IRJET- Image Forgery Detection using Support Vector Machine

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 06 Issue: 03 | Mar 2019 www.irjet.net p-ISSN: 2395-0072
© 2019, IRJET | Impact Factor value: 7.211 | ISO 9001:2008 Certified Journal | Page 4221
Image forgery detection using support vector machine
Dr PALANIVEL .N
Associate Professor,
Department of Computer Science and Engineering,
Manakula Vinayagar Institute of Technology,
Puducherry.
npalani76@gmail.com
ARTHI.Z
Department of Computer Science and
Engineering,
Manakula Vinayagar Institute of
Technology, Puducherry.
arthi033@gmail.com
DEEPIKA.G
Engineering,
deepikagurumurthy1998@gmail.com
LATHA.S
Engineering,
lathasl1509@gmail.com
Abstract- Now-a-days, it is very easy to manipulate an
image by adding or removing some features in an image
without leaving any clue of editing the original image. They
use advanced tools to digitally manipulate images to create
forged image without finding a clue on it to find the forged
region of an originality of images. These modifications are
not visible when we see it into the naked eye. Splicing and
Copy-move forgeries are most common forgery techniques.
In splicing forgery, a small region in one image is cut and
paste over an another image. Whereas in copy-move, a small
region of an image is copied and pasted over a same image.
The devices like cameras are getting more and more
digitized, there is an increase in the need for digital image
authentication, validation and forgery detection. This paper
has an approach for the Splicing and Copy-move forgery
detection. Copy-move and Splicing are the passive image
forgery detection techniques. Initially, an image is taken as
an input for both copy-move and splicing forgery. For both
copy-move and Splicing detection, pre-processing and
enhanced threshold methods are used to extract the features
in an image. After feature extraction, using SVM we find
whether it is authentic or forged by using RBF. If the given
input image is authentic then the output will be the Black
screen(No forged region). When the SVM identified it is
forged, then using PCA algorithm we remove the authentic
region and shows only the forged region as an output.
Keywords: Multimedia Technology; ulterior; Image
manipulation; investigations; Splicing ; Copy-Move ; Image
Forgery ; support- Vector Machine; Enhanced Threshold
Method; Radial Basis Function; Principle Component
Analysis.
I. INTRODUCTION :
With the availability of powerful digital image
processing platforms, such as Photoshop, it is relatively
very easy to create digital forgeries from one or more
images. Due to the development of computer technology
and image processing software, digital image forgery has
been increasingly easy to perform. However, digital images
are a popular source of information, and the reliability of
digital images is thus becoming an important issue. In
recent years, more and more researchers have begun to
focus on the problem of digital image tampering. Of the
existing types of image tampering, a common
manipulations of a digital image are cut-paste and copy-
move forgeries, which is to paste one or several copied
region of an image onto other parts of the same image or
on another image. plotting the cumulative graph shows
whether the image is splicing forged or not. During the
copy and move operations, some image processing
methods such as rotation, scaling, blurring, compression,
and noise addition are occasionally applied to make
convincing forgeries. Because the copy and move parts are
copied from the same image, the noise component, color
character and other important properties are compatible
with the remainder of the image some of the forgery
detection methods that are based on the related image
properties are not applicable in this case.
In previous years, many forgery detection
methods have been proposed for copy-move forgery
detection. According to the existing methods, the copy-

move forgery detection methods can be categorized into
two main categories: block based algorithms and feature
key point based algorithms. The existing block based
forgery detection methods divide the input images into
overlapping and regular image blocks then, the tampered
region can be obtained by matching blocks of image pixels
or transform coefficients. As an alternative to the block
based methods, key point based forgery detection methods
were proposed, where image key points are extracted and
matched over the whole image to resist some image
transformations while identifying duplicated regions.
II. LITERATURE SURVEY:
Pun et al. proposed a method which is based on noise
discrepancies between the original image and spliced
image. Initially, the noise level function is calculated and
analyzed at pixel level on various scales. The region which
is not under the noise level, termed as suspicious area and
inconsistent level of noise indicate the presence of
tampering in spliced segments. This technique performs
better for multiple spliced object detection. F. Hakim
proposed a method based on improved local binary
pattern (LBP) and discrete cosine transform (DCT). The
chrominance component of the image is divided into
nonoverlapping blocks. Then improved LBP is calculated
for all blocks and using 2D-DCT, it is transformed into
frequency domain. Further the frequency coefficients are
evaluated to find the standard deviation for all blocks
which are used as features for classification using k-
nearest neighbour.
Shi et al.proposed a natural image model, which educe
statistical moments of characteristics function by treating
the neighbouring differences of BDCT of an image as 1-D
signal and the dependencies between neighbouring nodes
along certain directions have sculpted as Markov model.
SVM classifier considered these features as discriminative
features for classification. Wang et al. proposed a method
in which gray level cooccurrence matrix (GLCM) is
considered along certain direction (horizontal, vertical,
main and minor diagonal) to extract edge images and
educed edge images serve as discriminative features for
classification. Xuefang Li et al. proposed a method in which
Hilbert-Huang transform and moment of characteristic
function of wavelet transform are used for forgery
detection. SVM is used as a classifier for spliced image
classification in their method and achieved an accuracy of
85.86%.
Qu et al. proposed a algorithm to detect splicing image
forgery with visual cues in 2009. Authors used a detection
window and divided it into nine sub-squares. VAM (visual
consideration model) is used to distinguish an obsession
point and afterward feature extraction is used to extract
the spliced region in the digital image.
III. TYPES OF DIGITAL IMAGE FORGERY:
Image forgery is characterized as ”inserting, replacement,
or removing some important features from an image
without leaving any clue. There are many different
techniques to utilize forging regions of an image. Taking
into account these method is used to make forged images.
Digital image forgery can be divided into three primary
classifications: Copy-Move, splicing, and Image re-
sampling.
A. Copy-move Forgery
In copy-move forgery, some portion of an image at any
size and shape is copied and pasted to another region in
the same image to modify some important feature. The
copied region is pasted over the same so that the pixel of
the block will varies. This is very difficult to find the forged
region in it. Digital image forgery techniques are used
many different techniques to manipulate an image without
leaving any clue on it.
B. Splicing Forgery:
Image splicing means a part of region of one image is cut
and paste over an another image to create a suspicious
image is called Image Splicing. When compared to copy-
move forgery detection Splicing Forgery detection is very
difficult to find. Previously many techniques are developed
to find the forged region in splicing but it results in failure.
When splicing is performed, the borders between the
spliced regions can visually be impossible. Splicing,
disturbs the high order Fourier statistics.
IV. PROPOSED WORK:
This paper proposes a technique to find a duplicate region
in an image. Duplicate region is finded by using the PCA
algorithm and SVM classifier.
Steps in Proposed work:
1. Dividing the grayscale image into fixed sized
overlapping blocks using SVM.

2. Extracting Gaussian RBF kernel PCA-based
features from each DCT square block.
3. Matching similar block pairs.
4. Removing the isolated block and output the
duplicated regions.
A. BLOCK DIAGRAM:
B. SUPPORT VECTOR MACHINE:
SVM is mainly used for classification purpose. In
order to avoid computational complexity it uses
recognition tools by high dimension. There are many
studies using SVM as a classifier in image forgery
Detection.
In this paper, a technique is developed to detect
image forgery which includes removal, addition, and
replacement of regions in an image. SVM classifier is used
to find similar regions of an image by matching image
blocks. Image, texture pixel value based features and edges
are extracted to analyze the images for forged regions.
After analyzing the image, hash values are calculated for
feature extraction. This process consists of two phases
they are training phase and a testing phase. In training
phase, SVM train a set of image. SVM is used to classify the
image whether it is authentic or forged.
SVM first identify the decision boundaries in the
training phase and then the technique will give the good
generalization in high dimensional input images.
Classification using SVM is mainly based on the concept of
decision making and that defines the decision boundaries.
A decision plane separates a set of objects having different
class memberships and a set of objects having different
class relationships. SVM determine a vectors called
"support vectors" that easily identify the separators which
gives the wide separation of classes and objects.
SVM classifier supports both the binary and
multiclass targets. Support Vector Machine models must
have a similar functional form for block based network and
radial basis functions, both are well-knowned data mining
techniques. Since, neither of these algorithms has the very
new theoretical approach to regularize the format, that
forms the basis of SVM. The quality of generalization and
ease of training in SVM is based on the capacities of those
traditional methods. The SVM map the original data points
from the input image to the high dimensional, feature
block making classification problem simpler in feature
space. This kind of mapping is done by a suitable choice of
a kernel basis function.
C. PCA:
The image is changed over from colour to grayscale.
The image is isolated into a few little sized blocks, which
are broken to vectors. This is vastly improved than the
BruteForce strategy for finding the matches. The PCA
technique is utilized to break the diverse blocks in an
option way. PCA is fit for recognizing even minor
variations because of noise and/or compression. This
strategy is just for grayscale images. Be that as it may, the
strategy can be made to work for colour images also by
preparing the image for every colour channel which
outcomes three duplication maps. At that point PCA is
connected to every map independently to recognize the
forgeries. This technique has a decent proficiency in
detecting Copy Move forgeries furthermore gives the less
number of false-positives.
In any case, the productivity drops as the piece
size reduces furthermore in case the nature of image is
low. The goal of PCA is to enlarge the variance between
data without considering class separation. There are
distinctive methodologies proposed for a feature
extraction piece. Distribution shows multidimensional raw

data which is every now and again troublesome. Normally,
evacuating features those are proposed to catch and
address the distributions in a lower-dimensional space
may unravel this mission.
The PCA is frequently used for pre-processing of
multi-spectral remote sensing images for the explanations
behind change detection. Change, regardless, is interesting
in connection to the interpretation that is used here. In
remote-sensing, the change is fathomed as the technique of
perceiving contrasts in the condition of an article in space
by watching it at various times, for instance a vegetable
canopy. In case there is no learning of what the change
might be, it is not clear whether the representations in a
lower-dimensional space will offer support. The PCA will
analyze each pixel of the image and then classify it.
D. PCA ALGORITHM:
Step 1: The input image should changed from color to
GreyScale Image.
Step 2: Then the image is isolated into few little sized
blocks and which are broken into vectors.
Step 3: Column or row vector of size N2 represents the set
of M images (B1, B2, B3…BM) with size N*N
Step 4: And then we need to calculate the co-variance
matrix.
Step 5: Then Measure the Eigen vectors and Eigen values
of these co-variance matrix of an image. The PCA is also
used to recognise even minor variation because of noise
and compression.
E. EXPERIMENTAL RESULTS:
The result shows only the duplicate region of the image.
 If the image which we give as an input is authentic
then the output result will be a black screen. The
black screen shows that the given image is
authenticated.
INPUT
Fig.1 Input image(Authentic image to
detect forged region)
OUTPUT:
Fig .2 Output Image(Which shows no forged
region in an image)
 If the given image is forged then the output result
will be the duplicate region of the image.

INPUT:
Fig 3: Original Image(Pure image with no
modification in it)
Fig 4: Forged image(small region is altered in an
image)
OUTPUT:
Fig 5: Output Image(shows the forged region)
V. CONCLUSION
With the increase in digital image forgery, the need
for forgery detection algorithms has increased. In this
paper, copy-move and splicing forgery detection are done
at the same time. The suspicious image is taken as an input
and then extract the features in it. Using SVM and PCA
algorithm the original region of an image is eliminated.
Then duplicate regions of an image is highlighted and
marked as an output. In future work, we focus on accuracy
of detecting duplicate regions.
VI. REFERENCE
[1]. Amerini S, Ballani P, Caldellan Y, Seeran G (2018) A
sift-based forensic method for copy–move detection and
transformation recovery. IEEE Trans Inf Forensics Secure
6(3):1099–1110
[2]. Amarini K, Bala K, Caldalli R, Bimbosh A, Tongo
LD,(2014) Copy-move forgery detection by means of
robust clustering with j-linkage. Signal Process Image
Communication 28(6):659–669
[3]. Andreopolos Z, Tsokso J (2012) Object recognition
directions forward. Computing Vis Image Understanding
117(8):827–891
[4]. Barne C, Shechtmoni P, Finkel L, Goldman CB (2009)
Patchmatch Algorithm: A randomized correspondence
algorithm for structural image editing. ACM Graph
28(3):24–1 Multimedia Tools Application.
[5]. Bayiram O, Sensar TH, Menon N (2008) An efficient
and robust method for detecting copy-move forgery. In:
Acoustics, Speech and Signal Processing, 2008. ICASSP
2009. IEEE International Conference on, pp 1053–1086.
IEEE
[6]. Bianchin M, Pivash K (2016) Image forgery localization
via block grained analysis of .jpeg artifacts. IEEE Transition
Information Forensics Secure 7(3):1003–1017
[7]. Bosh S, Yuansh Q, Wangelton S, Zhao C, Li S (2014)
Enhanced state selection markove model for image
splicing detection. EURASIP J Wirel Communication
Network 2014(1):7

[8]. Bravo-Solonio P, Nanditha TS (2011) Automatic
detection and localization of duplicate regions affected by
reflection, rotation and scaling in image forensics. Signal
Process 91(8):1754–1770
[9]. Campos FM, Correia L, Calado JMF (2015) Robot visual
localization through local feature fusion: an evaluation of
multiple classifiers combination approaches. J Intelligence
Robust System 77(2):377–390
[10]. Caoli Y, Gaola T, Li Feo, Yangush Q (2018) A robust
detection algorithm for copy-move forgery in digital
images. Forensic Sci Int 214(1):33–43
[11]. Chen L, Wei L, Ni J, Sun W, Huang J (2013) Region
duplication detection based on harrister corner points and
step sector statistics. J Vis Communication Image
Represent 24(3):244–254
[12]. Christleguin V, Riuess C, Jordan J, Rtiess C,
Angelopoulou E (2012) An evaluation of popular copy-
move forgery detection approaches. IEEE Trans
Information Forensics Secur 7(6):1841–1854
[13]. Dollar P, Wojevk C, Schiuele B, Peerona P (2012)
Pedestrian detection: An evaluation of the state of the art.
IEEE Trans Pattern Anal Mach Intellect 34(4):743–761
[14]. Donjbg J, Wakljng W, Tanley T (2013) Casia image
tampering detection evaluation database. In: Signal and
information processing (ChinasSIP), 2013 IEEE China
Summit & International Conference on, pp 422– 426. IEEE
[15]. El-Alfky E-SM, Quryeshi MA (2015) Combining spatial
and dct based markov features for enhanced blind
detection of image splicing. Pattern Anal Application
18(3):713–723
[16]. Fischler MA, Bollses RC (1981) Random sample
consensus: a paradigm for model fitting with applications
to image analysis and automated cartography. Commun
ACM 24(6):381–395
[17]. Fridrich AJ, Sodfukal BD, Luka´s AJ (2003) Detection
of copy-move forgery in digital images. In ˇ Inproceedings
of Digital Forensic Research Workshop. Citeseer
[18]. Gionisa A, Inidyk P, Maotwani R et al (1999)
Similarity search in high dimensions via hashing. In: VLDB,
vol. 99, pp 518–529
[19]. Gusdo J-M, Liu Y-F, Wu Z-J (2013) Duplication forgery
detection using improved daisy descriptor. Expert System
Appl 40(2):707–714
[20]. Hakshiimi F (2015) Image-splicing forgery detection
based on improved lbpogb and k-nearest neighbors
algorithm. Electron Inf Plan, 3

IRJET- Image Forgery Detection using Support Vector Machine

More Related Content

What's hot (20)

Similar to IRJET- Image Forgery Detection using Support Vector Machine (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- Image Forgery Detection using Support Vector Machine