International Journal of Recent Technology and Engineering (IJRTE)

ISSN: 2277-3878, Volume-8, Issue-2S3, July 2019

Digital Image Forgery Detection

[Link] Shyry, Saranya Meka, Mahitha Moganti 

image or leaving out certain parts of an image is

ABSTRACT--- With the growing challenges in authenticity acceptable unless they are mentioned and differentiated
and integrity of images, image manipulation has crumbled preferable by using boxes that portray the different parts of
assurance over digital image. The major motivation of the
the image.
forgery in image is manipulating the image in such a way that it
cannot be distinguished to the naked eye. Image manipulation Semantic sense is really changed in Malignant
has increased the demand to assess the trustworthiness of digital manipulation and this fashion should not be repeated.
images when used in crime investigation, as witness of law and Moreover it never performs adding, moving or removing
for surveillance purposes. In this paper, various types of image objects within the frame, changing the color other than to
forgery and detection techniques have been explained. Initially restore what the picture actually looks like, to alter its
different kinds of forgery attacks are categorized and summary of
interpretation, cropping a frame in order, flopping an image
passive approach is discussed.
Index Terms: Digital Image; Forgery; Forgery detection;
either left or right reversal, and lastly painting a photograph
Copy move forgery; Tampering in other than its true orientation. Image forgery has two
flavours namely Active and passive based approach. Non-
I. INTRODUCTION blind/intrusive methods are commonly referred to as Active
methods and it need major processed data to be embedded in
Forgery is an illegal means of manipulating images or
the original image during the recording.
documents without prior access. Images are tampered for
Due to this necessitous, active approaches have restricted
different reasons either to create false evidence or to earn
scope. Digital watermarking and digital signatures are some
money in an illegal way. An pictorial representation of
of the examples. Watermarking involves injecting a
image conveys much better idea than the words of human.
watermark which is used for the authenticity of the digital
Due to the progression in digital technology, images are
image which is indivisible from the image. On the other
proceesed using several tools like Adobe Photoshop, GIMP
hand, Passive methods are the non-intrusive/blind methods
and Corel Paint Shop and they ended up with a threat for the
and it never needs any prior information to include in the
authenticity of digital images. Generally, image
digital image. A digital image can be tampered by different
manipulations are of two types a) Allowed manipulation b)
attacks like resizing, adition of noise, blurring, rotation,
Malignant manipulation.
scaling compressing, image splicing, copy-move and many
more.

II. LITERATURE WORK

Toqeer et al detected the copy-move forgery attack, in
which the images are splitted in overlapping square blocks
and for the block representations DCT components are
adopted. Reduced dimensional nature of the feature space is
required to improve the efficiency of image matching and so
Gaussian RBF Kernel PCA is implemented. The proposed
Fig [Link] Image Fig 1 Fake Image method is compared with the traditional methods and their
results revealed best credibility of images with better
Allowed or incidental manipulations are the ones which efficiency.
never alters the semantic sense of information and are Anita Sahani et al inroduced a method for image
acceptable by any authentication system. The edits made forgery detection which plays the role for replacement,
should be very minor and subtle. Manipulation of images is insertion and removal of [Link] classifier which has a
generally allowed when correcting the color, tuning the comparable utilitarian shape like neural systems is utilized.
brightness and contrast of the photo, fiiting a layout using At first Image, the values of surface and pixel are extracted
cropping a frame, reducing the noise like dust ,dirt or and after analyzation hash values are [Link]
scratches in the photo. Combining certain parts of whole strategy comprises of two stages namely preparing or
training stage and testing stage. SVM classifies both original
and fake images and the information is secured using RSA
Revised Manuscript Received on July 10, 2019. algorithm.
[Link] Shyry, CSE, Sathyabama Institute of Science and
technology,Chennai,Tamil Nadu. India.
Saranya Meka, CSE, Sathyabama Institute of Science and technology
Chennai,Tamil Nadu. India.
Mahitha Moganthi, CSE, Sathyabama Institute of Science and
technology ,Chennai,Tamil Nadu, India.

Published By:
Retrieval Number: B11210782S319/19©BEIESP Blue Eyes Intelligence Engineering
DOI : 10.35940/ijrte.B1121.0782S319 658 & Sciences Publication
 DIGITAL IMAGE FORGERY DETECTION

Shivani et al developed a technique of SIFT algorithm Digital signatures are part of an active approach in which
which is applied to mark the objects in the image. They used some bit patterns are embedded in the digital image to avoid
the SIFT algorithm where a whole image is scanned and image manipulation. Dynamic picture validation utilizes a
from the scanned image objects are marked. The properties realized verification code while picture gaining or sending
of the marked object are accessed and objects which have which is infused into the picture or sent alongside it for
similar properties are classified into a group and other are surveying its legitimacy or honesty at recipients side. It is
into second. To classify the similar type of objects varied from exemplary information validation. Sensitivity
techniques like block-based & Key Point based technique, and robustness are two primary needs of active image
shift key point can be used. authentication. Digital watermarking and digital signatures
Jian et al proposed a novel forgery detection schemes are the main active image authentication approaches.
using copy-move At first, the test picture is splitted into
semantic free fixes of preceding key point extraction. V. PASSIVE APPROACH
Accordingly, a match between patches can identify the Though there are different traditional methods which can
duplicate move locales. The coordinating procedure detect forgery in unique way, there is no exact method that
comprises of two phases. At first, the sets of can treat all these cases. The stream of latent altering
patches(suspicious) contains duplicate move fraud locales identification manages breaking down the crude picture
and generally evaluated a relative change grid. Expectation- dependent on various measurements and semantics of
Maximization-based calculation is intended to refine the picture substance to restrict altering of the picture. Neither
evaluated network and to affirm the presence of duplicate develop is infused in the picture and nor related with it for
move fabrication. Trial results on general society databases security, as like dynamic methodologies and consequently
demonstrate the upstanding execution of the proposed plan this technique is otherwise called crude picture
by means of contrasting it and the cutting edge plans. investigation. The confinement of altering is just relying
Mohammad et al integrated Hidden Markov Model upon picture highlight measurements. Along these lines,
(HMM) and Support Vector Machine(SVM) Classifier for calculations and strategies for location and limitation of
image forgery authentication. They consolidated discrete picture dependent on aloof recognition change contingent on
cosine change (DCT) measurements, LBP highlights with the kind of security develop utilized. In any case,
curvelet insights and Gabor change of the pictures to speak uninvolved altering location is usually goal to the limitation
to a picture in the changed area. To verify the technique of altering on crude picture.
CASIA image dataset is used which contains seven
thousand authentic and tampered images. To perform VI. TYPES OF PASSIVE APPROACH
training and testing they divided the dataset into two halves.
Changed pictures are utilized to prepare the Hidden Markov I. Image splicing
display as HMM can separate probabilistic state data from a Image splicing is the method of making a composite
huge measurable model. image by cropping and combining two or more images
.Unlike image retouching this technique is more aggressive
III. CATEGORIZATION OF IMAGE FORGERY for creating forgery images. A number of sharp
TECHNIQUES transformations such as edges, lines, and corners can be
introduced in a spliced image. Examples include the use of
fake images by several infamous news reporting cases.
Image splicing is used with later post-processing techniques
like smoothing boundaries of fragments or even without
post-processing.
II. Image retouching
Image retouching is adding or removing something from
the image for enhancing features of an image. This
technique is applicable for used for aesthetic, commercial
users and it degrades the image [Link], it is
popular among magazine photo editors, where they try to
make an image more attractive by enhancing some features.
Compared to all available forgery techniques Image
IV. ACTIVE APPROACH Retouching is considered as less harmful. Removing
blemishes on a picture of a model would be a great example
In an active approach, the first phase consists of pre- of image Retouching.
processing technique that is watermark injecting.
Watermarking makes active tampering detection, which III .Copy-Move attack
involves injecting a special pattern into the owner (source) Copy Move copies and paste a specific portion of the
image so that piece of information gets authorized. This image. Since the copied region represents the same
special pattern can be further used to notify the user either
the image is tampered or not. But today large portion of the
imaging gadgets don't contain any watermarking or mark
module.

Published By:
Retrieval Number: B11210782S319/19©BEIESP Blue Eyes Intelligence Engineering
DOI : 10.35940/ijrte.B1121.0782S319 659 & Sciences Publication
 International Journal of Recent Technology and Engineering (IJRTE)
ISSN: 2277-3878, Volume-8, Issue-2S3, July 2019

image,the dynamic range and color remains same. In this IX. ALGORITHM & RESULTS
technique, we add or remove information to cover a part of i. Images are divided into small overlapping or either
the image. Some image or text is masked in original image. non-overlapping blocks
A. SVM Classifier ii. Extract the features using traditional techniques
iii. Extracted feature values corresponding to each
SVM classifier detects forgery in images by calculating blockare stored in matrix.
the hash values for extracted features. In the training phase, iv. Apply sorting techniques to get similar features that
the RSAis used in testing phase to ensure the authenticity of lie in nearness.
person. Image classification, bioinformatics, bio-sequence v. Introduce shift vector concept to find blocks with
analysis, hand-writing recognition, and many more complex similar shifting
real world problems can be attained through SVM. SVM vi. Use the counter vector to count the occurrence same
works in two phases –the training phase and testing phase. shifting blocks and set the counter to 1
Initially, a database is created with a larger number of jpg vii. Similar regions are identified with the help of
or jpeg images and trained in the training phase. These threshold value
images can be of any size and can be captured through a Above steps are used to identify the forged blocks in an
camera or downloaded from the internet. RSA key is fixed image. Block-based techniques can be further divided into
in the database after training images. Authorization is [Link] move forgery detection using Key point based
provided by entering the same RSA key provided during the approach
training phase. These images are further converted from
RGB to grayscale which the noise is removed by applying Key-point based methods can distinguish foreground to
Median filter. background. Unlike block based approach it forms
Image enhancement techniques are applied which include descriptors from specific areas. Using SURF(Speed Up
contrast manipulation & gray level, interpolation and Robust Features),SIFT(Scale in-variant Feature
magnification, pseudo coloring, edge crisping and Transform),GLOH,ORB (Oriented FAST (Features from
sharpening, filtering, noise reduction, etc. Feature Extraction Accelerated Segment Test) etc., detection algorithms on
is done using image analysis, pixel value analysis, and descriptors yields better results. In image, regions with high
texture analysis and hash values are calculated entropy are collected to form feature vectors which follow a
correspondingly. Decision boundaries are defined by SVM series of steps including greyscale conversion, image
classification whereas no algorithms have nice theoritical subdivision. Matching is done on similar feature vectors
approach. forming a cluster into large areas which reports a forgery.
The resultant output is more efficient that block based
VII. B. NAÏVE BAYES CLASSIFIER method. Post-processing may be done such as filtering, edge
detection etc.
Naïve Bayes is used as an image classifier due to its
effectiveness for larger datasets. It assumes that there is an b. Passive based Image forgery detection
independent relation between the presences of a feature in a In the passive approach, the digital signature is not used
class is independent of the presence of any other feature. for the purpose of authentication. Assumptions are made
Bayes rule is applied to the image for the classification. while tapering the image that digital forensics may not leave
Bayes' theorem follows simply from the principles of any visual clue. It alters the underlying statistics of the
conditional probability. image. Passive based Image forgery detection can be done
Posterior probability P(CK|x) is the updated probability of using any of these techniques:
an event occurring after taking into consideration new
information. Before collecting the empirical data, a prior a) Pixel based technique
probability is based on established knowledge. b) Format based technique
c) Camera based technique
A(CK|x) = A(ck)p(x|ck) (1) d) Physically based technique
A(x)
e) Geometric based technique
Where A(CK|x) – Posterior
A(ck)p(x|ck) - prior * likelihood X. CONCLUSION AND FUTURE WORK
A(x) - Evidence We conclude our study on image forgery techniques
.Study of different image forgery techniques has been done
VIII. COPY MOVE FORGERY DETECTION USING elaborately with pros and cons. This paper reviewed various
BLOCK BASED traditional techniques which are been used. Though
In this method, Blocks of images are used for analyzing accuracy of detecting forgery in image using traditional
the forgery. Images are to divided into overlapping or non- methods is attained to certain level, improvement in existing
overlapping blocks than analysing the entire image at the techniques is required for better accuracy. Combination of
same time. machine learning algorithms could be a better option to

Published By:
Retrieval Number: B11210782S319/19©BEIESP Blue Eyes Intelligence Engineering
DOI : 10.35940/ijrte.B1121.0782S319 660 & Sciences Publication
 DIGITAL IMAGE FORGERY DETECTION

yield accuracy. In future, SVM(Support Vector Classifier) Technology, Chennai. Her areas of interest include Network
can be used along with other machine learning techniques security, Overlay networks. She has 14years of experience
for better results. in teaching. She has published more than 45 research papers
in International/National conferences and
REFERENCES International/National Journals. She has actively
1. A.C. Popscu, and [Link], “Statistical Tools for Digital participated as session chair person, member, organizing
Forensics” ,in [Link] 6th international workshop on committee in various International and National
information hiding ,Toronto, Canada 2004. conferences. She has also received best paper Award for her
2. Shivani Thakur, RamanpreetKaur, Dr. Raman presentation in National conference.
Chadha,JasmeetKaur, “A Review Paper on Image
Forgery Detection In Image Processing”, IOSR Journal Saranya Meka is doing her final year year in the
of Computer Engineering (IOSR-JCE) e-ISSN: 2278- department of Computer Science and Engineering,
0661,p-ISSN: 2278-8727, Volume 18, Issue 4, Ver. I
(Jul.-Aug. 2016), PP 86-89 Sathyabama Institute of science and [Link] has
3. DevanshiChauhan, DipaliKasat, SanjeevJain, published 4 papers in international conferences and journals
VilasThakare, “Survery on KeyPoint based Copymove
Forgery Detection Methods on Images”, science direct Mahitha Moganti is doing her final year year in the
volume 85 2016. department of Computer Science and Engineering,
4. Ali Qureshi, M., and M. Deriche. “A review on copy Sathyabama Institute of science and [Link] has
move image forgery detection techniques.” IEEE, 2014.. published 4 papers in international conferences and journals.
5. Qazi, Tanzeela. “Survey on blind image forgery
detection.”IET, 2013.
6. M. Qiao, A. Sung, Q. Liu and B. Ribeiro, ”A novel
approach for detection of copy-move forgery,” Fifth
International Conference on ADVCOMP (Advanced
Engineering Computing and Applications in Sciences,
2011.
7. Gajanan K. Birajdar, Vijay H. Mankar, “Digital image
forgery detection using passive techniques: A survey,”
Digital Investigation, vol. 10, no.3, 2013, pp. 226-245.
8. GagandeepKaur, Manoj Kumar, “Study Of Various Copy
Move Forgery Attack Detection In Digital Images”,
International Journal Of Research In Computer
Applications And Robotics, Vol.3 Issue 9, Pg.: 30-34
September 2015
9. [Link], AlkaKhade, [Link], “Image Copy
Move Forgery Detection Using Block Representing
Method”, International Journal Of Soft Computing And
Engineering (Ijsce) Issn: 2231-2307, Volume-4, Issue-2,
May 2014.
10. R.C. Gonzalez, R.E. Woods, “Digital Image Processing”,
2nd edition, Addison- Wesley, 2003.
11. [Link], X.-P. Cheng, “Copy-Move Forgery Detection in
Digital Image”, 3rd International Congress on Image and
Signal Processing, IEEE Computer Society,2010, pp.
2419-21
12. BarnaliSarma, Gypsy Nandi, “A Study on Digital Image
Forgery Detection”, International Journal of Advanced
Research in Computer Science and Software
Engineering, Volume 4, Issue 11, November 2014
13. ManpreetKaur, Richa Sharma, “Optimization of Copy-
Move Forgery Detection Technique”, International
Journal of advanced Research in Computer Science and
software Engineering, Volume 3, Issue 4, April 2013
14. J. Fridrich, “Methods for "Methods for Tamper
Detection in Digital Images", Proc. ACMWorkshop on
Multimedia and Security, Orlando, FL, October 30−31,
1999, pp. 19−23.
15. S. Saic, J. Flusser, B. Zitová, and J. Lukáš, “Methods for
Detection of AdditionalManipulations with Digital
Images”, Research Report, Project
RN19992001003"Detection of Deliberate Changes in
Digital Images", ÚTIA AV ČR, Prague, December1999
(partially in Czech).

XI. AUTHORS PROFILE

Dr [Link] Shyry is presently working as Professor in
Faculty of Computing, Sathyabama Institute of Science and

Published By:
Retrieval Number: B11210782S319/19©BEIESP Blue Eyes Intelligence Engineering
DOI : 10.35940/ijrte.B1121.0782S319 661 & Sciences Publication