Image analysis – Applications
Reexamination Certificate
2001-07-17
2002-07-23
Johns, Andrew W. (Department: 2621)
Image analysis
Applications
C382S251000
Reexamination Certificate
active
06424726
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a digital image processing system and, in particular, to an image data encoding system for embedding identification data with special information (hereinafter, referred to as electronic watermark data) to a digital image. In addition, the present invention relates to an image inputting apparatus for use in, for example, a personal computer and, in particular, to an image inputting apparatus equipped with an illegal copy prohibiting function.
2. Description of the Related Art
In recent years, the act of illegally copying digital images causes a social problem.
To prevent digital images from being illegally copied, an encryption system has been proposed. In this system, digital image data is encrypted. Only a reproducing system with a valid decryption key can reproduce the encrypted digital image data. However, in such a system, once encrypted data is decrypted, there is no way to prevent the data from being copied any more.
The purpose of a conventional illegal copy prohibiting method for an image inputting apparatus was to prevent the instance of copying image data.
FIG. 9
is a block diagram showing an example of an image inputting apparatus equipped with a conventional illegal copy prohibiting function. An input image is supplied to image pickup means
901
, analog-to-digital converting means
902
, converting means
903
, quantizing means
904
, and variable-length encoding means
905
. After the input image is converted into compressed image data such as an MPEG data stream, the resultant data is supplied to scrambling means
906
. Scrambling means
906
scrambles the input data and outputs compressed and scrambled image data. The compressed and scrambled image data can be reproduced only by an apparatus with a de-scrambling function.
As explained above, in the conventional system, images are scrambled to be prevented from being illegally copied.
In the conventional system, once scrambled images were descrambled, it was impossible to prevent them from being illegally copied.
In addition to such a conventional system, in order to prevent bills and securities from being illegally copied, a method for embedding identification information in pixel components of an image has been proposed in, for example, Japanese Patent Laid-Open Publication Nos. 4-351164, 6-22062, and 6-22119.
In the method for embedding identification information to pixel components of an image, there was the disadvantage that the identification information could be easily forged and removed.
Therefore, a method for embedding electronic watermark data in a digital image has been proposed to prevent digital images from being illegally used and copied.
There are two types of electronic watermark data for digital images, i.e. visible electronic watermark data and invisible electronic watermark data.
The visible electronic watermark data is composed of special characters or symbols so that it can be recognized by visual sensation. Although the visible electronic watermark data causes deterioration of the image quality, the user of the digital image can distinguish it from a forged one, whereby illegal circulation of bills or securities can be prevented.
An example of a method for embedding visible electronic watermark data in an electronic image is disclosed in Japanese Patent Laid-Open Publication No. 8-241403. In this method, when visible electronic watermark data is combined with an original image, only the brightness of pixels corresponding to an opaque portion of the electronic watermark data is varied, not color components. In this method, scaling values which vary the brightness components of the pixels are determined corresponding to color components, random numbers, pixel values of electronic watermark data, or the like.
On the other hand, the invisible electronic watermark data is embedded in an image in such a manner that the electronic watermark data does not affect the image quality. Thus, since the invisible electronic watermark faintly deteriorates the image quality, the deterioration is not perceivable by visual sensation. When special information that identifies a copyright holder of a original image is embedded in the form of the electronic watermark data, even after the image has been illegally copied, the copyright holder of the image can be identified by detecting the electronic watermark data. In addition, in the case that information inhibiting duplication is embedded in a image in the form of electric watermark data, when a relevant reproducing unit such as VTR detects the information, the unit can inform the user that the duplication of the image is inhibited or the unit can prevent duplication of the image by activating duplication inhibiting mechanism.
As one method for embedding invisible electronic watermark data in a digital image, special information representing invisible electronic watermark is embedded in a portion where the information faintly affects the picture quality such as the least significant bits (LSBs) of pixel data. However, in this method, it is easy to erase the electronic watermark data from the image. For example, with a low-pass filter, the information of LSBs of the pixel data can be removed. Additionally, in the image compressing process, redundant data that faintly affects the image quality is removed so as to reduce the data amount and the electric watermark data is embedded in the place where redundant data exists. Thus, when the image compressing process is performed, the electronic watermark data is lost. Consequently, it is difficult to detect the electronic watermark data of an image that has been compressed.
To solve this problem, a method for transforming an image into frequency components and embedding electronic watermark data in the frequency spectrum has been proposed (Nikkei Electronics, p. 13, No. 660, Apr. 22, 1996). In this method, since electronic watermark data is embedded in frequency components, even if an image process such as a compressing process or a filtering process is performed for an image, the electronic watermark data is not lost. In addition, when random numbers that follow a normal distribution are used as electronic watermark data, different pieces of electronic watermark data do not interfere with each other. Thus, it is difficult to destroy the electronic watermark data without largely deteriorating the image.
Referring to
FIG. 10
, the method for embedding electronic watermark data in an image is performed as follows. First of all, a discrete cosine transforming means
1020
transforms an original image into frequency components. In the frequency components, n components are selected as f(
1
), f(
2
), . . . , f(n) according to amplitude order. Electronic watermark data pieces w(
1
), w(
2
), . . . , w(n) are extracted from random data following a normal distribution with means=0 and variance=1. An electronic watermark data embedding means
1030
calculates the following equation for each i:
F
(
i
)=
f
(
i
)+&agr;|
f
(
i
)
|·w
(
i
)|,
where 1≦I≦n and where &agr; is a scaling factor. Finally, image data in which electronic watermark data has been embedded is obtained by transforming F(I) by inverse discrete cosine transform.
The electronic watermark data is detected in the following manner. In this case, it is assumed that the original image and electronic watermark data candidate set {w(i)} (where i=1, 2, . . . n) are known.
With reference to
FIG. 11
, a discrete cosine transforming means
1120
transforms an image in which electronic watermark data has been embedded into frequency components F(
1
), F(
2
) , . . . , F(n). A discrete cosine transforming means
1110
transforms original image data into frequency components f(
1
), f(
2
), . . . , f(n). With f(i) and F(i), electronic watermark data estimated values W(i) are calculated and extracted by the following equation:
W
(
i
)=(
F
(
i
)
−f
(
i
))/
f
(
i
).
Next, an inner product calculating me
Hashimoto Masahiro
Manabe Koji
Nakano Hirotaka
Alavi Amir
Johns Andrew W.
NEC Corporation
LandOfFree
Image data encoding system and image inputting apparatus does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image data encoding system and image inputting apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image data encoding system and image inputting apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2829554