You are on page 1of 20

DUAL IMAGE

WATERMARKING
Presented By-
Sachin Deshmukh.
9970406068
Guided By-
INDEX
Introduction
SVD
Primary Watermarking
Secondary Watermarking
Applications
Advantages & Disadvantages
Conclusion
References
INTRODUCTION
Watermark
What is Digital Watermarking?
Dual Watermarking
SINGULAR VALUE DECOMPOSITION:
(SVD)
The singular value decomposition (SVD) is
a factorization of a real or complex matrix, with
many useful applications in signal processing and
statistics.
It is a one type of transformation.
CONTINUED
T
V S U A =
j j
T
N
N
N
v v v u u u ,..., ,
0 0
0 0 0
0 0
0 0
,..., ,
2 1
2
1
2 1

(
(
(
(

Assume that rank of Ais r.


U and V are N x N orthogonal matrices,
U x U
T
=I, V x V
T
=I
S is an N x N diagonal matrix,

r+1
==
N
=0
PRIMARYWATERMARKING :
LL1 HL1
LH1 HH1
The frequency sub band LL1 is divided into
small blocks bk of size 4*4
EMBEDDINGALGORITHM :
Input : Block bk, watermark w with co-ordinates (i,j)
1. Let k=0
2. Perform SVD on the block bk
3. It will give the three matrices Uk, Dk and Vk.
1 0 0 0 1,2, 3, 4
Dk= 0 2 0 0 are the singular values of bk.
0 0 3 0 If w(i,j)==1, 4 = 2- 3.
0 0 0 4 This will give the watermarked Dk.
4.Perform inverse SVD and reconstruct the all blocks.
It will give the watermarked Wbk.
5. k=k+1, go to step 2.
6. Combine all watermark blocks Wbk and we will get the
watermarked image I
EXTRACTION ALGORITHM :
Input : Watermarked Block Wbk, co-ordinates (i,j)
1. Let k=0
2. Perform SVD on the block Wbk
3. It will give the three matrices UWk, DWk and
VWk.
w1 0 0 0 w1, w2, w3, w4
DWk= 0 w2 0 0 are singular
0 0 w3 0 values.
0 0 0 w4
4.EW(i,j)=1
5. k=k+1, go to step 2 until
all watermark bits are extracted.
SECONDARY WATERMARKING
The watermarked image I is decomposed into
four sub bands.
The global mean g of sub band LL1is
calculated by taking the mean of all pixels in the
sub band LL1.
LL1 HL1
LH1 HH1
GENERATION OF OWNERS SHARE
Input: Low frequency sub band LL1 ,watermark
W and a key K2.
1. Compute the global mean g of sub band LL1.
2. Generate a list of two-dimensional random
number pair (m,n) using key K2.
CONTINUED
3. For each pixel value in LL1at the random
location (m,n) and the global mean g from and
each watermark value W(i,j)at location (i,j)from,
generate the owners block o.
4. Repeat the step 3 until all pixels
are processed.
5. The ownership share O is made
up of blocks of o.
OWNERS SHARE :
Rule Comparison Between
LL1(m,n) and g
Watermark
W(i,j)
Owners block
o
1
2
LL1(m,n) g
LL1(m,n) g
0
1
3
4
LL1(m,n) g
LL1(m,n) g
0
1
GENERATION OF IDENTIFIERS SHARE :
Input: Low frequency sub band LL1 of
controversial image and key K2.
1. Compute the global mean g of sub band LL1
of the image.
2. Generate a list of two-dimensional random
number pair (m,n) using key K2.
CONTINUED
3. For each pixel value in LL1at the random
location (m,n) and the global mean g and each
sub watermark value W(i,j) at location (i,j),
generate the master block m.
4. Repeat the step 3 until all random locations are
being processed.
IDENTIFIERSS SHARE :
Rule Comparison Between LL1(m,n)
and g
Owners block o
1 LL1(m,n) g
2 LL1(m,n) g
APPLICATIONS :
1. Copyright Protection
2. Ownership Assertion
3. ID card security
4. Military Applications
5. Medical Applications
ADVANTAGES :
1. It is the robust watermarking scheme.
2. It is best against many attacks.
3. Low distortion
REFERENCES :
1. www.ijetae.com (ISSN 2250-2459, Volume 2,
Issue 7, July 2012)
2. S. Voloshynovkiy, S. Pereira, T. Pun, J. Eggers
and J. Su, Attacks on digital watermarks, IEEE
communications Magazine, vol.39, no. 9
3. I.J. Cox and M. Miller, Electronic watermarking:
Journal of Applied Signal Processing, vol. 2002,
Issue 2.
4. http://www.petitcolas.net/fabien/watermarking

You might also like