Image stegnogrpahy(muqeed)

Group Members
Muqeed Hussain(UW-13-CPS-BS 077)
Kafeel-ur-rehman(UW-13-CPS-BS 077)
Muhammad Bilal(UW-13-CPS-BS 078)
Department of Computer ScienceDepartment of Computer Science
University of WAHUniversity of WAH
Session(2013-2017Session(2013-2017((

What is Steganography?
Steganography is the art and science of
writing hidden messages in such a way
that no one, apart from the sender and
Intended recipient, suspects the
existence of the message.

What is Image Steganography?
Image Steganography is the technique of hiding
the data within the image in such a way that
prevents the unintended user from the detection
of the hidden messages or data.
For Example:
Cover image Data/message Stego image

Applications of Image Steganography
Secure Private Files and Documents.
Hide Passwords and Encryption Keys.
Transport Highly Private Documents
between International Governments.
Transmit message/data without revealing
the existence of available message.

Image Domain
What Images are made up of?:-What Images are made up of?:-
Images are made up of lots of little dots called pixels. Each
pixel is represented as 3 bytes – one for RedRed, one for
GreenGreen and one for BlueBlue.
Each byte is interpreted as an integer numberinteger number, which is how
much of that color is used to make the final color of the
pixel.
248 + 201 + 3 = Orange Color
248 201 3
11111000 11001001 00000011

Image Domain(Cont)
The difference between two colors that differ by one bit in
either one red, green or blue value is impossible detect for
a human eyehuman eye.
So we can change the least significant (last) bit in a byte,
we either add or subtract one or more values from the
value it represents.
This means we can overwrite the last bit in a byte without
affecting the colors it appears to be.

Image Domain
A common approach of hiding data within an image
file is Least Significant Bit (LSB) SubstitutionLeast Significant Bit (LSB) Substitution.
In this method, we can take the binary
representation of the hidden data and overwrite
the LSB of each byte within the cover image.

Least Significant Bit Substitution
Suppose we have the following binary representation for the Cover
Image.
10010101 00001101
10010110 00001111
Suppose we want to "hide" the following 4 bits of data: 10111011,
we get the following,
100101011 000011011
100101100 000011111
Where the each data bits are accommodated in the least significant bits
of each byte of the image.

Least Significant Bit Substitution
Least Significant Bit Substitution results in a very
minor distortion of the image which is very much
negligible for the human eyes.
Stego ImageCover Image

Substitution Levels
4-bits
6-bits
5-bits
7-bits

Pixel Indicator
This method uses the least two significant bits of oneone
of the channelsof the channels to indicate existence of data in the
other two channelsother two channels.

Pixel Indicator
Example,
Initial Pixel Bytes: 10101101 11011010 11100101
Data to be Embedded: 11011101
Channel R: 1010111111
Channel G: 1101100101
Channel B: 1110011111
• SteganographySteganography
• CategoriesCategories
• FrameworkFramework
• Image SteganographyImage Steganography
• ApplicationsApplications
• Image DomainImage Domain
• -- Methods ---- Methods --
• LSBLSB
• PIPI
• SCCSCC
• Triple-ATriple-A
• Max-bitMax-bit
• StatisticsStatistics
• OPAPOPAP
• Inverted PatternInverted Pattern
• MSE basedMSE based
• Entropy basedEntropy based
• PVDPVD
Pixel Indicator Bits
RR GG BB
Indicating Channel
Channels in
which Data is
Embedded

Pixel Indicator
Example,
Initial Pixel Bytes: 10101101 11011010 11100101
Data to be Embedded: 11011101
Channel R: 1010111111
Channel G: 1101100101
Channel B: 1110011111

Stego Color Cycle
The SCC technique uses the
RGB images to hide the data
in different channels.
It keeps cycling the hidden
data between the Red,
Green and Blue channels,
utilizing one channel at a
cycle time.
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

Triple-A
Triple-A technique uses the same principle
of LSB, where the secret is hidden in the
least significant bits of the pixels, with
more randomizationrandomization in selection of the
number of bits used and the color
channels that are used.
• Two Seeds:
• To determine the used channels
• To determine the number of bits used
This randomization is expected to increase
the security of the system.
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

Max-bit
This method measures the intensityintensity of
the pixel and then hides data by
random pixel selectionrandom pixel selection with a goal to
hide maximum data in each pixel.
This method is divided into three
parts:
Encryption
Image Intensity Calculation
Steganography.
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

Max-bit
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD
Original Image

Max-bit
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD
Grayscale Image Intense Pixels*
* All Black colored pixels are considered as Intense pixels.

Triple-A SCC Max-bit
Bits/pixel 3.428 3 6.281
Capacity
Ratio
3.43/24 =>
14.28%
3/24 =>
12.5%
6.28/24 =>
26.1%
Statistics
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

Optimal Pixel adjustment
Procedure (OPAP) reduces the
distortion caused by the LSB
substitution method.
 In OPAP method the pixel value
is adjusted after the hiding of the
secret data.
This done to improve the quality
of the stego image without
disturbing the data hidden.
Optimum Pixel Adjustment Procedure
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

First a few least significant bits are substituted
with the data to be hidden.
Then in the pixel, the bits before the hidden
bits are adjusted suitably if necessary to give
less error.
Let nn LSBs be substituted in each pixel.
Let dd= decimal value of the pixel after the
substitution.
d1d1 = decimal value of last n bits of the pixel.
d2d2 = decimal value of n bits hidden in that
pixel.
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

If(d1~d2)<=(2^n)/2
then no adjustment is made
in that pixel.
Else
If(d1<d2)
d = d – 2^n .
If(d1>d2)
d = d + 2^n .
Where,
d is converted to binary and
written back to pixel
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

Retrieval Process of DataRetrieval Process of Data:
The retrieval follows
the extraction of the
least significant
bits(LSB) as hiding is
done using simple LSB
substitution.
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

This inverted pattern (IP) LSB
substitution approach uses the idea of
processing secret messages prior to
embedding.
In this method each section of secret
images is determined to be inverted or
not inverted before it is embedded.
In addition, the bits which are used to
record the transformation are treated
as secret keys or extra data to be re-
embedded.
Inverted Pattern
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

The embedded string is S, the replaced string is R,
and the embedded bit string to divided to P parts.
Let us consider n-bit LSB substitution to be made.
Then S and R are of n-bits length.
For P part in i = 1 to P
If MSE(Si,Ri) ≤ MSE(S’i,Ri)
Choose Si for embedding
Mark key(i) as logic ‘0’
If MSE(Si,Ri) ≥ MSE( S‘i,Ri)
Choose S‘ i for embedding
End For
Where,
MSE = Mean Squared Error
S is the data to be hidden
S‘ is the data to be hidden in inverted form.
Inverted Pattern
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

Retrieval Process of DataRetrieval Process of Data:
The stego-image and the key file are
required at the retrieval side.
First corresponding numbers of LSB
bits are retrieved from the stego-image.
If the key is ‘0’, then the retrieved bits
are kept as such.
Else if the key is ‘1’, then the bits are
inverted.
The bits retrieved in this manner from
every pixel of the stego-image gives the
data hidden.
Inverted Pattern
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

Relative entropy measures the
information discrepancy between
two different sources with an
optimal threshold obtained by
minimizing relative entropy.
In this method, instead of finding
the mean square error for inverted
pattern approach, the relative
entropy is calculated to decide
whether S or S‘ suites the pixel.
IP Method Using Relative Entropy
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

Divide the cover image into P blocks of
same size, the embedding string is S,
and the replaced string is R.
For P part in i =1 to P
If rel.entropy(Si,Ri) ≤ rel.entropy (S‘i,Ri)
Choose Si for embedding
If rel.entropy (Si,Ri) ≥ rel.entropy (S‘i,Ri)
Choose S‘i for embedding
End For
Where,
S is the data to be hidden
S‘ is the data to be hidden in inverted form.
IP Method Using Relative Entropy
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

 Peak Signal to Noise Ratio (PSNR):-Peak Signal to Noise Ratio (PSNR):-
 The PSNR is calculated using the equation,
 where Imax is the intensity value of each pixel which is
equal to 255 for 8 bit gray scale images.
 Mean Square Error (MSE):-Mean Square Error (MSE):-
 The MSE is calculated using the equation,
 where M and N denote the total number of pixels in
the horizontal and the vertical dimensions of the image
Xi, j represents the pixels in the original image and Yi,
j, represents the pixels of the stego image.
Decision Factors
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

Pixel Value Differencing (PVD) is able to
provide a high quality stego image in
spite of the high capacity of the concealed
information.
That is, the number of insertion bits is
dependent on whether the pixel is an
edge area or smooth area.
In edge area the difference between the
adjacent pixels is more, whereas in
smooth area it is less.
Pixel Value Differencing
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD
• While human perception is less sensitive to subtle
changes in edge areas of a pixel, it is more
sensitive to changes in the smooth areas.
Smooth Area
Edge Area

This method hides the data in the
target pixel by finding the
characteristics of four pixels
surrounding it, indicated in the table
below:
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

Select the maximum and the minimum
values(g) among the three pixel values that
have already finished the embedding process.
Consider upper pixel (g1), left pixel (g2) and
the upper left pixel (g3) in a given target pixel
g(x,y)
Calculate d using following:
d= [max (g1, g2, g3) – min (g1, g2, g3)]
 Using d , we get an idea as to whether the
target pixel is included in an edge area or in
smooth area.
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

Calculation of n: the number
of the insertion bits in a target
pixel Px,y is calculated, using
the following formula:
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

OPAP IP PVD
Size of secret
data
Image Size
25 kb
242 kb
24.5 kb
242 kb
22 kb
242kb
MSE 0.1503 0.2635 12.4715
PSNR 56.71 51.86 37.87
Time (s) 7.65 7.96 8.6
Statistics
• LSBLSB
• PIPI
• SCCSCC
• Max-bitMax-bit
• OPAPOPAP
• PVDPVD

Image stegnogrpahy(muqeed)

More Related Content

What's hot

Viewers also liked

Similar to Image stegnogrpahy(muqeed)

Recently uploaded

Image stegnogrpahy(muqeed)