Professional Documents
Culture Documents
I.
robustness;
fragile-
wateramrk;
INTRODUCTION
Original
Image
Scrambled
Image Iscr
K-times
iterations of
Arnold cat
map
Extract 8-bits
from each
pixel of Iscr
Calculate
XOR of 8 bits
of each pixel
Is XOR(x, y)
= W(x, y)?
Binary
watermark (W)
yes
no
Watermarked
Image
No
change
in the
pixels of
Iscr
Watermarked
Image
Cat maps
K times
Scrambled
Image
Calculate XOR
of 8 bits of
each pixel
Original
watermark
Extracted watermark
(XOR results)
XOR
Cat maps (T-K)
times
Locate modified
areas
(1)
(2)
Fig.3: Periodic phenomenon in Arnolds cat map. (a)-is the original image
[256 256]; (b) & (c) are the scrambled images after K=20 & 160 iterations,
respectively; (d) cameraman image after completing a period (T=K=192).
{
LSBm(scr)(x, y) = LSBscr(x, y) + 1 ; if LSBscr(x, y)=0
LSBm(scr)(x, y) = LSBscr(x, y) - 1 ; if LSBscr(x, y)=1
a
Fig.4 (a) original image; (b) watermark image; (c) watermarked image and (d)
extracted watermark from watermarked image.
B. Watermark extraction
The watermark is extracted as follows, step by step-
III.
Fig.5: (a), (b), (c) & (d)- original images; (e), (f), (g) & (h)- watermarked images; (i), (j), (k) & (l)- tampered images; (m), (n), (o) & (p)- extracted watermark with
right key (Key = 20); (q), (r), (s) & (t)-extracted watermark with wrong key (Key=21); (u), (v), (w) & (x)- detected tampered regions.
Fig.8: the relation between EBR (%) with (a) salt and pepper noise intensity
and (b) cropped image (%).
Fig.6: (a), (b) & (c) are the watermarked images attacked by salt & pepper
noise with intensities 0.02, 0.10 & 0.15 respectively; (d), (e) & (f)
corresponding extracted watermarks.
(5)
Fig.7: (a), (b) & (c) are the watermarked images cropped in different
positions; (d), (e) & (f) corresponding extracted watermarks.
(3)
(4)
TABLE I.
Test
images
Method [14]
Proposed
method
MSE
PSNR
MSE
PSNR
MSE
PSNR
Fig. 4(a)
0.5482
50.7413
1.897
45.35
0.4997
51.1437
Fig. 5(a)
0.5518
50.7129
1.932
45.27
0.5032
51.1134
Fig. 5(b)
0.5501
50.7261
2.255
44.6
0.5014
51.1289
Fig. 5(c)
0.5410
50.7986
1.892
45.36
0.4927
51.2049
Fig. 5(d)
1.025
48.025
3.163
43.13
0.9299
48.4464
[2]
[3]
[4]
[5]
TABLE II.
Attacks
Cropping
Salt &
pepper
noise
Cox IJ, Matthew LM, Jeffrey AB, et al. Digital Watermarking and
Steganography. Second edition, Burlington, MA: Morgan Kaufmann
Publishers (Elsevier); 2007.
Chandra M. B. and Srinivas K. S., Robust Multiple Image
Watermarking Scheme using Discrete Cosine Transform with Multiple
Descriptions, International Journal of Computer Theory and
Engineering, vol. 1, no. 5, pp. 17938201, 2009.
W. H. Lin, Y. R. Wang, S. J. Horng, A wavelet-tree-based
watermarking method using distance vector of binary cluster, Sci.
Direct Expert Systems with Applications, vol. 36, pp. 98699878, 2009.
Ramani K.; Prasad E.V, Varadarajan S.; Subramanyam A,"A Robust
Watermarking Scheme for Information Hiding", Advanced Computing
and Communications, 16th International Conference, pp. 58 64, 14-17
Dec. 2008.
Ayman, M.A. and D.D. Dwight, Applications of the naturalness
preserving transform to image watermarking and data hiding, Sci.
Direct Digital Signal Process., vol. 14, pp. 531-549, 2004.
Xiao-Li, N., L. Ju, S. Jian-De and Q. Jian-Ping, A Novel watermarking
Method with Image Signature, Springer-Verlag Berlin Heidelberg, pp:
293-298, 2006.
Method [8]
Method
[14]
Proposed
method
[6]
EBR (%)
EBR (%)
EBR (%)
[7]
5%
4.69
6.00
2.57
[8]
10%
9.10
11.00
4.95
15%
14.05
15.00
7.68
0.10
3.82
4.00
3.64
0.15
7.69
8.00
7.56
0.20
10.05
12.00
9.84
Intensity
[9]
[10]
[11]
[12]
[13]
[14]
V. CONCLUSION
In this paper, chaotic system based fragile image
watermarking scheme has been proposed. Two dimensional
Arnolds cat map has been used to scramble the image. Three
secret keys are obtained for the proposed method. The
watermark can be extracted from the watermarked image
using the correct keys. By investigating the experimental
results, it can be said that the proposed method gives low
values of MSE and high values of PSNR compared to the
existing methods in [8] and [14] which is required for image
watermarking. The proposed method also gives low value of
EBR compared to the existing methods in [8] and [14]. This
property proves the robustness of the proposed method. The
[15]
[16]