMENTAL RESULTS
The proposed method is evaluated with a 2.0 GHz In-
tel Pentium processor and 4 GB of RAM and Mat-
lab 2013a. The performance of the proposed forgery
detection method is evaluated on a dataset that con-
sists of 100 images with different contents from the
Columbia photographic image repository (Ng et al.,
2005) and our personal collection. The images varied
in size, format, and shape of the duplicated areas.
To evaluate the robustness and sensitivity of the
proposed method, detection performance is measu-
red in terms of TPR equation (7) and FPR equation
(8). TPR is the fraction of forged images correctly
recognized as forged, and FPR is the fraction of ori-
ginal images that are not correctly recognized as ori-
ginal (Mishra et al., 2013). The value of FPR, TPR,
and time is calculated, and an evaluation is performed
with other existing methods.
T PR =
# forged images detected as forged
# forged images
(8)
FPR =
# original images detected as forged
# original images
(9)
Table 1 shows the processing time on average (in
seconds) for an image. The results indicate that the
proposed method performs better with respect to the
others methods; in fact the processing time of Kangs
method is 60 seconds for a grayscale image with a
dimension of 256×256.
Table 1: Comparison result of proposed method with other
methods.
Method
Time(S)/
Grayscale
Time (S)/
Colour
(Kang and Wei, 2008) 60 120
(Li and Yu, 2010) N/A 44
Proposed Method 7 15
Proposed Method 5.44 8.19
By contrast, the detection time for the same image
is 5 seconds for the proposed algorithm. The average
ICPRAM 2017 - 6th International Conference on Pattern Recognition Applications and Methods
616
(a) (b) (c) (d)
(e) (f) (g) (h)
(i) (j) (k) (l)
Figure 3: The original images are in the top row (a-d). The forged images are in the middle (e-h) and the last row shows the
detected region(i-l).
runtime of the Kangs algorithm (Kang and Wei, 2008)
for a 256×256 color image is approximately 120 se-
conds; compare with the proposed algorithm for the
same size of color image, it takes 8 seconds to iden-
tify the duplicated areas. Several different images
are used in the experiment, which are challenging for
copy-move forgery detection with different sizes of
copied regions.
Figure 3 illustrates the detection results of the pro-
posed method. Figure 3i to Figure 3l show the dupli-
cation detection map of the proposed method applied
Figure 4: FPR and TPR for different JPEG quality factors.
to the tempered images in Figure 3e to Figure 3h. As
Figure 3i to Figure 3l illustrated, the proposed method
can accurately detect the duplicated regions.
As illustrated in Figure 4, when the different qua-
lity factors of JPEG compression is applied, there is
only a slight change in the value of FPR and TPR .
To evaluate the accuracy of the proposed algorithm,
detection rates (FPR and TPR) are computed for dif-
ferent quality factors applied to all the images in the
Columbia dataset. Figure 4 demonstrates the accepta-
ble results of the proposed algorithm in terms of FPR
and TPR.
In table 2, the performances of the proposed al-
gorithm in terms of authenticity detection (FPR and
TPR) is reported. The experimental results have il-
Table 2: Comparison of proposed method based on TPR
and FPR.
Method FPR(%) TPR(%)
(Fridrich et al., 2003) 84 89
(Popescu and Farid, 2004) 86 87
(Bashar et al., 2010) 0.12 32.1
(Mishra et al., 2013) 3.64 73.64
(Silva et al., 2015) 0.12 71.92
Proposed Method 6.2 95.4
Electromagnetismlike Mechanism Descriptor with Fourier Transform for a Passive Copy-move Forgery Detection in Digital Image Forensics
617
(a) (b) (c) (d)
(e) (f) (g) (h)
(i) (j) (k) (l)
Figure 5: Detection results (i-l) of our method on a set of forgery images with duplicated regions undergone different types of
distortion (e-h), and (a-d) shows the original images.
lustrated the superiority of the proposed algorithm in
comparison to other schemes in terms of efficiency.
Figure 5 shows the detection results of the propo-
sed method on some realistic forgeries. The forgery
images in the second row are generated with Pho-
toshop and Gimp. The results show the robustness of
the proposed method against JPEG compression and
Gaussian noise. The duplicated regions in Figure 5e
are small, and the forged image is under JPEG com-
pression. Figure 5f is a tampered image with Gaus-
sian noise. Figure 5h shows that the proposed method
can detect forgery; even when the copied and pasted
regions are similar to each other.
5 CONCLUSIONS
In this paper, an efficient copy-move forgery detection
algorithm based on Electromagnetism Mechanism is
proposed. The proposed EMag descriptors are uti-
lized for locating the copy-move tampered regions.
The degree of the attraction-repulsion force between
each pixel and its neighbour is calculated, and an
accurate description of the electrical forces between
two objects are utilized to distinguish between forged
and original pixels. The proposed false block elimi-
nation, eliminates the incorrect identified blocks with
similar structures which greatly influence on the result
of TPR and FPR. The proposed block size of 21×21
pixels and sub-blocks of 3×3 pixels creates a high ef-
ficiency in locating small tampered regions. Howe-
ver, the performance analyse and experimental result
clearly demonstrate the efficiency of the proposed al-
gorithm in terms of scaling, detection time, robust-
ness against different noises ratio, JPEG compression
and rotation. Future research include capability of de-
tecting image splicing and image retouching.
ACKNOWLEDGEMENTS
The author of this article would like to thank Kyushu
Institute of Technology, University of Malaya and
Universiti Teknologi Malaysia for thier educational
support.
REFERENCES
Ardizzone, E., Bruno, A., and Mazzola, G. (2015). Copy–
move forgery detection by matching triangles of
ICPRAM 2017 - 6th International Conference on Pattern Recognition Applications and Methods
618
keypoints. IEEE Transactions on Information Foren-
sics and Security, 10(10):2084–2094.
Ardizzone, E. and Mazzola, G. (2009). Detection of dupli-
cated regions in tampered digital images by bit-plane
analysis. In Image Analysis and Processing–ICIAP
2009, pages 893–901. Springer.
Bashar, M., Noda, K., Ohnishi, N., and Mori, K. (2010).
Exploring duplicated regions in natural images.
Bentley, J. L. (1975). Multidimensional binary search trees
used for associative searching. Communications of the
ACM, 18(9):509–517.
Chen, W., Shi, Y. Q., and Su, W. (2007). Image splicing de-
tection using 2-d phase congruency and statistical mo-
ments of characteristic function. In Society of photo-
optical instrumentation engineers (SPIE) conference
series, volume 6505, page 26. Citeseer.
Cuevas, E., Oliva, D., Zaldivar, D., P
´
erez-Cisneros,
M., and Sossa, H. (2012). Circle detection using
electro-magnetism optimization. Information Scien-
ces, 182(1):40–55.
Dadkhah, S., Manaf, A. A., and Sadeghi, S. (2014). An ef-
fective svd-based image tampering detection and self-
recovery using active watermarking. Signal Proces-
sing: Image Communication.
Farid, H. and Lyu, S. (2003). Higher-order wavelet statis-
tics and their application to digital forensics. In IEEE
workshop on statistical analysis in computer vision,
volume 8, page 94. Citeseer.
Fridrich, A. J., Soukal, B. D., and Luk
´
a
ˇ
s, A. J. (2003). De-
tection of copy-move forgery in digital images. In in
Proceedings of Digital Forensic Research Workshop.
Citeseer.
Fridrich, J. (1999). Methods for tamper detection in digi-
tal images. In Multimedia and Security, Workshop at
ACM Multimedia, volume 99, pages 29–34.
Hou, D. M., Bai, Z. Y., and Liu, S. C. (2012). A new algo-
rithm for image copy-move forgery detection. In Ad-
vanced Materials Research, volume 433, pages 5930–
5934. Trans Tech Publ.
Hu, J., Zhang, H., Gao, Q., and Huang, H. (2011).
An improved lexicographical sort algorithm of copy-
move forgery detection. In Networking and Distribu-
ted Computing (ICNDC), 2011 Second International
Conference on, pages 23–27. IEEE.
Jalab, H. A. and Abdullah, N. A. (2013). Content-based
image retrieval based on electromagnetism-like me-
chanism. Mathematical Problems in Engineering,
2013.
Jalab, H. A. and Shaker, K. (2014). Training the neural
networks by electromagnetism-like mechanism based
algorithm. In INTERNATIONAL CONFERENCE ON
QUANTITATIVE SCIENCES AND ITS APPLICATI-
ONS (ICOQSIA 2014): Proceedings of the 3rd Inter-
national Conference on Quantitative Sciences and Its
Applications, volume 1635, pages 582–586. AIP Pu-
blishing.
Jing, L. and Shao, C. (2012). Image copy-move forgery
detecting based on local invariant feature. Journal of
Multimedia, 7(1).
Kang, X. B. and Wei, S. M. (2008). Identifying tampered
regions using singular value decomposition in digital
image forensics. In Computer Science and Software
Engineering, 2008 International Conference on, vo-
lume 3, pages 926–930. IEEE.
Kirchner, M., Sch
¨
ottle, P., and Riess, C. (2015). Thin-
king beyond the block: block matching for copy-move
forgery detection revisited. In SPIE/IS&T Electronic
Imaging, pages 940903–940903. International Society
for Optics and Photonics.
Li, W. and Yu, N. (2010). Rotation robust detection of copy-
move forgery. In Image Processing (ICIP), 2010 17th
IEEE International Conference on, pages 2113–2116.
IEEE.
Li, Y. and Wang, H. (2012). An efficient and robust met-
hod for detecting region duplication forgery based on
non-parametric local transforms. In Image and Signal
Processing (CISP), 2012 5th International Congress
on, pages 567–571. IEEE.
Lin, Z., He, J., Tang, X., and Tang, C.-K. (2009). Fast,
automatic and fine-grained tampered jpeg image de-
tection via dct coefficient analysis. Pattern Recogni-
tion, 42(11):2492–2501.
Liu, G., Wang, J., Lian, S., and Wang, Z. (2011). A passive
image authentication scheme for detecting region-
duplication forgery with rotation. Journal of Network
and Computer Applications, 34(5):1557–1565.
Mishra, P., Mishra, N., Sharma, S., and Patel, R. (2013).
Region duplication forgery detection technique based
on surf and hac. The Scientific World Journal, 2013.
Ng, T.-T., Chang, S.-F., Hsu, J., and Pepeljugoski, M.
(2005). Columbia photographic images and photorea-
listic computer graphics dataset. Columbia Univ., New
York, ADVENT Tech. Rep, pages 205–2004.
Piva, A. (2013). An overview on image forensics. ISRN
Signal Processing, 2013.
Popescu, A. C. and Farid, H. (2004). Exposing digital for-
geries by detecting duplicated image regions. Dept.
Comput. Sci., Dartmouth College, Tech. Rep. TR2004-
515.
Rosenfeld, A. and Kak, A. C. (2014). Digital picture pro-
cessing, volume 1. Elsevier.
Ryu, S.-J., Lee, M.-J., and Lee, H.-K. (2010). Detection of
copy-rotate-move forgery using zernike moments. In
International Workshop on Information Hiding, pages
51–65. Springer.
Silva, E., Carvalho, T., Ferreira, A., and Rocha, A. (2015).
Going deeper into copy-move forgery detection: Ex-
ploring image telltales via multi-scale analysis and vo-
ting processes. Journal of Visual Communication and
Image Representation, 29:16–32.
Turabieh, H. and Abdullah, S. (2011). An integrated hy-
brid approach to the examination timetabling problem.
Omega, 39(6):598–607.
Wu, H.-C. and Chang, C.-C. (2002). Detection and restora-
tion of tampered jpeg compressed images. Journal of
Systems and Software, 64(2):151–161.
Electromagnetismlike Mechanism Descriptor with Fourier Transform for a Passive Copy-move Forgery Detection in Digital Image Forensics
619