Image analysis – Applications
Reexamination Certificate
2000-02-03
2003-10-28
Patel, Jayanti K. (Department: 2721)
Image analysis
Applications
C713S186000, C381S094200
Reexamination Certificate
active
06639997
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an apparatus for and a method of embedding and extracting digital information, and a medium having a program for carrying out the method recorded thereon, more particularly to, for copyright protection, an apparatus for and a method of embedding digital data such as copyright information in an image signal and extracting the same, and a medium having a program for carrying out the method recorded thereon.
2. Description of the Background Art
In recent years, developments in digital technology accelerate digitalization of multimedia data including audio, image and video. The digitalized multimedia data has been getting popular through high-speed transmission in volume. Such multimedia data, however, is easy to duplicate and hence any unauthorized person is free to duplicate copyrighted digital images for secondary utilization.
To get around such problem, an electron (digital) watermark technique is applied. The digital watermarking is a technique for embedding digital information in image data in an insensible form for human being. With such digital watermark technique, a copyright holder can claim that his/her copyright is illegally used by extracting embedded information as a proof.
The conventional digital watermark technique includes a method disclosed in Japanese Patent Laying-Open No.9-191394 (hereinafter, referred to as first document). The first document proposes a method of embedding, according to normal or random distribution, an embedding value in a key component for image quality after spectrum decomposition is done on data. More specifically, in the first document, embedment is done with an equation (1) and extraction with an equation (2), where Wi is a pseudo-random number string (embedding value), fi is a frequency coefficient of before-watermarking data, fi′ is a frequency coefficient of after-watermarking data, and &agr; is a scaling parameter.
f
i
′=f
i
+&agr;f
i
W
i
(1)
W
i
=
f
i
′
-
f
i
α
f
i
(
2
)
Although the method advantageously renders the embedding value difficult to eliminate, it does not store positional information on the frequency coefficient into which the embedding value is inserted. To extract the embedding value, the method accordingly requires before-embedding original data.
Differently, Japanese Patent Laying-Open No.10-308867 (hereinafter, second document) discloses a method which does not require original data. In the second document, embedment is done with an equation (3) and extraction with (4), where avg(fi) is a partial average of the frequency coefficient fi of before-watermarking data.
f
i
′=f
i
+&agr;avg
(
f
i
)
W
i
(3)
f
i
′
avg
(
f
i
′
)
=
f
i
+
α
avg
(
f
i
)
W
i
avg
(
f
i
′
)
=
f
i
avg
(
f
i
′
)
+
α
W
i
≈
α
W
i
(
∵
avg
(
f
i
)
=
avg
(
f
i
′
)
)
(
4
)
The equation (4) having no fi on the right part thereof indicates that there is no more need for the original data. However, the method is still required to perform DCT (Discrete Cosine Transform) and compute the reciprocal of the partial average of data. In this manner, applying such method to data in volume like image data results in great computation effort.
For betterment, another method is disclosed in Japanese Patent Laying-Open No.10-145757 (hereinafter, third document).
In the third document, embedment is done with an equation (5) and extraction with an equation (6), where |fi| is an absolute value of the frequency coefficient fi of before-watermarking data.
f
i
′=f
i
+&agr;avg
(
f
i
)
W
i
(5)
f
i
′
avg
(
|
f
i
′
|
)
=
f
i
+
α
avg
(
|
f
i
|
)
W
i
avg
(
|
f
i
′
|
)
=
f
i
avg
(
|
f
i
′
|
)
+
α
W
i
≈
α
W
i
(
∵
avg
(
|
f
i
′
|
)
=
avg
(
|
f
i
′
|
)
)
(
6
)
For the purpose of reducing the effort great in the second document, the pseudo-random number string (embedding value) is first divided into several units, and then each unit of numbers is subjected to watermarking. When each unit is 8 by 8 in size, this method can be carried out by utilizing the procedure of MPEG, which is a manner of encoding moving images. In this manner, computational complexity accordingly gets lower, but the method is still required to compute the reciprocal of the partial average of data and great in computation effort.
The method in the third document is relevant to MPEG. Described below are methods of embedding (inserting) and extracting an embedding value disclosed in the second document.
First, the method of embedding is described by referring to FIG.
10
. An embedding signal first goes through an error correction encoder
81
, secondly through a spread spectrum modulator
82
, and lastly through a first spectrum transformer
83
and reaches a spectrum shaping device
84
as a first input. On the other hand, before-watermarking data is provided to a second spectrum transformer
85
. An output of the second spectrum transformer
85
is partially averaged temporally or spatially in a partial averaging device
86
before provided to the spectrum shaping device
84
as a second input, and is also provided to a delaying device
87
. An output of the spectrum shaping device
84
is added to that of the delaying device
87
in an adder
88
. An output of the adder
88
is inversely-transformed in an inverse transformer
89
, and thus data is watermarked.
Next, the method of extracting is described by referring to
FIG. 11. A
spectrum normalization unit
91
receives the watermarked data and then subjects the data to spectrum normalization so as to put the data back to a state before watermarking. Then, the normalized signal is analyzed in correlators
92
A to
92
Z. The correlators
92
A to
92
Z each detects a specific pseudo-random number sequence in the signal, if any, correlates the normalized signal to the pseudo-random number sequence, and then provides an output indicating a degree of the correlation therebetween to a judgment circuit
93
. The judgement circuit
93
sequentially selects an output being maximum as a most-probable current symbol among current symbols received from the correlators. Further, a sequence of the selected maximum likelihood current symbol is provided to the error corrector
94
so as to correct any error in the judgement made in the judgement circuit
93
. In this manner, the embedding signal is extracted as an output of the error corrector
94
.
To eliminate the need for original data, the method in the foregoing results in another need for a partial average to embed an embedding signal and for spectrum shaping by using the partial average. Further, to extract the embedding signal, there is still another need for spectrum normalization to put the embedding signal back to a state before watermarking. In a practical manner, the reciprocal of the partial average is computed. Still further, the method requires a plurality of correlators for extraction, and is required to correct errors at the end. Accordingly, the method bears a problem of great computation effort for extraction.
SUMMARY OF THE INVENTION
Therefore, in view of the above problems, an object of the present invention is, with the help of simple spectrum transformation for embedding and an inner product computed for extraction, to provide an apparatus for and a method of embedding and extracting digital information, in a simplified manner with less effort, with little clue to a third party as to the embedded digital information, and a medium on which a program for carrying out the method is recorded.
The present invention has the following features to achieve the object above.
A first aspect of the present invention is directed to a digital information embedding
Inoue Hisashi
Katsura Takashi
Choobin Barry
Matsushita Electric - Industrial Co., Ltd.
Patel Jayanti K.
Wenderoth , Lind & Ponack, L.L.P.
LandOfFree
Apparatus for and method of embedding and extracting digital... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus for and method of embedding and extracting digital..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus for and method of embedding and extracting digital... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3134642