Graphics Reference
In-Depth Information
Ta b l e 6 .
BER comparison between the proposed method and GDWM [10]
Attacks
Proposed
GDWM
Attacks
Proposed
GDWM
Rotation (0
.
5
◦
)
JPEG (Q=20)
0.0018
0.0154
0.2154
0.3715
−
0
.
5
◦
)
JPEG (Q=30)
0
0.0034
Rotation (
0.2307
0.3785
JPEG (Q=40)
0
0.0013
Average filtering
(3
×
3)
0
-
Gaussian noise
(
˃
=10)
0
0.0146
Average filtering
(5
×
5)
0.0164
-
Gaussian noise
(
˃
=20)
0.1433
0.1309
Gaussian filtering
(3
×
3)
0
0
Salt&Pepper
(
p
=0
.
01)
0.0064
0.0021
Gaussian filtering
(5
×
5)
0.0104
0.0046
Salt&Pepper
(
p
=0
.
02)
0.0080
0.0088
Median filtering (3
×
3)
0.0051
0.0182
Salt&Pepper
(
p
=0
.
04)
0.0199
0.0310
Median filtering (5
×
5)
0.0613
0.1041
5Con lu ion
In this paper, we have proposed an improved RDM watermarking method. Three as-
pects are applied to improve the robustness of our algorithm: 1) We increase the quan-
tization step size by modifying two coefficients instead of only one coefficient in the
basic RDM method. In this way, the quantization step size is increased. 2) Several mod-
ification rules are defined to reduce embedding distortion and to improve robustness.
For example, we modify the coefficients according to their magnitude and the relation-
ship between the original ratio and its watermarked ratio. 3) Significant coefficients are
selected to embed watermark, because they are more robust and can resist various at-
tacks. A wide range of attacks are tested. Experimental results have verified that our
method is not only robust to amplitude scaling attack but also robust to common sig-
nal processing attacks. Experiments have also demonstrated that our method has better
robustness than the basic RDM and two state-of-the-art watermarking methods, though
the capacity of our method is less than that of the basic RDM method. Hence, when
considering a robust watermarking, our method is a better choice.
Acknowledgments.
The work on this paper was supported by Nature Science Foun-
dation of China (Grant No.61303262) and National Key Technology R&D Program
(Grant No.2012BAH04F02).
References
1. Chen, B., Wornell, G.W.: Quantization index modulation: A class of provably good meth-
ods for digital watermarking and information embedding. IEEE Trans. Inf. Theory 47(4),
1423-1443 (2001)
2. Chen, L.H., Lin, J.J.: Mean quantization based image watermarking. Image and Vision Com-
puting. 21(8), 717-727 (2003)
Search WWH ::
Custom Search