Cryptography Reference
In-Depth Information
encrypted data. Once such data is decrypted there is no way to keep track of
its subsequent reproduction or retransmission. Over the last decade, digital
watermarking has been used to complement cryptographic processes. Invisible
watermarks can be broadly classified into two types, Robust and Fragile (or
Semi-Fragile) watermarks. Robust watermarks [6] are generally used for copy-
right protection and ownership verification because they are robust to nearly
all kinds of image processing operations. In comparison, fragile or semi-fragile
watermarks [9] are mainly applied to content authentication because they are
fragile to most types of modifications. To fulfill copyright protection and con-
tent authentication simultaneously, the multipurpose watermarking algorithm
based on wavelet transform [11] has been suggested.
In general, the above two issues are separately taken into account. This
chapter presents a simple multipurpose watermarking scheme which is able
to solve these two problems simultaneously. The main idea is to embed o
ine
three watermarks. These are the Copyright, the Denotation, and Features
into each of the images in the database. During the online retrieval, we can
make queries based on the Copyright, the Denotation and the Features, or
the combination of them. In the following sections, we describe the features
which we use. Then we give the proposed O
ine Multipurpose Watermarking
Scheme. We then discuss how to make the query and how to retrieve online.
We next provide the experimental results and conclusions.
11.4.1 Feature Extraction
In general, we should extract as many as possible of the available features.
In our system, because the embedding of watermarks will affect the image
quality, we should select as few as possible of the best representative features.
We use three kinds of features, that is, the Global Invariant Feature based on
the Haar Integral [38], Statistical Moments of the Color Histogram and the
Hu Moments, which are now described in detail.
Global Invariant Feature
Invariant features remain unchanged even when the image is transformed
according to a group of transformations. In [38], an example of global feature
invariant to rotations and translations is considered. Given a gray-scale image
M =M (x) , x =(x
0
,x
1
) , 0≤x
0
<N
0
, 0≤x
1
<N
1
,
(11.22)
and an element g∈G of the group of image translations and rotations. The
transformation can be expressed as follows.
′
(gM)(x)=M (x
) ,
(11.23)
cos φ sin φ
−sin φ cos φ
t
0
t
1
N
0
N
1
′
with
x
=
x +
mod
.
