-
A REVIEW ON MODIFIED ANTI FORENSIC TECHNIQUE FOR REMOVING
DETECTABLE
TRACES FORM DIGITAL IMAGES 1M.GOWTHAM RAJU. 2N.PUSHPALATHA,
M.Tech (DECS) Student, Assistant professor
Department of ECE, AITS Department of ECE, AITS Annamacharya
Institute of Technology and Sciences, Tirupati, India-517520
[email protected] [email protected]
Abstract: The increasing attractiveness and trust on digital
photography has given rise to new acceptability issues in the field
of image forensics. There are many advantages to using digital
images. Digital cameras produce immediate images, allowing the
photographer to outlook the images and immediately decide whether
the photographs are sufficient without the postponement of waiting
for the film and prints to be processed. It does not require
external developing or reproduction. Furthermore, digital images
are easily stored. No conventional "original image" is prepared
here like traditional camera. Therefore when forensic researchers
analyze the images they dont have access to the original image to
compare. Fraud through conventional photograph is relatively
difficult, requiring technical expertise. Whereas significant
features of digital photography is the ease and the decreased cost
in altering the image. Manipulation of digital images is simpler.
With some fundamental software, digitally-recorded image can easily
be edited. The most of the alterations include borrowing, cloning,
removal and switching parts of a digital image. A number of
techniques are available to verify the authenticity of images. But
the fact is that number of image tampering is also increasing. The
forensic researchers need to find new techniques to detect the
tampering. For this purpose they have to find the new anti-forensic
techniques and solutions for them. In this paper a new
anti-forensic technique is considered, which is capable of removing
the evidences of compression and filtering. It is done by adding a
specially designed noise called tailored noise to the image after
processing. This method can be used to cover the history of
processing in addition to that it can be also used to remove the
signature traces of filtering. Keywords: Digital forensic, jpeg
compression, image coefficients, image history, filtering,
Quantization, DCT coefficients.
Introduction Digital images become very popular for transferring
visual information. And there are many advantages using these
images instead of traditional camera film. The digital
cameras produce instant images which can be viewed without delay
of waiting for film processing. it does not require external
development they can be store easily. And there should not be taken
any time delay. Images can be processed in different ways. They are
processed as jpeg images, in some other cases they are processed in
bit mat format. When they are used in bitmap format it does need to
use without any information of past processing. to know about the
past processing information it is desirable to know the artifacts
of image. These techniques are capable of finding the earlier
processing information. Therefore forensic researches need to
examine the authenticity of images to find how much the trust can
be put up on the techniques and this can also be used to find out
the drawback of this techniques. Person with good knowledge in
image processing can do undetectable manipulation. it is also
desirable to find the draw backs of these techniques. For this
purpose research has to develop both forensic and anti-forensic
techniques to understand the weaknesses. Consider the situation
that already tried to remove the artifacts of compression. The
forensic experts can easily find out the existing techniques such
as quantized estimation. It is useful when image processing unit
receives compression details and quantization table used for
processing and compression. Some of the existing techniques like
detection of blocking signature estimation of quantization table
this allow the mismatches and forgeries in jpeg blocks by finding
the evidence of compression. To solve this problem of image
forensic the research has to develop tools that are capable of
fooling the existing methodologies. Even though the existing
methods have advantages some limitations too. The main drawback of
these techniquesis that they do not report for the risk that new
technique may be design and used to conceal the traces
ofmanipulations. As mention earlier it may possible for an image
forger to generate undetectable compression and other image
forgeries. This modified anti-forensic technique approach is
presented which is capable of hiding the traces of earlier
processing including both compression and filtering. This concept
is that adding specially designed noise to the images blocks will
help to hide the proof of tampering.
18
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING
AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR
SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in
-
1. RELATED TO PROJECT WORK: 1.1. ANTI FORENSIC OF DIGITAL IMAGE
COMPRESSION: As society has become increasingly reliant upon
digital image to communicate visual information, a number of
forensic techniques have developed. Among the most successful of
these are techniques that make use of an images compression history
and its associate compression finger prints. Anti-forensic
techniques capable of fooling forensicAlgorithms this paper
represents set of anti-forensic techniques designed to remove
forensically significant indicators of compression of an image. in
this technique first distributes the image transform coefficients
before compression then adding anti-forensic transform coefficients
of compressed image so that distribution matches estimation one.
When we use these frame work of anti-forensic techniques specially
targeted at erased finger prints left by both JPEG and wavelet
based coders. 1.1.1. ANTI-FORENSIC FRAMEWORK: All image compression
techniques are subbing band coders, which are themselves a subset
of transform coders. Transform coders are mathematically applying
to the signals of compressing the transforms coefficients. Sub band
coders are transform coders that decompose the signal in to
different frequency bands. By applying two dimensional invertible
transform, such as DCT to as image as a whole that has been
segmented into a series of disjoint sets. Each quantized transform
coefficient value can be directly related to its corresponding
original transform coefficient value by equation. = < + 1 (1) If
the image was divided into segment during compression, another
compression finger print may arise. Because of the loss
Fig1: anti forensic of digital image compression When the
anti-forensically modify each quantized transform coefficient by
adding specially designed noise,
which we refer to as anti-forensic dither, to it value according
totheequation Z=Y+D The segment length is equal to the length of
the quantization interval the probability that the quantized
coefficient value is qk is given by. ( = ) (,) (2) The
anti-forensic dithers distribution is given by the formula
P (D=d)= (,) (.) ( + < + 1)(3)
1.1.2. JPEG ANTI-FORENSICS: Brief over view of JPEG compression
then present our anti-forensic technique designed to remove
compression finger prints from JPEG compressed image DCT
coefficients. For gray scale image, JPEG compression begins by
segmenting an image into a series of non over lapping 8x8 pixel
blocks then computing the two dimensional DCT of each block.
Dividing each coefficient value by its corresponding entry in
predetermined quantization matrix rounding the resulting value to
the nearest integer. First image transformed from the RGB to the
YCBCrcolorspace. After this can been performed, compression
continues as if each color layer were an independent gray scale
image. 1.1.3. DCT Coefficient Quantization Fingerprint Removal:
Anti-forensic frame work which we outlined in section 2.we begins
by modeling the distribution of coefficients values with in a
particular ac sub band using the Laplace distribution. ( = ) =
x (4)
Using this model and the quantization rule described above the
coefficient values of an ac sub band of DCT coefficients with in a
JPEG compressed image will be distributed according to the discrete
Laplace distribution.
P(Y=y)=1 , ify = 0
sin( 2 )0 if y=kQi,j (5)
19
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING
AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR
SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in
-
Fig2: Histogram of perturbed DCT coefficient values from a DCT
sub band in which all coefficients were quantized to zero during
JPEG compression. Wavelet-Based Compression Overview: Through
several wavelet-based image compression techniques exists such as
SPIHT,EZW,and most popularly JPEG 2000.they all operate in a
similar fashion and leave behind similar compression finger
prints.JPEG 2000 begins compression by first segmenting an image
into fixed sized non over lapping rectangular blocks known as tiles
while other operate on the image as a whole. Two dimensional DWT of
the image or each image tile is computed these sub bands of the
wave let coefficient. Because of these sub bands corresponding to
either high or low frequency DWT coefficients in each spatial
dimension, the four sub bands are referred to using the notation
LL, LH, HL, and HH. Image compression techniques achieve loss
compression through different processes they each introduce DWT
coefficient quantization finger prints into an image Quantization
and dequantization process causes DWT coefficient in image
compression in the multiples of their respective sub bands.
As a result only the n most significant bits of each DWT
coefficients are retained. This is equivalent to applying the
quantization rule. Where X is a DWT coefficient from an
uncompressed imager y is the corresponding DWT coefficient in its
SPIHT compressed counterpart.
Fig3: Top: Histogram of wavelet coefficient from an uncompressed
image. Bottom: wavelet coefficient from same image after SPIHT
compression.
Fig4: Top: peg compressed image using quality factor. Bottom:
Anti forensically modified version of same image. 2. UNDETECTABLE
IMAGE TAMPERING THROUGH JPEG COMPRESSION Number of digital image
forensic techniques have been developed which are capable of
identifying an images origin, tracing its processing history, and
detecting image forgeries. Though these techniques are capable of
identifying standard image manipulation, they do not address the
possibility t be that anti forensic operations may be designed and
used to hide evidence of image tampering .we propose anti-forensic
operation capable of removing blocking artifacts from a previously
JPEG compressed image. We can show that by help of this operation
along another anti-forensic operation we are able to fool
forensic
20
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING
AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR
SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in
-
methods designed to detect evidence of JPEG compression in
decoded images, determine an images origin, detect double JPEG
compression, and identify cut and paste image forgeries. A digital
image forgery has resulted in an environment where the authenticity
of digital images cannot be trusted. Many of these digital forensic
techniques rely on detecting artifacts left in image by JPEG
compression. Because most of the digital cameras make use of
proprietary quantization tables, an image compression history can
be used to help identify the camera used to capture it. These
techniques are quite adept at detecting standard image
manipulation, they do not account for the possibility that
anti-forensic operation designed to hide traces of image
manipulation may applied to an image. Recent work as shown such
operations can be constructed to successfully fool existing image
forensic techniques. Back Ground: When an image is subjected to
JPEG compression, it is first segmented into 8X8 pixel blocks. The
DCT of each block is computed and resulting set of DCT coefficients
are quantized by dividing each coefficient by its corresponding
entry in a quantization table then rounding the result to the
nearest integer. The set of quantized coefficients read into a
single bit stream and lossless encoded. so decompressed begins by
bit stream of quantized DCT coefficients and reforming into a set
of 8X8 pixel blocks. As a result two forensically significant
artifacts are left in an image by JPEG compression. That is DCT
coefficient quantization artifact sand blocking artifacts. Blocking
artifacts are the discontinuities which occur across 8X8 pixel
block boundaries because of JPEGs loss nature antiforensic
technique capable of removing DCT coefficient artifacts from a
previously compressed image.
2.1. ANTI-FORENSIC DEBLOCKING OPERATION
JPEGblocking artifacts must be removed from an image after
anti-forensic dither has been applied to its DCT coefficients.
Number of de blocking algorithms proposed since the introduction of
JPEG compression, these are all suited for anti-forensic purposes.
To be successful it must remove all visual and statistical traces
of block anti-facts. We found that light smoothing an image
followed by adding low-power white Gaussiannoise. Able to remove
statistical traces of JPEG blocking artifacts without causing the
image DCT coefficient distribution to deviate from the Laplace
distribution. in the anti-forensically deblocked image according to
the equation.
A measure of blocking artifacts strength is obtained by
calculating the difference between the histograms of Z and Z values
denoted by H1 and H2 respectively, using the equation.
K=|HI (Z= n) HII (Z= n)|. The values of K lying above a fixed
detection threshold indicate the presence of blocking
artifacts.
Fig5: Histogram of DCT coefficients from an image before
compression (top left), after JPEG compression (top right), and
after addition anti-forensic dither to the coefficients of the JPEG
compressed image.
2.2. IMAGE TAMPERING THROUGH ANTI-FORENSIC:
We show that anti-forensic dither and our proposedanti-forensic
deblocking operation can be used to deceive several existing image
forensic algorithms that rely on detecting JPEG compression
artifacts.
21
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING
AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR
SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in
-
Fig 6:Result of the proposed anti-forensic deblocking algorithm
applied to a typical image after it has been JPEG compression using
a quality factor of 90 (far left), 70(center left), 30(center
right),and 10 (far right) followed by the addition of anti-forensic
dither to its DCT coefficients.
2.3. Hiding Traces of Double JPEG compression: An image forger
may wish to remove evidence of corresponding a previously JPEG
compressed image. Such image forger wishes to alter a previously
compressed image, and then save the altered image as JPEG.Several
methods have been proposed to detect recompression of JPEG
compressed image commonly known as double JPEG compression. 2.4.
Falsifying an Images Origin: In some scenarios, an image forger may
wish to falsify the origin of digital image simply altering the
Mata data tags associated with an images originating device is
insufficient to accomplish this because several origin identifying
features are intrinsically contained with a digital
image.Anti-forensic dither of an images DCT coefficient, then
re-compressing the image using quantization tables associated with
another device. by doing an image in this manner, we are able to
insert the quantization signature associated with a different
camera into an image while preventing the occurrence of double JPEG
compression artifacts that may alert forensic investigators of such
a forgery.
Fig 7: Histogramof (3, 3) DCT coefficients from an image JPEG
compressed once using a quality factor of 85(left),
image after being double JPEG compressed using a quality factor
of 75 followed by 85(center),and the image after being JPEG
compressed using a quality factor of 75,followed by the application
of anti forensic dither, then recompressed using a quality factor
of 85(right).
3. PROPOSED METHOD:
Tothe best knowledge increased in the field of anti-forensics.
Most of the methods of this an forensics is to find out the process
that which the image compression is takes places, such of that
methods involves in like JPEG detection and quantization table
estimation.in this method of anti-forensic the JPEG compression of
an image history also produces the information of camera used to
produce an image.
Although it can be used to discover the forged areas along with
in the picture.in case of image compression this technique is also
developed to use as evidence of image manipulation.so in this anti
forensic technique traces left by compression and other processing
are discussed
4. CONCLUSION:
By the above two existing methods, one of the method of
anti-forensic method of digital image compression it has
increasingly up on digital images to communicate and this method is
considered anti forensics method is fooling forensic algorithms.
This technique is designed to remove forensically significant
indicators of compression of an image. First developing frame work
its design the anti-forensic techniques to remove compression
finger prints from image transform coefficients. This anti forensic
dither to the transform coefficient of compressed image
distribution matches the estimated one. When we use this frame work
it specifically targeted at erasing compression finger prints left
by both JPEG and wavelet based coders. These techniques are capable
of removing forensically detectable traces of image compression
without significantly impacting an images visual quality. The
second method of undetectable image tampering through JPEG
compression anti forensics digital
22
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING
AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR
SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in
-
forensics are developed which are capable of identifying an
images origin. Thesetechniques are capable of identifying standard
image manipulations. This anti forensic technique capable of
removing blocking artifacts from previously JPEG compression
image.in this method we are able to fool forensic methods to
designed to detect evidence of JPEG compression in decoded images,
determine an images origin. When comparing above two existing
methods, the anti-forensic method of removing detectable traces
from digital images has advanced technique increases attractive
ness and more over trust in the digital images it has capable of
removing evidences of compression and filtering of in digital
images history processing.by adding tailored noise in the image
processing we can find out the where the images is tampered and
compressed, weather its fake or original this can be used in the
medical department as well as in the police department cases. This
method is to be used to cover history of processing and it can be
also used to remove the signature traces of filtering.
REFERENCES
[1]M.chen, J.fridrich, M.goljan and Lukas, Determining image
origin and integrity using sensor noise IEEE trans.inf.forensic
security, vol.3, no.1, pp.74-90, march.2008.
[2]A.swaminathan,M.Wu,andK.R>Liu,Digital image forensics via
intrinsic finger prints,IEEEtrans.inf.forensicssecurity, vol.3,
no.1, pp.101-117, mar.2008. [3] M.Kirchner and R.Bohme,Hiding
traces of resampling in digital images,IEEEtrans.inf.forensics
security, vol.3, no.4, pp.582-592, Dec.2008. [4] I. Ascribes, S.
Bayram, N.Memon, M. Ram Kumar, and B. Sankur, A classifier design
for detecting image manipulations, in Proc. IEEE Int.Conf. Image
Process. Oct. 2004, vol. 4, pp. 26452648. [5] M. C. Stamm and K. J.
R. Liu, Forensic detection of image manipulation using statistical
intrinsic fingerprints, IEEE Trans. Inf. ForensicsSecurity, vol. 5,
no. 3, pp. 492506, Sep. 2010. [6] Z. Fan and R. de Queiroz,
Identification of bitmap compression history: JPEG detection and
quantizer estimation, IEEE Trans. ImageProcess. vol. 12, no. 2, pp.
230235, Feb. 2003. [7] M. C. Stamm and K. J. R. Liu, Wavelet-based
image compressionanti-forensics, in Proc. IEEE Int. Conf. Image
Process., Sept. 2010, pp. 17371740.
[8] W. S. Lin, S. K. Tajo, H. V. Zhao, and K. J. Ray Liu,
Digital image source coder forensics via intrinsic fingerprints,
IEEE Trans. InformationForensics and Security, vol. 4, no. 3, pp.
460475, Sept. 2009. [9] H. Farid, Digital image ballistics from
JPEG quantization, Tech. Rep.TR2006-583, Dept. of Computer Science,
Dartmouth College, 2006. [10] A.C. Popes cu and H. Farid,
Statistical tools for digital forensics, in 6th International
Workshop on Information Hiding, Toronto, Canada, 2004. [11] T.
Pevny and J. Fridrich, Detection of double-compression in JPEG
images for applications in steganography, IEEE Trans.
InformationForensics and Security, vol. 3, no. 2, pp. 247258, June
2008. [12] M. Kirchner and R. Bohme, Hiding traces of resampling in
digital images, IEEE Trans. Information Forensics and Security,
vol. 3, no.4, pp. 582592, Dec. 2008.
23
INTERNATIONAL CONFERENCE ON CURRENT INNOVATIONS IN ENGINEERING
AND TECHNOLOGY
INTERNATIONAL ASSOCIATION OF ENGINEERING & TECHNOLOGY FOR
SKILL DEVELOPMENT
ISBN: 378 - 26 - 138420 - 5
www.iaetsd.in