Positif Journal Issn No : 0048-4911

FORENSIC TECHNIQUES FOR IDENTIFYING COPY-MOVE

MANIPULATIONS IN DIGITAL IMAGES
BHASKAR RAO KASIREDDI, Assistant Professor, Department of CSE, Avanthi’s Research &
Technological Academy, Bhogapuram,Vizianagaram, Email- [email protected]

SAILAJA BODDU, Assistant Professor. Department of CSE, Avanthi’s Research & Technological
Academy, Bhogapuram,Vizianagaram, [email protected]

GAYATRI YANDRAPU, Assistant Professor. Department of CSE, Avanthi’s Research &

Technological Academy, Bhogapuram,Vizianagaram, Email- [email protected]

PILLA JAGADAMBA ALEKHYA , Assistant Professor. Department of CSE, Avanthi’s Research &
Technological Academy, Bhogapuram,Vizianagaram, [email protected]

LAVANYA YEDLA, Assistant Professor. Department of CSE, Avanthi’s Research & Technological
Academy, Vizianagaram, [email protected]
ABSTRACT: overall accuracy rate of 84.33% and
The development of picture editing 82.79%, respectively. When applied to
software over the past several years has photos that had been manipulated in a
led to the establishment of a topic of active number of ways, such as rotated,
research in the field of digital image fraud magnified, or object translated, the True
detection. Passive forgery detection, or Positive Rate for forgery detection was
Copy Move Forgery Detection (CMFD), is above 91%.
the focus of this work. Using the copy
I. INTRODUCTION
move approach, it is applied to
photographs that have been altered. The Nowadays, with the widespread
proposed feature extraction method for a availability of image altering programs
CMFD technique that uses 2 Nearest like Adobe Photoshop, tampering with
Neighbor (2NN) with Hierarchical digital images has become simple.
Agglomerative Clustering (HAC) as the With the development of picture
feature matching method is Oriented editing software, it is now possible to
Features from Accelerated Segment Test alter images without causing
and rotated Binary Robust Independent noticeable alterations or deteriorating
Elementary Features (Oriented FAST and their quality. Images are now used as
rotated BRIEF). It is suggested that this historical records and supporting
method be used for CMFD. The suggested evidence in a variety of sectors,
CMFD method was tested on pictures that including forensic investigation, law
were subjected to different geometrical enforcement, journalistic photography,
attacks at different times. The suggested and medical imaging. This is
approach for evaluations, which utilizes concerning. Furthermore, modified
images from the MICC-F600 and photographs have frequently been
MICCF2000 databases, can achieve an published in the press or on social
media. One example is the July 9,
Vol 23, Issue 10, 2023 Page No: 172
Positif Journal Issn No : 0048-4911

2008, broadcast of manipulated images Determining the authenticity of a given

of an Iranian missile launch by Sepah image is one of the main issues in image
press, the Iranian Revolutionary forensics. In situations where photos are
Guard's official media outlet. Fig. 1's utilized as fundamental evidence to sway
manipulated image is intended to judgment, such as in a court of law, this
inflate the nation's military prowess. can be an extremely important task. The
The counterfeit was discovered a day literature has produced a number of
later when the same source published a technology tools to perform such forensic
second image with different content analysis. This study examines the issue of
that was taken from the same determining whether a picture has been
perspective at nearly the same time. forged; specifically, it looks at situations
The scientific community is not where a portion of an image is duplicated
immune to picture manipulation either. and then pasted into another region to
20% of accepted submissions in the make a duplicate or to erase an
Journal of Cell Biology, according to uncomfortable element. The picture patch
Farid et al., contain improper figure usually requires a geometric adjustment to
manipulation. Image tampering and fit the new context. A unique methodology
detection have therefore attracted a lot based on the scale invariant features
of attention since modified images transform (SIFT) is proposed to detect
might be used maliciously to falsify such alterations. This technique enables us
their meaning. to determine whether a copy-move attack
has taken place and, moreover, to retrieve
the geometric transformation that was
II. LITERATURE SURVEY
utilized to carry out the cloning process.
Identifying digital fakes from JPEG ghosts Numerous experimental findings are
It is frequently required to merge many provided to demonstrate the method's
photos while producing a digital fraud, capacity to accurately identify the changed
such as when compositing the head of one region and, additionally, to very reliably
person onto the torso of another. The estimate the geometric transformation
digital composite might have remnants of parameters. The technique also addresses
the original JPEG compression quality if multiple cloning.
these pictures had differing compression Finding Copy-Move Forgeries in Digital
settings at first. We therefore outline a Pictures
method for determining if a portion of an
Image forensic techniques use a variety of
image was originally compressed at a
high-tech mechanisms developed in the
lesser quality than the remainder of the
literature to determine the integrity of
image. This method works with both high-
images. Because of the powerful image
and lowquality and high-resolution photos.
editing tools, images are susceptible to
A SIFT-based forensic technique for various manipulations, making their
transformation recovery and copymove authenticity questionable, especially when
attack detection images have influential power, such as in
news reports, insurance claims, and court

Vol 23, Issue 10, 2023 Page No: 173

Positif Journal Issn No : 0048-4911

cases. In order to make a duplicate or hide IV. PROPOSED SYSTEM

some existing items, a portion of an image
is copied and pasted onto another image. Reducing superfluous information in a
This sort of forgeries is examined in this picture and increasing computational
work. DCT components are used as the efficiency in the subsequent CMFD phases
block representations after images are are the two main goals of image
initially split into overlapping square preprocessing. Image scaling, tampered
blocks in order to identify the copy-move region recognition, and RGB to grayscale
forgery attempt. Given the high conversion are the pre-processing steps in
dimensionality of the feature space, a our work. An approach to CMFD is shown
lower dimensional feature vector in this study, which uses oriented FAST
representation is obtained by using the and rotating BRIEF (ORB) for feature
Gaussian RBF kernel PCA, which also extraction and 2NN with HAC for feature
increases feature matching efficiency. The matching.
suggested approach is evaluated against
the state of the art through extensive
experiments. According to the SYSTEM ARCHITECTURE
experimental results, the suggested method
can successfully identify multiple copy-
move forgeries and accurately identify the
forgery even when the photos are tainted
by noise, blurring, and compression. In
order to improve the trustworthiness of
photographs in evidence-based
applications, the suggested strategy offers
a computationally effective and
trustworthy method of copymove forgery
detection.
V. IMPLEMENTATION

Modules: The author has utilized the

III. EXISTING SYSTEM
following modules in the suggested
A comparison is made with the work of algorithm.
Kaur and Kaur (2016), in which the feature Getting images: This module will be used
extraction and matching methods are ORB to read every image from the dataset.
and SVM, respectively. The MICC-F600 Pictures Getting ready: RGB to greyscale
database's photos are used to assess the picture conversion
performance of the current task. Negative The process of extracting important points
aspects and descriptors will be carried out utilizing
1. A decrease in precision ORB.
Feature Matching: descriptors will be used
to find picture matching using 2NN
(nearest neighbors), and keypoints will be
Vol 23, Issue 10, 2023 Page No: 174
Positif Journal Issn No : 0048-4911

used to plot the descriptors that match. Its

accuracy will rise if there is a lot of
similarity, and false positives will rise if
there is none.

VI. SCREEN SHOTS

Click the "Acquire Images" button to read
To launch the project, double-click the
and preprocess all of the images after the
"run.bat" file to see the screen below.
dataset has loaded in the screen above.

To upload photos and obtain the screen

Images are loaded and preprocessed by
below, click the "Upload MICC Dataset"
turning their color to grey, as shown in the
button in the above screen.
screen above. I'm only showing one image
for illustration purposes. Click "Run
Existing SVM Algorithm" to train SVM
and obtain the output below.

To load the dataset and view the screen

below, choose and upload the "Dataset"
folder in the above page, then click the
"Open" button.
Click the "Run Propose ORB Algorithm"
button to obtain the output below. In the
SVM screen above, we obtained 100%
accuracy and 0% False Positive Rate.

Vol 23, Issue 10, 2023 Page No: 175

Positif Journal Issn No : 0048-4911

In the screens above, we can observe that

the proposed ORB is analyzing each image
and then identifying/classifying those that
are FORGE. The forge portion of the
image is displayed with connecting lines,
where the original image is in the first part
and the forgery image is in the second. The
application will then display all of the
detected FORGE images; you must close
each image as you receive the output until In above screen we got accuracy as 90%
you obtain the proposed algorithm but we got FPR (false positive rate) as 0.1
accuracy, as shown in the screen below. and SVM give it as 0.

VII. CONCLUSION
In the screen above, we obtained 90%
accuracy; however, the false positive rate
(FPR) was 0.1 and the SVM gave it Our
goal in this research was to identify
methods for ensuring that copy-move
forgeries in digital photographs are
detected. Reducing the feature length
dimension and identifying the forged items
in the suspected image were the primary
considerations of this work. In order to
extract features that take into account the
similar things present in the forged image,

Vol 23, Issue 10, 2023 Page No: 176

Positif Journal Issn No : 0048-4911

we have utilized DCT and kernel PCA. [multimedia authentication technologies],”

This method also functions without a IEEE Signal Processing Magazine, vol. 21, no.
digital signature or watermark and doesn't 2, pp. 40–49, 2004.
require any previous information to be [6] H. Farid, “Image forgery detection: a
included in the image. According to the survey,” IEEE Signal Processing
findings, the suggested method not only Magazine, vol. 26, no. 2, pp. 16–25,
successfully identifies numerous copy- 2009.
[7] I. Cox, M. Miller, J. Bloom, J.
move forgeries and accurately pinpoints
Fridrich, and T. Kalker, Digital
the forged regions, but it also exhibits
Watermarking and Steganography,
good resilience to postprocessing
Morgan Kaufmann, Burling-ton,
techniques including compression, AWGN, Mass, USA, 2007.
and Gaussian blurring. Furthermore, the [8] M. A. Qureshi and M. Deriche, “A
suggested technique's detection bibliography of pixel-based blind
performance is comparatively good in image forgery detection techniques,”
terms of average TPR and FPR when Signal Processing: Image
compared to the current standard copy- Communication, vol. 39, pp. 46–74,
move forgery systems [11– 2015.
14]. [9] T. Qazi, K. Hayat, S. U. Khan et al.,
“Survey on blind image forgery
REFERENCES detection,” IET
[1] N. Krawetz, “A pictures worth digital Image Processing, vol. 7, no. 7, pp. 660– 670,
image analysis and forensics,” Black 2013.
Hat Briefings, 2007. [10] T. Mahmood, T. Nawaz, R. Ashraf et
al., “A survey on block based copy
[2] S. Lian and Y. Zhang, “Multimedia
move image forgery detection
forensics for detecting forgeries,” in
techniques,” in Proceedings of the
Handbook of Information and
International Conference on Emerging
Communication Security, pp. 809–
Tech-nologies (ICET ’15), pp. 1–6,
828, Springer, New York, NY, USA,
Peshawar, Pakistan, December 2015.
2010.
[11] J. Fridrich, D. Soukal, and J. Luka´ˇs,
[3] Y. Li, “Image copy-move forgery
“Detection of copy-move forgery in
detection based on polar cosine
digital images,” in Proceedings of
transform and approximate nearest
Digital Forensic Research Workshop,
neighbor searching,” Forensic Science
Cleveland, Ohio, USA, August 2003.
International, vol. 224, no. 1– 3, pp.
[12] S. Bayram, H. T. Sencar, and N.
59–67, 2013.
Memon, “An efficient and robust
[4] H. Farid, “Digital doctoring: how to
method for detecting copy-move
tell the real from the fake,”
forgery,” in Proceedings of the IEEE
Significance, vol. 3, no. 4, pp. 162–
International Conference on Acoustics,
166, 2006.
Speech, and Signal Processing (ICASSP ’09),
[5] B. B. Zhu, M. D. Swanson, and A. H.
pp. 1053–1056, April 2009.
Tewfik, “When seeing isn’t believing
[13] A. C. Popescu and H. Farid,
“Exposing digital forgeries by
detecting duplicated image regions,”

Vol 23, Issue 10, 2023 Page No: 177

Positif Journal Issn No : 0048-4911

Tech. Rep. TR2004-515, Dartmouth

College, Hanover, NH, USA, 2004.
[14] Y. Huang, W. Lu, W. Sun, and D.
Long, “Improved DCT-based
detection of copy-move forgery in
images,” Forensic Science
International, vol. 206, no. 1–3, pp.
178–184, 2011.
[15] G. Li, Q. Wu, D. Tu, and S. Sun, “A
sorted neighborhood approach for
detecting duplicated regions in image
forgeries based on DWT and SVD,” in
Proceedings of IEEE International
Conference on Multimedia and
Expo (ICME ’07), pp. 1750–1753, IEEE,
Beijing, China, 2007.
[16] B. Mahdian and S. Saic, “Detection of
copy-move forgery using a method
based on blur moment invariants,”
Forensic Science
International, vol. 171, no. 2-3, pp. 180–189,
2007.
[17] J. Zhao and J. Guo, “Passive forensics
for copy-move image forgery using a
method based on DCT and SVD,”
Forensic Science
International, vol. 233, no. 1–3, pp. 158–166,
2013.

Vol 23, Issue 10, 2023 Page No: 178