Transcript 下載/瀏覽
Robust and transparent
watermarking scheme for colour
images
Speaker : Po-Hung Lai
Adviser : Chih-Hung Lin
Date : 2009.1.5
1
Article History and Author
Article History
Published in IET Image Processing
Received on 6th April 2008
Revised on 2nd February 2009
Author
K.-C. Liu & C.-H. Chou
2
Outline
Introduction
Estimation
of JND profiles of colour
images
Watermark embedding and
extraction
Simulation results
Conclusions
3
Introduction
Business
4
Introduction
Business
WATERMARK
5
Introduction
圖的正確性
Y?N?
WA
ARK
6
Introduction
texture,
luminance, corner and edge
7
Introduction (visual part)
discrete
cosine transform (DCT)
by Watson’s perceptual model Li et al.
adaptively adjust quantisation step sizes
small number of perceptual works can be
found
Human
Visual System (HVS)
watermark was inserted into the perceptually
important sub-image
8
Introduction (visual part)
Just
noticeable colour difference
(JNCD)
in
the uniform CIELAB colour space
perceptually lossless colour
quantisation watermark transparency
JNCD threshold
9
Introduction (embed part)
Properties
of histograms
Lin
et al. measured the numerous
global features of all pixels and threedimensional feature space
Invisible
watermarking method
cryptography
watermarking
10
Introduction (embed part)
Modify
pixels
modifying the intensities of pixels in the
blue component of the colour image
modifying quantisation indices of colour pixels
Spread
spectrum watermarking
scheme
watermark into a four-fork tree in the
wavelet domain of the Y component
11
Introduction (defect & improvement)
Fixed
step size to quantise
Worse
robustness
Modification
of the quantisation
help
to improve the performance of the
watermarking scheme.
12
Introduction (in this paper)
perceptual
redundancy inherent
in
each wavelet coefficient of the
colour image in a more reliable YCbCr
13
Estimation of JND profiles of
colour images
Tristimulus
values
Perceptually indistinguishable.
14
Estimation of JND profiles of colour
images
Color
space,CIELAB
(Commission International L’E’ clairage)
RGB,YUV,YCbCr
to overcome the non-uniform colour
metric that had been discussed by
MacAdam
15
Estimation of JND profiles of colour
images
JNCD
In CIELAB space any two colours
difference is greater than 2.3
Performance In CIELAB space use JNCD to
of
embed transparent watermark
robustness
Perceptually Modifying quantisation indices
lossless colour embedded transparent
quantisation watermark
16
Estimation of JND profiles of
colour images
adaptive
JNCD (AJNCD)
due to local variations in luminance
magnitude is considered to estimate
the adaptive JNCD (AJNCD) for each
colour pixel
17
Estimation of JND profiles of colour
images
is a weighting function
between AJNCD
is the average background
luminance
is the maximum of the four weighted
average of luminance
18
Estimation of JND profiles of colour
images
19
Estimation of JND profiles of colour
images
20
Estimation of JND profiles of colour
images
Luminance contrast is given by f1
texture masking effect is given by f2
21
Estimation of JND profiles of colour
images
22
Estimation of JND profiles of colour
images
23
Estimation of JND profiles of colour
images
Choose
YCbCr colour space
Provides
large perceptual redundancy
Without degrading the visual quality
Can resist JPEG,JPEG2000,H.26x
compression attacks
24
Estimation of JND profiles of colour
images
is the JND threshold
is the MND threshold
is a lifting factor
minimally noticeable distortion (MND)
25
Estimation of JND profiles of colour
images
26
Watermark embedding and extraction
27
Watermark embedding and extraction
28
Watermark embedding and extraction
29
Watermark embedding and extraction
30
Watermark embedding and extraction
is the energy of the estimated JND
profile
31
Watermark embedding and extraction
The plots of relative distortion capacity
versus band number are obtained
32
Watermark embedding and extraction
33
Watermark embedding and extraction
To guarantee the transparency and
robustness
JND
tuned by the associated JND value
key information
facilitate the portability of key
34
Watermark embedding and extraction
full-band JND profiles
Bitstream image is too bad
JPEG2000 coder
reduced to 1 k bytes
Sub-band JND profile
tuned colour quantisation
JNDs decodedwatermarked coefficients
35
Watermark embedding and extraction
= quantisation index
= quantisation step size
36
Watermark embedding and extraction
= decoded JND threshold
37
Watermark embedding and extraction
, B=bit sequence
= binary value of 1 or 0
is a random value of 1 or 21 for
modulating qZ,i to higher or lower index
38
Watermark embedding and extraction
39
Simulation results
40
Simulation results
41
Simulation results
To verity the robustness
42
Simulation results
43
Simulation results
44
Simulation results
45
Simulation results
Second experiment
Huang et al.
46
Simulation results
Stimark software
47
Simulation results
48
Simulation results
49
Simulation results
50
Simulation results
51
Simulation results
52
Conclusions
By using the perceptual redundancy
Watermark transparency is achieved
Quite robust in the face of various attacks.
53