Professional Documents
Culture Documents
Abstract---Image forgery detection is emerging as one of the hot practice [3, 4]. The main drawback of this approach remains
research topic in the area of image forensics. In this modern digital era that they are to be inserted i.e., water mark or digital signature in
due to availability of advanced technology, powerful computer photo to the images at the time of recording using special equipments.
editing tools and software packages digital images can be easily It also decreases image quality and security and limited to
forged. In the fields such as forensics, medical imaging, industrial
photography and e-commerce authenticating the originality of images controlled environments.
and detecting traces of manipulation without any prior knowledge of The passive approach does not require any prior information
the image content or any embedded information is a challenging task about the image. To detect traces of tampering it will use the
and quite impossible to say images are authentic. As a result, image function, statistics and content of the available image [4].
photographs have almost lost their trustworthiness. In this paper an It is based on the assumption that digital forgeries may leave no
attempt is made to present a survey on classification of Image forgery
visual clues of having been tampered with and they may likely
detection techniques and a state-of-the-art review of the forgery
detection techniques with complete bibliography. disturb the underlying statistical property. Passive approach
determines the location and amount of forgery in the image. The
Keywords---Image Forgery Detection, Image Forensics,
following are the some of the applications.
Copy-Move Forgery, Image Processing Document verification in various fields
Finger print recognition
I. INTRODUCTION Authenticity of evidence in various conditions
Authenticity of information captured from cameras
MAGE tampering is a process of intentional manipulation of
I images to conceal some information of image. Due to
advancement of sophisticated image editing tools and image
Detailed classification of Image forgery detecting techniques is
as shown in figure.1 and is self explanatory.
processing software packages a digital image can be modified or
II. ACTIVE APPROACHES
tampered very easily [1, 26]. Digital images play a important
roles in many fields as a evidence of events of the depicted Active approaches are related to preprocessing concepts.
happenings [29]. If the digital forged images neither are nor There are two approaches namely.
recognized properly then it will lost its authenticity Digital Water marking
automatically. Hence integrity and authenticity validation of Digital signature
digital image has got much attention in the field of image These two methods work only when there is some prior
processing field [3]. So forgery detection has become an information about the image.
important and most popular research field of image processing. A. Digital Water Marking:
Digital image forgery detection is an emerging research field
Digital water marking consists of inserting a digital water
with important implication of ensuring the trustworthiness of
mark in the image at the time of capture and then verifying at the
digital image [13]. Many image forgery detection techniques
recipient for integrity of image. This technique is used to ensure
have been proposed in the field of digital image forgery
and facilitate data authentication, security and copyright
detection. The exhaustive surveys of the existing methodologies
protection of digital media by inserting a digital watermark and
are presented in this paper. From the literature review its found
verifying through certain efficient algorithms. But this method
that the existing techniques can be classified into two main
decreases image quality and security and moreover it is limited
approaches.
to controlled environment [3]. Limitation of this approach is
1. Active approaches (Non- Blind Approaches)
that it needs to be embedded in image at the time of recording by
2. Passive Approaches (Blind Approaches)
an authorized person with specialized equipment. Classification
In active technique prior information about the image is
of digital watermarking is domain specific namely.
indispensable to the process of authentication. It is concerned
1. Spatial domain technique
with data hiding where some code is embedded into the image at
2. Frequency domain technique
the time of generation which would limit their application in
Manuscript received on May 18, 2017, review completed on May 22, 2017
and revised on May 29, 2017.
S. Manjunatha is with the Department of ISE, GAT Bengaluru. E-Mail:
manjunaths@gat.ac.in
Dr. Malini. M. Patil is with the Department of ISE, JSSATE, Bengaluru.
E-Mail: drmalinimpatil@gmail.com
Digital Object Identifier: DIP052017003.
clues that indicate tampering, they may alter the underlying quantization is the source of the lossy compression. In this
statistics of an image. technique, such compression gives rise to certain artifacts which
These passive approaches can be classified mainly into five can be exploited to detect tampering or forgery. There are three
categories. techniques detect forgery in images.
Pixel-based image forgery detection
1) JPEG Quantization:
Format based image forgery detection
Most of cameras encode images in the JPEG format, this
Camera based image forgery detection
image is representing the image as DCT blocks & quantizing the
Physical environment based image forgery detection
resulting coefficient. The manner in which the DCT coefficients
Geometry based image forgery detection
in each block are quantized is determined by a table called
A. Pixel-based Image Forgery Detection: quantization table. The size and quality of image is determined
Pixel based techniques highlight on the pixels that by quantization table and these tables tend to differ between
constituting the image. The various techniques either indirectly camera manufacturers [7, 21, 22]. This difference between the
of directly between pixels as a result of forgery [23]. These tables is used to perform a forensics analysis on the image.
techniques roughly classified into types. 2) Double JPEG:
1. Copy-move (Cloning/ copy paste)
During forgery the image being tampered must be loaded in
2. Resampling
to photo-editing tool and resaved. Since both the images stored
3. Splicing
in the JPEG format, this results in the JPGE image being
4. Statistical
compressed twice. Because of the lossy nature of the JPGE
1) Copy – Move: format this double compression introduces specific artifacts
In this particular type of forgery one region of the image is not visible in singly compressed image. These artifacts can be
concealed by using another region from the same to cover it. As used as evidence of manipulation [15, 18]. However it is to be
the closed region can be of any shape and location, it is feasible noted that detection of such double compressed does not
to search all possible image locations and sizes. There are 2 necessarily imply malicious intent.
most efficient algorithms, Discrete Cosine Transform (DCT) [7,
3) JPEG Blocking:
14, 27] and Principal Component Analysis (PCA). Besides they
are also robust to minor changes in the image due to additive For JPEG image compression DCT transform is the basis. In
noise or lossy compression. this process an image is break down into 8 x 8 blocks and these
blocks are introduces Horizontal and Vertical edges are called
2) Resampling: blocking artifact grid [16]. When an image is manipulated these
In order to create convincing tampering it is necessary to do blocking artifacts may be disturbed. These artifacts can be used
some image processing operations (Resize, Rotate or to detect manipulations and manipulated regions in the images.
Stretch). In this Resampling method an unnatural
C. Camera Based Techniques:
correlation between neighboring pixels can be found to detect
specific manipulations [17, 28]. The Expectation/ In modern world digital camera is the major device to take
Maximization Algorithm is used to solve a problem in images. When picture has taken it involves a series of
Resampling that is to know which pixels are correlated with processing steps on the path from sensor to memory. There are
their neighbor. many techniques which is specifically model artifacts induced
by the different stages of the imaging process. These techniques
3) Splicing: can be used to estimate different camera artifacts. The
In this, digital splicing of two or more images is done into a inconsistencies among these artifacts can be used for images
single composite, this gives rise to certain abnormalies at the forgery detection. Camera response, sensor noise color filter
splicing boundary. When splicing done carefully, the array and chromatic aberration are major techniques.
boundary between two regions can be visually hardly noticeable
1) Chromatic Aberration:
[6, 14].
Chromatic Aberration can be employed in a camera to detect
4) Statistical: the image tampering and source of camera [25]. There is a
This is another method to find authenticity. It examined chance of expansion or contraction of the color channels due to
statistical properties [7, 1, 2, 27] make used of statistical different wave length of light [10]. If any object is inserted in to
moments from a wavelet decomposition of the image and local image it is likely that contraction or expansion pattern will be
co-occurrence characteristics in image bit-planes distributed allowing for the tampered regions to be detected.
The disadvantage is this will work well only for
B. Format based Image Forgery Detection:
non-compressed or non-uniform pacts of the image. It is more
Now a day’s JPEG lossy compression is the most common regions dependent.
image format. This is based on representing the image as DCT
blocks and quantizing the resulting coefficient. This
2) Color Filter Array: surface normal’s. The required 3-D surface normal are
In today modern world digital camera have a single CCD or determined by 3-D model of human eye [14]. These obtained
CMOS sensor in order to capture images. This is accomplished directions can be compared for various peoples in the image.
by color filter array at the top of sensor. Only one sample color 3) Light Environment:
(Red/Green/Blue) is recorded at the pixel level & remaining two
In this method, distinct lighting condition can be present and
colors are estimated from neighboring samples. This process of
it can be created by using multiple lights source. By assumption
estimating these missing samples is called CFA interpolation
that light striking a Lambertain surface can be calculated the
and particular algorithm is adopted and it can be used to
direction of light source may be estimated and checked for
distinguish between cameras. The specific type of interpolation
consistencies across the image.
can be identified from a periodic correlation exists between
subset of pixels in each color channels [19, 24]. Deviation from E. Geometry based Techniques:
the periodic correlation pattern can be used as evidence of These techniques measure the relative positions of camera
global or local tampering. and world of objects. There are two main techniques includes
3) Camera Response: principal point and metric measurements.
The sensors in cameras are tending to be linear. There is an 1) Principal Point:
existing of linear relationships among cameras in terms of The projection of the camera Centre to the image plane is
amount of light measured by different sensor and corresponding known as the principal point. When an object or image is
final pixel value. However in order to enhance in final image, manipulated the principal point also shifted accordingly or
cameras often apply a point wise non linearity [8]. This proportionally [12]. This estimated difference across image can
distinctiveness of response function in the entire image can be be used as evidence for tampering.
used to detect tampering.
2) Metric Measurement:
4) Sensor Noise: In this there are some tools for projective geometry which
When a camera capture digital image during copying of gives a way to rectify of planner surfaces and provides the
image to computer it will undergoes a series of processing i.e., ability to make real world measurements from a planner surface
white balancing, quantization, filtering etc. this process [11]. In general the image is captured at different angle it is very
introduces a unique pattern it may contain various sources of difficult to discover certain details of image. Under certain
noise [20]. It is possible to model these operations and detect if assumption we can make metric measurement from a single
an image has undergone any subsequent processing. A camera image with suggested geometric tools. These depend on
sensor also contains various sources of noise [7, 22]. Later this knowledge of polygons, vanishing points and co-planarity of
has being showed to be distinct for specific sensor and it can be circles in order to remove planner distortions that enable metric
used to identify specific camera with which the image was measurements to be made on the plane.
captured.
D. Physics Based Techniques: IV. CONCLUSION
An image can be manipulated by splicing together individual In this modern era with the rapid progress of image
images of any two objects. During this process one of biggest processing technology, the need of digital image forgery
challenge in creating a convincing spliced image is to match the detection is more essential in various fields i.e., Defense,
light source directions of the images being combined. The Medical, legal evidence etc., which will play a vital role in the
differences in lighting can be used as evidence of forgery in an decision making. In this paper an attempt is made to understand
image. There are three techniques for estimating different the existing image forgery detection techniques available in
properties of the lighting environment. literature. It reveals that these techniques address various
aspects of forgery detection that help to detect forgeries in
1) Light Direction 2-D: digital images.
During illumination of light on an object the lighting
direction can be estimated at various points in an image from the ACKNOWLEDGEMENT
2-D surface normals at the obstruct object boundary, it is One of the author acknowledges Global Academy of
assumed that the surface of interest is Lambertain surfaces with Technology and JSSMVP’s JSS Academy of Technical
constant reflective values and a point lights surface at infinite Education for providing the facilities to carry out the research
distance, a set of equations can be solved for the lighting work.
directions and ambient light terms [9]. Inconsistencies in
lightning can and have been used to identified the tampering. REFERENCES
2) Light Direction 3-D: [1]. J.a. Redi, W. Taktak, and J. L. Dugelay.” Digital image forensics: a
booklet for beginners,” multimedia Tool Appl., Vol 51, no.1, pp 133-62,
There is an ambiguity in the light direction in 2-D or in a Jan.2011.
normal image. Because it is usually difficult to determine 3-D
[2]. J. Wang.G.Liu, Z. Zhang. Y. Dainad Z. Wang. “Fast and robust [27]. A. Anoop, "Image forgery and its detection: A survey," IEEE Sponsored
forensics for image region- duplication forgery,” Acta Automatica 2J/d International Conference on Innovations in Information, Embedded
Sinica, Vol.35, no.12, pp 1488-95,Dec 2009. and Communication systems (ICIIECS), 2015.
[3]. V.Tyagi, “Detection of forgery in images stored in digital form,” [28]. A. Kaur, and R. Sharma, "Copy-move forgery detection using DCT and
project report submitted to DRDO, New Delhi, 2010. SIFT," International Journal of Computer Applications 70, no. 7 :30-34
[4]. Henry Farid, “Image Forgery Detection”, IEEE Signal Processing ,2013
Magazine, March 2007. [29]. K. Li and C. Xiao-ping. "Copy-move Forgery Detection in Digital
[5]. S. Bayram, I. Avcibas, B. Sankur, and N. Memon, “Image manipulation Image," 3rd International Congress on Image and Signal Processing
detection,” J. Electron. Imaging, vol. 15, no. 4, p. 41102, 2006. (CISP2010)978-1-4244-6516-3/10/$26.00 © IEEE 2010.
[6]. H. Farid, “Detecting digital forgeries using bi spectral analysis,” AI
Lab, Massachusetts Institute of Technology, Tech. Rep. AIM-1657,
1999. S. Manjunatha is a research scholar at JSSATE
[7]. J. Fridrich, D. Soukal, and J. Lukás, “Detection of copy move forgery in Research Centre, Dept. of CSE, JSSATE, affiliated to
digital images,” in Proc. Digital Forensic Research Workshop, Aug. VTU Belagavi. He completed M.Tech Computer
2003. Science and Engineering from NMAMIT Nitte
[8]. Y.-F. Hsu and S.-F. Chang, “Image splicing detection using camera Mangalore, affiliated to VTU Belagavi, Karnataka.
response function consistency and automatic segmentation,” in Proc. Now he is working as a Associate Professor Dept. Of
Int. Conf. Multimedia and Expo, Beijing, China, 2007. ISE, Global Academy of Technology, Bengaluru.
[9]. M. K. Johnson and H. Farid, “Exposing digital forgeries by detecting E-mail: manjunaths@gat.ac.in
inconsistencies in lighting,” in Proc. ACM Multimedia and Security
Workshop, New York, NY, 2005, pp. 1–10.
[10]. M. K. Johnson and H. Farid, “Exposing digital forgeries through Dr. Malini M. Patil is presently working as Associate
chromatic aberration,” in Proc. ACM Multimedia and Security Professor in the Department of Information Science and
Workshop, Geneva, Switzerland, 2006, pp. 48–55. Engineering at J.S.S. Academy of Technical Education,
[11]. M. K. Johnson and H. Farid, “Metric measurements on a plane from a Bangalore, Karnataka, India. She received her Ph.D.
single image,” Dept. Computer. Sci., Dartmouth College, Tech. Rep. Degree from Bharathiar University in the year 2015. Her
TR2006- 579, 2006. research interests are big data analytics, bioinformatics,
[12]. M. K. Johnson and H. Farid, “Detecting photographic composites of cloud computing, image processing. She has published
people,” in Proc. 6th Int. Workshop on Digital Watermarking, more than 20 research papers in many reputed
Guangzhou, China, 2007. international journals. She is a member of IEEE, IEI, ISTE, CSI. She has
[13]. M. K. Johnson and H. Farid, “Exposing digital forgeries in complex attended and presented papers in many international conferences in India and
lighting environments,” IEEE Trans. Inform. Forensics Security, vol. 3, Abroad. E-mail: drmalinimpatil@gmail.com
no. 2, pp. 450– 461, 2007.
[14]. S. Katzenbeisser and F. A. P. Petitcolas, "Information Techniques for
Steganography and Digital Watermarking". Norwood, MA: Artec House,
2000.
[15]. J. Lukas and J. Fridrich, “Estimation of primary quantization matrix in
double compressed JPEG images,” in Proc. Digital Forensic Research
Workshop, Cleveland, OH, Aug. 2003.
[16]. T.-T. Ng and S.-F. Chang, “A model for image splicing,” in Proc. IEEE
Int. Conf. Image Processing, Singapore, 2004, vol. 2, pp. 1169–1172.
[17]. A.C. Popescu and H. Farid, “Exposing digital forgeries by detecting
duplicated image regions,” Dept. Computer. Sci., Dartmouth College,
Tech. Rep. TR2004-515, 2004.
[18]. A. C. Popescu and H. Farid, “Exposing digital forgeries by detecting
traces of re-sampling,” IEEE Trans. Signal Processing, vol. 53, no. 2, pp.
758–767, 2005.
[19]. A. C. Popescu and H. Farid, “Exposing digital forgeries in color filter
array interpolated images,” IEEE Trans. Signal Processing, vol. 53, no.
10, pp. 3948–3959, 2005.
[20]. A. Swaminathan, M. Wu, and K. J. R. Liu, “Digital image forensics via
intrinsic fingerprints,” IEEE Trans. Inform. Forensics Security, vol. 3, no.
1, pp. 101–117, 2008.
[21]. H. Farid, “Digital ballistics from jpeg quantization: A follow-up study,”
Dept. Comp. Sci., Dartmouth College, Tech. Rep. TR2008-638, 2008.
[22]. Minati Mishra and Munesh Chandra Adahikary, "Digital Image Tamper
Detection Techniques – A Comprehensive Study", international Journal
of Computer Science and Business Informatics, 2013, Vol.2, No.1,
Pp.1-12.
[23]. Pradyumna Deshpande and Prashasti Kanikar, Pixel Based "Digital
Image Forgery Detection Techniques", International Journal of
Engineering Research and Applications, 2012, Vol. 2, Issue 3, Pp. 539-
543.
[24]. Wei Wang et al, "A Survey of Passive Image Tampering Detection,
Springer", 2009, Pp. 308–322.
[25]. M. K. Johnson and H. Farid, “Exposing digital forgeries through
chromatic aberration,” in Proc. ACM Multimedia Security Workshop,
Geneva, Switzerland, 2006, pp. 48–55.
[26]. S. Bayram, I.Avcibas, B. Sankur, and N. Memon, “Image manipulation
detection,” J. Electron. Imaging, vol. 15, no. 4, p. 041102, 2006.