Image analysis – Applications
Reexamination Certificate
2000-03-23
2004-07-27
Patel, Jayanti K. (Department: 2625)
Image analysis
Applications
C713S176000
Reexamination Certificate
active
06768807
ABSTRACT:
CROSS-REFERENCE TO RELATED APPLICATIONS
This application is based upon and claims the benefit of priority from the prior Japanese Patent Application No. 11-080268, filed Mar. 24, 1999, the entire contents of which are incorporated herein by reference.
BACKGROUND OF THE INVENTION
The present invention relates to a digital watermark embedding device, a digital watermark detection device, a digital information distribution device, and a storage medium.
Recently, a technology to which information, which is called as watermark information, is embedded is used for the digital author data of a moving picture, a static picture, and the voice and music, etc. In this technology, copyright protection and usage control of digital data (prevention of illegal use etc.) is performed by embedding watermark information to become a hard state to perceive visually or audibly, and thereafter detecting watermark information in the data if necessary. Identification information of the author of the user of the digital data, right information of copyrighter, a usage condition of the data, secret information, usage control information, and copy control information, etc. which are required to use them are used, for example, as watermark information.
The digital watermarking method applied, for example, image data includes a method to embed digital watermark information in a spatial (pixel value) domain and a method to embed it in a frequency domain. First, watermark information is embedded in the specimen values in the method of embedding watermark information in the spatial domain. Though, generally, the processing of embedding and detection is light in this method, it is easy to loose watermark information by the noise added by the third party.
On the other hand, the orthogonal transformation is performed to the specimen values, and watermark information is embedded to the data converted into the frequency domain by the conversion in the method of embedding watermark information in the frequency domain. Thereafter, an inverse-orthogonal transformation is performed, and the specimen value is reproduced. In this method, since watermark information is not embedded directly in the specimen value, it is comparatively difficult to remove information by addition of the noise. However, since it is necessary to perform an orthogonal transformation and an inverse-orthogonal transformation, the processing of embedding and detection is heavy.
There is a method to apply the idea of spread spectrum to the method to embed watermark information in the frequency component. The spread spectrum is the communication method which widely distributes information in a large band enough compared with a necessary band for the signal to communicate and transmits it. The tolerance to the noise on the transmission line is excellent in this method.
On the other hand, the method to use the spread spectrum which Prof. Matsui et al. of Defense Academy proposed is called direct sequence spread spectrum method. In this method, it is a spread spectrum according to multiplying the PN (pseudo-random noise) sequence to the pixel value. The orthogonal transformation is performed to the obtained image further, watermark information is embedded in the frequency domain, and an inverse-orthogonal transformation is performed again. And the same PN sequence is multiplied. The detection of watermark information is performed by using the direct sequence spread spectrum of the pixel value by the PN (pseudo-random noise) sequence. The orthogonal transformation is performed to the obtained image, and judges from the value of the frequency component in which watermark information is embedded.
Since watermark information is spreaded and embedded by multiplying the PN sequence from information converted into the frequency domain in the perturbation method, it is possible to embed and control so that the energy of the digital watermark is distributed to from low to intermediate frequency domain as described above. On the other hand, in the direct sequence spread spectrum method, after data manipulation is performed beforehand by the previous execution of the spreading processing so that the energy distribution of the digital watermark in the frequency domain becomes a Gaussian distribution, the orthogonal transform is performed and than watermark information is embedded. Therefore, in a case of the direct sequence spread spectrum method, the energy of the digital watermark embedded by the inverse spreading by the PN sequence after embedding is distributed to the high frequency domain. Therefore, watermark information is easily lost for the D-A-D conversion and the StirMark attack in the conventional direct sequence spread spectrum method compared with the perturbation method. The energy (intensity) of the digital watermark attenuates since the high frequency component is deleted even in DCT conversion in image compressions such as MPEG and JPEG.
Direct sequence spread spectrum method will be explained referring to
FIG. 1A
to
FIG. 2
in more detail.
FIG. 1A
to
FIG. 1C
are figures, which show an appearance to perform the spread spectrum in the direct sequence spread spectrum method, first.
FIG. 2
is a figure which shows an appearance to embed watermark information by the direct sequence spread spectrum method and performing an inverse-spectrum spreading.
First,
FIG. 1A
shows an appearance to multiply the PN sequence to the original picture image and convert (spread spectrum) into a random image. Information is in the pixel domain in this stage. FIG.
1
B and
FIG. 1C
show a case that information of
FIG. 1A
is applied to information on the frequency domain. FIG.
1
B and
FIG. 1C
are shown for convenience sake for the explanation, and conversion into the frequency domain is actually performed after the spread spectrum.
As shown in
FIG. 1B
, the direct sequence spread spectrum method temporarily converts into random information which extends to the high frequency domain pixel information which concentrates on the low frequency domain (especially, DC component) by performing the spread spectrum in the pixel domain.
FIG. 1C
is a figure which changes this to frequency-frequency component value distribution, and it is known to be converted into the Gaussian distribution after spreading. This distribution is according to the given PN sequence.
Though the processing of
FIG. 1A
shows the spread spectrum in
FIG. 2
, in the direct sequence spread spectrum method, in addition, orthogonal transformation is performed and watermark information is embedded. FIG.
2
(
d
) shows a figure in which watermark information is embedded.
That is, the information embedding is performed by changing the data point in the original picture image data converted into information on the frequency domain and changing the frequency component value. In detection and the extraction of data, whether the watermark information is embedded is judged by whether the data point moved at this time exceeds the threshold. In the direct sequence spread spectrum method, watermark information is prevented from modifying, deleting, and leaking by the third party by assuming the embedded position of watermark information in the PN sequence and the frequency domain to be secret information.
Though embedded watermark information is on the frequency domain, and the information image data which is embedded on the frequency domain is returned to image data on the pixel domain by an inverse-orthogonal transformation and an inverse-spread spectrum (FIG.
2
(
d
) to FIG.
2
(
f
)). Here, watermark information embedded in the frequency domain exists on image data as a noise superimposed on the original picture image by the inverse-spreading. Watermark information as this noise is information (noise) to which the frequency is modulated by the PN sequence, and, in general, it becomes the frequency over a high frequency from low frequency (FIG.
2
(
f
)). That is, the energy of watermark information is diffused and distributed to a wide frequency band contrary to the original picture image data, by
Finnegan Henderson Farabow Garrett & Dunner L.L.P.
Patel Jayanti K.
Tabatabai Abolfazl
LandOfFree
Digital watermark embedding device, digital watermark... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Digital watermark embedding device, digital watermark..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Digital watermark embedding device, digital watermark... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3256917