1. Guided by,
Dr. Jyoti Pareek
Department of Computer Science
Rollwala Computer Center
Gujarat University
Ahmedabad
Hushen Savani (24)
Vikas Kantiya (10)
MCA-V
2. What is Steganography?
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Framework
Categories
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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.
5. What is Image Steganography?
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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
6. Applications of Image Steganography
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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.
7. Image Domain
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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 Red, one for Green and one for Blue.
Red
Blue
11111000 11001001 00000011
248
201
3
Each byte is interpreted as an integer
number, which is how much of that color is
number
used to make the final color of the pixel.
248 + 201 + 3 = Orange Color
8. Image Domain
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
The difference between two colors
that differ by one bit in either one red,
green or blue value is impossible
detect for a human eye.
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.
9. Image Domain
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
A common approach of
hiding data within an image file
is Least Significant Bit (LSB)
Substitution.
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.
10. Least Significant Bit Substitution
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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: 1011,
we get the following,
10010101 00001101
10010110 00001111
Where the each data bits are accommodated in
the least significant bits of each byte of the image.
11. Least Significant Bit Substitution
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
Least Significant Bit Substitution
results in a very minor distortion of
the image which is very much
negligible for the human eyes.
Cover Image
Stego Image
15. Stego Color Cycle
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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.
16. Triple-A
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
Triple-A technique uses the same principle
of LSB, where the secret is hidden in the
least significant bits of the pixels, with
more randomization 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.
21. Optimum Pixel Adjustment Procedure
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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.
22. Optimum Pixel Adjustment Procedure
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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 n LSBs be substituted in each pixel.
Let d= decimal value of the pixel after the
substitution.
d1 = decimal value of last n bits of the pixel.
d2 = decimal value of n bits hidden in that
pixel.
23. Optimum Pixel Adjustment Procedure
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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
24. Optimum Pixel Adjustment Procedure
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
Retrieval Process of Data:
Data
The retrieval follows
the extraction of the
least significant
bits(LSB) as hiding is
done using simple LSB
substitution.
25. Inverted Pattern
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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 reembedded.
26. Inverted Pattern
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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
Mark key(i) as logic ‘1’
End For
Where,
MSE = Mean Squared Error
S is the data to be hidden
S‘ is the data to be hidden in inverted form.
27. Inverted Pattern
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
Retrieval Process of Data:
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.
28. IP Method Using Relative Entropy
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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.
29. IP Method Using Relative Entropy
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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
Mark key(i) as logic ‘0’
If rel.entropy (Si,Ri) ≥ rel.entropy (S‘i,Ri)
Choose S‘i for embedding
Mark key(i) as logic ‘1’
End For
Where,
S is the data to be hidden
S‘ is the data to be hidden in inverted form.
30. Decision Factors
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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): 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.
31. Pixel Value Differencing
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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.
32. Pixel Value Differencing
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
Smooth Area
Edge Area
• 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.
33. Pixel Value Differencing
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
This method hides the data in the
target pixel by finding the
characteristics of four pixels
surrounding it, indicated in the table
below:
34. Pixel Value Differencing
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Steganography
Categories
Framework
Image Steganography
Applications
Image Domain
-- Methods -LSB
PI
SCC
Triple-A
Max-bit
•
Statistics
OPAP
Inverted Pattern
•
MSE based
•
Entropy based
PVD
•
Statistics
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.