1. DIGITAL IMAGE WATERMARKING
ALGORITHM USING HUMAN VISUAL
SYSTEN ANANLYSIS IN DWT
PRESENTED BYPRESENTED BY
V.SUNDHARARAJV.SUNDHARARAJ M.EM.E
ASSISTANT PROFESSOR/ECEASSISTANT PROFESSOR/ECE
PAAVAI COLLEGE OF ENGINEERINGPAAVAI COLLEGE OF ENGINEERING

2. 2
OUTLINE
• Introduction
• DWT (Discrete Wavelet Transformation)
• HVS （ Human Visual System ）
• Proposed Scheme
• Experimental results
• Conclusions

3. 3
Introduction
• The digital watermarking techniques can be
classified into two categories:
– Spatial domain
• Less complex
• Not more robust
– Frequency domain
• Complex
• More robust

4. INTRODUCTION
Watermark embedding:
A digital watermark is a piece of information embedded
into a digital image using Human visual system
technique .
WATERMARKING DETECTION:
Information can be recovered from the watermarked
image.

5. EXISTING METHOD
ORIGINAL
IMAGE
SECRET
IMAGE
EMBED
DING
EMBED
DING
WATERMA
RKED
IMAGE
WATERMA
RKED
IMAGE
Embedding in this context means to add the
information directly into the image data in
such a way that it is not easily removed.
Less complex
Not more robust

6. PROPOSED METHOD
Complex
More robust

7. 7
DWT (Discrete Wavelet
Transformation)1/2
LL1 HL1
LH1 HH1
Original image
LL2 HL2
LH2 HH2
DWT
1-level
DWT
2-levels
A B C D A+B C+D A-B C-D
L H
A C
B D
L H
A+B C+D
C-DA-B
LL HL
LH HH

8. 8
DWT (Discrete Wavelet
Transformation)2/2
DWT
2-levels
LH1LH1 HH1HH1
HL1HL1
LL2LL2 HL2HL2
LH2LH2 HH2HH2

9. 9
HVS （ Human Visual
System ）
• The human eye is less sensitive to noise in
– High frequency sub-bands
– Brightness is high or low
– Textured area and more near the edges

10. Watermark embedding algorithm
)(i,j)x(m,nαw(i,j)I(i,j)I' θ
l
θ
l
θ
l +=
Example:
206
50*1.2*0.1200
1112*0.11212 0
2
0
2
0
2
=
+=
+= ),)x(,(w),(I),(I'
Step 1:
Step 2: IDWT

11. Watermark detection algorithm
• Watermark images is recovered following the
expression,
(i,j)αw
(i,j)(i,j)-II'
x'(m,n)
θ
l
θ
l
θ
l=
Example:
50
1.2*0.1
200206
==
-
x'(m,n)
Step 2: IDWT
Step 1:

12. 12
Conclusions
• The proposed scheme is based on
HVS(Human Visual system) characteristics.
• The proposed scheme has better
performance in terms of robustness.

13. CONFERENCE
[1]. V.sundhararaj , ”image watermarking using human visual
system scheme in wavelet domain”, 2011,GOVERNMENT
COLLEGE OF TECHNOLOGY,COIMBATORE.
[2]. V.sundhararaj ,” image watermarking detector using Gauss
hermite expansion in wavelet domain human visual system”,
2011,ANNA UNIVERSITY OF
TECHNOLOGY,COIMBATORE.
[3]. V.sundhararaj ,”watermarking detector using HVS analysis in
wavelet domain human visual system”, 2012,jayalaksmi college
Engg &tech.
[4]. V.sundhararaj ,” image fusion detection in satellite image”,
2013,ncret, Gujrat.

14. REFERENCE
[1]. Yaohui Dai,Chunxian wang, “ Digital watermarking Algorithm
based on wavelet transform”, Control, Automation Systems
Engineering (case), 2011 International Conference on IEEE.
[2] M. M. Rahman, M. O. Ahmad, and M. N. S. Swamy,
“Statistical detector for wavelet-based image watermarking
using modified GH PDF,” in Proc. IEEE Int. Symp. Circuits and
Systems, Seattle, WA, 2008, pp. 712–715.
[3]. M. Mahbubur Rahman, M. Omair Ahmad, AUGUST 2009 “A
New Statistical Detector for DWT-Based Additive Image
Watermarking Using the Gauss Hermite Expansion”IEEE
TRANSACTIONS ON IMAGE PROCESSING VOL. 18, NO.
8, AUGUST 2009.

15. 15
Watermark embedding
algorithm(1/3)
x0
3
x0
0
x0
2
x0
1
Image
Watermark
LL HL
LH HH
DWT
4-levels
DWT
1-levels
Step 1:
Θ=0
Θ=1Θ=2
Θ=3
θ
lI : The sub-band (θ) at resolution level (l) of image.

16. 16
Watermark embedding
algorithm(2/3)
• Find the Weight factors for wavelet- coefficient.
(Barni et al.,2001)
(i,j)Wθ
l
)(p*GT θ
lI
θ
l S=
LL2 HL2
LH2 HH2
0.21 0.1 0.12 1.13
0.25 0.36 0.37 1.38
1.40 1.2 1.3 1.4
1.6 1.7 2.1 2.3
Example:
1.21650%ST0
2 == )*(
0.1, 0.12, 0.21, 0.25, 0.36, 0.37, 1.13, 1.2, 1.3, 1.38,
1.4,1.41, 1.6, 1.7, 2.1, 2.3
Step 2:

17. 17
Watermark embedding
algorithm(3/3)
)(i,j)x(m,nαw(i,j)I(i,j)I' θ
l
θ
l
θ
l +=
Example:
206
50*1.2*0.1200
1112*0.11212 0
2
0
2
0
2
=
+=
+= ),)x(,(w),(I),(I'
Step 3:
Step 4: IDWT

18. 18
Watermark extracting algorithm
• Both the original and the watermark images are
needed.
(i,j)αw
(i,j)(i,j)-II'
x'(m,n)
θ
l
θ
l
θ
l=
Example:
50
1.2*0.1
200206
==
-
x'(m,n)
Step 2: IDWT
Step 1:

19. 19
Experimental results (1/3)
(a) Lena 512*512
(b) Watermark 64*64
(C) Watermarked Lena
PSNR=44.7 dB

20. 20
Experimental results (2/3)
(a) 64 times compressed
watermarked Lena
(b) Extracted
Watermark
(d) Extracted
Watermark
(c) 1.37% remained
watermarked Lena after
cropping

21. 21
Experimental results (3/3)
(b) Extracted
Watermark
(a) Warped watermarked Lena

22. 22
Weight factor (1/4)
63 34 49 10
31 23 14 -13
15 14 3 -12
-9 -7 -14 8
0
0 1
1
0
1
10
LL3 HL3
LH3 HH3
i
j
10(0,1)I,63(0,0)I 0
3
3
3 ==
Example:

23. 23
Weight factor (2/4)
0.10.1*1(3,3) ==Θ
2
),,(),,(),(
),(
2.0
jiljill
jiwl
Ξ⋅Λ⋅Θ
=
θθ














=
=
=
=
⋅






 =
=Θ
3if,10.0
2if,16.0
1if,32.0
0if,00.1
otherwise,1
1if,2
),(
l
l
l
l
l
θ
θ
The human eye is less sensitive to noise in high frequency sub-bands:
Example:
2
(3,0,0)(3,0,0)(3,3)
(0,0)W
0.2
3
3
Ξ⋅Λ⋅Θ
=

24. 24
Weight factor (3/4)
)
j
,
i
(I
L(l,i,j)Λ(l,i,j)
ll 











+=
+=
−− 33
3
3
22256
1
1
1


<



=
otherwise,
50)if,1 .L(l,i,j
L(l,i,j)
-L(l,i,j)
L'(l,i,j)
The eye is less sensitive to noise in the those areas of the image
where brightness is high or low.
Example:
1.75
0.751
0.251
256
64
1
2
0
2
0
256
1
1
0031003
00
3
3
=
+=
+=
+=












+=
+=
),(I
),,L(),,Λ(

25. 25
Weight factor (4/4)
10
1033
3
3
3
0
2
0
1
0
21
0
22
Var
2216
1
,y
,xll
l
k x y
kklkk
j
x,
i
yI
j
x,
i
yI
)j,i,l(
=
=−−
−
= =θ = =
θ
+














++⋅














++
=Ξ
∑ ∑∑∑
717164
228.6875(56)
228.68758)14-12-37-9-141513-1410(49003
2
2
=
⋅=
⋅++++++=Ξ ),,(
The is less sensitive to noise in highly texture areas but,The is less sensitive to noise in highly texture areas but,
among these, more sensitive near the edges.among these, more sensitive near the edges.
1.3
2
(14.83)1.750.1
2
(717164)1.750.1
(0,0)W
0.2
3
3
=
⋅⋅
=
⋅⋅
=
Example: