JP2003125189A - Image processing system, information embedding device, information detection device, information embedding method, and information detection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an information embedding device capable of reducing adverse effect onto an image when embedding prescribed information to the image. SOLUTION: The information embedding device 10 is characterized in to comprise a frequency conversion means 12 for converting a frequency of an image; an information embedding means for embedding each bit of the information to the image by using a plurality of conversion coefficients subjected to frequency conversion; and a frequency inverse conversion means for applying frequency inverse conversion to the image to which the information is embedded. When number of the conversion coefficients or number of pixels able to be used for embedding to the image is more than number of bits of the embedded information, one bit of the information is embedded to the image by using a plurality of the conversion coefficients or pixels for an object. Thus, the information is embedded to the image while reducing the adverse effect onto the image quality more than that of a conventional watermark embedding.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing system, an information embedding device, an information detecting device for embedding predetermined information (message) in digital data (hereinafter referred to as an image) such as a frame of a still image or a moving image. The present invention relates to an information embedding method and an information detecting method.

[0002]

2. Description of the Related Art With the development of electronic communication technology in recent years, problems regarding image tampering and copyright protection have become apparent. As one of the solutions to such a problem, an electronic watermark is embedded in an image. Electronic watermarks are roughly classified into the following two types. One is a tampering detection digital watermark for detecting tampering made in an image by easily erasing the information of the digital watermark embedded by tampering with the image. Such a digital watermark is a fragile digital watermark or a semi-fragile digital watermark (semi-fr).
agile digital watermark). The other is
Even if the image is tampered with, the information contained in the digital watermark is not destroyed, and this is a digital watermark for copyright protection to protect the copyright of the image. Such a digital watermark is called a hard-to-break digital watermark (robust digital watermark).

An electronic watermark for detecting tampering with an image (fr)
Regarding embedding of the authentication code of the agile / semi-fragile digital watermark, the wavelet (Wa
A digital watermark embedding method by frequency conversion (orthogonal coordinate conversion) such as vellet conversion is described in the document "This system is
partially described at Ching-Yung Lim and Shif-Fu
Chang, “Semi-Fragile Watermarking for Authenticat
ing JPEG Visual SPIESecurity and Watermarking of M
ultimedia Contents II, EI '00, San Jose, CA, Jan.200
0 "". As disclosed in this document, in the conventional digital watermark technology, as a method of embedding an authentication code of a digital watermark (hereinafter, simply referred to as digital watermark embedding), frequency conversion is performed on one bit of the digital watermark. It was performed using only one conversion coefficient.

[0004]

As disclosed in the above-mentioned document, it is possible to embed a digital watermark by using one frequency-transformed transform coefficient for one bit of the digital watermark. You can embed as much information as possible. However, such a digital watermark embedding method has a great influence on the image quality, and the image has a so-called rough feeling. This not only has an adverse effect on the appearance, but also has the risk of notifying that the digital watermark is embedded in the image. In particular, the digital watermark for falsification detection (fragile / s).
In the case of the emi-fragile digital watermark, there has been a problem that a malicious tamperer is made aware of the existence of the digital watermark and gives room for countermeasures against it.

The present invention has been made in view of the above problems of the conventional digital watermark technology, and an object of the present invention is to reduce adverse effects on an image when embedding predetermined information in the image. It is to provide a new and improved information embedding device and information embedding method that are capable of doing so.

Still another object of the present invention is to provide a new and improved information detecting device capable of detecting predetermined information embedded by an information embedding method capable of reducing adverse effects on an image. And to provide an information detection method.

Still another object of the present invention is to provide a comprehensive image processing system by combining the above information embedding device and image detecting device.

[0008]

In order to solve the above-mentioned problems, according to the present invention, an image processing system (claims 1 to 1)
2), information embedding device (claims 3 to 10), information detecting device (claims 11 to 12), information embedding method (claims 13 to 20), information detecting method (claims 21 to 2).
2) is provided.

According to a first aspect of the present invention, an information embedding device for embedding predetermined information consisting of one or more bits in an image and an information detecting device for detecting the predetermined information from the image. Image processing system (10,
20) is provided. The information embedding device (10) includes frequency conversion means (13) for frequency-converting the image to generate a conversion coefficient group, and groups of a number equal to the number of bits of the information with respect to the conversion coefficient group. And an information embedding means (14) for embedding each 1 bit of the information for each group as a target, and the information detecting device (20) extracts from each group of transform coefficients in which the information is embedded. An information detecting means (23) for detecting 1 bit of the information is provided.

Further, according to the present invention, another information embedding device for embedding predetermined information consisting of one or more bits in an image and an information detecting device for detecting the predetermined information from the image are provided. Image processing system (30,4
0) is provided. The information embedding device (3
0) includes an information embedding unit (34) for generating a number of groups equal to the number of bits of the information for the pixels of the image, and embedding each 1 bit of the information for each group. The information detecting device (40) is characterized by comprising an information detecting means (43) for detecting one bit of the information from each group of pixels in which the information is embedded.

According to such an image processing system, on the side of the information embedding device, when the number of conversion coefficients or the number of pixels that can be used for embedding the image is sufficiently larger than the number of bits of the information to be embedded, It is possible to embed a plurality of transform coefficients or pixels for 1 bit. In this way, it is possible to embed information in an image while reducing the adverse effect on the image quality as compared with the conventional watermark embedding. Further, the information detecting device side does not need the information of the place where the information is embedded when the information is detected. When embedding information in an image, it is not necessary to store header information etc. in the image, so it is not necessary to increase the image data amount more than necessary, and it is speculated that the information is embedded in a malicious tamperer. There is an effect that it is hard to be done.

According to a second aspect of the present invention, there is provided an information embedding device for embedding predetermined information consisting of one or more bits in an image. The information embedding device (10) of the present invention comprises a frequency conversion means (13) for frequency-converting the image to generate a conversion coefficient group, and a group of a number equal to the number of bits of the information with respect to the conversion coefficient group. And an information embedding unit (14) for embedding each 1 bit of the information for each group.

Further, according to the present invention, there is provided another information embedding device for embedding predetermined information composed of one or more bits in an image. An information embedding device (30) of the present invention generates information embedding means for the pixels of the image, the groups having a number equal to the number of bits of the information, and embedding each 1 bit of the information for each group. (34) is provided.

According to such an information embedding apparatus, when the number of transform coefficients or the number of pixels that can be used for embedding an image is sufficiently larger than the number of bits of information to be embedded, a plurality of bits are provided for each bit of information. It is possible to embed the transform coefficient or pixel of the target. In this way, it is possible to embed information in an image while reducing the adverse effect on the image quality as compared with the conventional watermark embedding.

According to a third aspect of the present invention, from the image 1
Alternatively, there is provided an information detection device that detects predetermined information composed of two or more bits. Information detecting device of the present invention (2
0) is characterized by including an information detecting means (23) for detecting 1 bit of the information from each group of transform coefficients in which the information is embedded.

Further, according to the present invention, there is provided another information detecting apparatus for detecting predetermined information consisting of one or more bits from an image. Information detecting device (40) of the present invention
Includes an information detecting means (43) for detecting 1 bit of the information from each group of pixels in which the information is embedded.

According to such an information detecting device, the information of the place where the information is embedded is not required at the time of detecting the information. When embedding information in an image, it is not necessary to store header information etc. in the image, so it is not necessary to increase the image data amount more than necessary, and it is speculated that the information is embedded in a malicious tamperer. There is an effect that it is hard to be done.

According to a fourth aspect of the present invention, there is provided an information embedding method for embedding predetermined information consisting of one or more bits in an image. The information embedding method of the present invention comprises a step of frequency-converting the image to generate a transform coefficient group, and a group of a number equal to the number of bits of the information is generated for the transform coefficient group, and each group is Embedding each one bit of the information as a target.

Further, according to the present invention, another information embedding method for embedding predetermined information consisting of one or more bits in an image is provided. The information embedding method of the present invention includes the step of generating groups of the number equal to the number of bits of the information for pixels of the image, and embedding each 1 bit of the information for each group. And

According to such an information embedding method, when the number of transform coefficients or the number of pixels that can be used for embedding an image is sufficiently larger than the number of bits of information to be embedded, a plurality of bits are included for each bit of information. It is possible to embed the transform coefficient or pixel of the target. In this way, it is possible to embed information in an image while reducing the adverse effect on the image quality as compared with the conventional watermark embedding.

In the information embedding apparatus and the information embedding method of the present invention, in order to further reduce the influence on the image quality, the change amount of the conversion coefficient value (or pixel value) due to embedding 1 bit of the information is the smallest. A transform coefficient (or pixel) is selected and 1 bit of the information is embedded. The conversion coefficient (or pixel value) having the largest conversion coefficient value (or pixel value) is selected and 1 bit of the information is embedded. That is, the conversion coefficient (or pixel value) that minimizes the relative change in the conversion coefficient value (or pixel value) is selected, and 1 bit of the information is embedded. For each of the plurality of frequency-converted transform coefficient groups (or for each of the plurality of pixel regions), the change amount of the transform coefficient value (or the pixel value) is accumulated, and according to the accumulated value, , A transform coefficient group (or pixel area) in which 1 bit of the information is embedded is determined. Various methods can be adopted, such as. Details will be described in the embodiments of the invention.

According to a fifth aspect of the present invention, 1 from the image
Alternatively, an information detecting method for detecting predetermined information composed of two or more bits is provided. The information detecting method of the present invention is characterized by including a step of detecting one bit of the information from each group of transform coefficients in which the information is embedded.

Further, according to the present invention, there is provided another information detecting method for detecting predetermined information consisting of one or more bits from an image. The information detection method of the present invention is characterized by including a step of detecting 1 bit of the information from each group of pixels in which the information is embedded.

According to such an information detecting method, the information of the place where the information is embedded is not required at the time of detecting the information. When embedding information in an image, it is not necessary to store header information etc. in the image, so it is not necessary to increase the image data amount more than necessary, and it is speculated that the information is embedded in a malicious tamperer. There is an effect that it is hard to be done.

[0025]

BEST MODE FOR CARRYING OUT THE INVENTION Referring to the accompanying drawings,
An image processing system according to the present invention, an image processing system,
Information embedding device, information detecting device, information embedding method,
Also, a preferred embodiment of the information detection method will be described in detail. In the present specification and the drawings, components having substantially the same functional configuration are designated by the same reference numerals, and duplicate description will be omitted.

(First Embodiment) As an embodiment of an image processing system according to the present invention, a digital watermark embedding device for embedding a predetermined digital watermark consisting of one or more bits into an image, and A system including a digital watermark detection device that detects the predetermined information will be described. In the present embodiment, the discrete cosine transform (D
It is characterized by embedding electronic watermark information (hereinafter, simply referred to as information) for a conversion coefficient of an image subjected to frequency conversion (orthogonal coordinate conversion) such as CT conversion) or wavelet conversion.

First, the digital watermark embedding device side will be described. FIG. 1 is an explanatory diagram showing a configuration of a digital watermark embedding device 10 according to the present exemplary embodiment.

(Digital Watermark Embedding Device 10) As shown in FIG. 1, the digital watermark embedding device 10 includes an image input terminal 11, an information input terminal 12, and a frequency conversion means 13.
The digital watermark embedding unit 14, the frequency inverse conversion unit 15, and the image output terminal 16 are provided. The arrows in the figure indicate that images and information are processed in that direction.

(Image Input Terminal 11) From the image input terminal 11, an image to which predetermined information (digital watermark information) is to be embedded is input. A buffer unit for temporarily accumulating and storing the image may be provided until the process of embedding information in the image is performed.

(Information Input Terminal 12) From the information input terminal 12, predetermined information to be embedded in the image input from the image input terminal 11 is input. A buffer unit for temporarily storing the information may be provided until the process of embedding the information in the image is performed.

(Frequency conversion means 13) Frequency conversion means 1
3 performs frequency conversion on the image input from the image input terminal 11.

(Digital Watermark Embedding Unit 14) The digital watermark embedding unit 13 receives the information input terminal 1 for the conversion coefficient of the image frequency-converted by the frequency conversion unit 12.
The predetermined information input from 2 is embedded. The information embedding operation will be described later.

(Frequency Inverse Transform Unit 15) The frequency inverse transform unit 15 performs the frequency inverse transform corresponding to the frequency transform unit 13 on the image in which the predetermined information is embedded by the digital watermark embedding unit 14.

(Image Output Terminal 16) From the image output terminal 16, the image in which the predetermined information obtained by the frequency inverse conversion means 15 is embedded is output as required. A buffer unit for temporarily storing and accumulating the image may be provided until the image is output.

The digital watermark embedding device 10 according to this embodiment is configured as described above. Next, a digital watermark embedding method in the digital watermark embedding device 10 will be described with reference to FIG.

(Operation of Digital Watermark Embedding Device 10) For the digital watermark embedding device 10, the image input terminal 1
An image is input from 1 and information is input from the information input terminal 12. The image is frequency-converted by the frequency conversion means 13.

The digital watermark embedding means 14 targets a plurality of transform coefficients (transform coefficient groups) of the image, which is frequency-converted by the frequency converting means 13, for each 1 bit of the information input from the information input terminal 12. Embed as.
For example, when information consists of 10 bits and 1 bit of information is embedded for 3 transform coefficients, 10 groups of 3 transform coefficients are generated from the transform coefficient group, and 1 bit from the beginning of the information. Are embedded in each group.

FIG. 2 is an explanatory view showing a case where 1 bit of information is embedded for 12 transform coefficients.
In the example shown in FIG. 2, transform coefficient groups 2LH, 2HL, 2H
Four transform coefficients are extracted from each H. The extraction position of the conversion coefficient may be determined according to a certain rule, or the extraction position may be randomized.

In this embodiment, the digital watermark embedding unit 14 embeds one bit of information for a plurality of transform coefficients by using the following calculation.

N pieces (n
≧ 2, the conversion coefficient a of n = 12 in the example of FIG. 2)₁, A_Two，
a_Three・ ・ ・ ・ ・ ・ A_nConversion coefficient sequence {a₁, A_Two，
a_Three・ ・ ・ ・ ・ ・ A_n}, N parameters for embedding
Data {Q₁, Q_Two, Q_Three・ ・ ・ ・ ・ ・ Q_n}, The following
From (Equation 1), the 1-bit sequence {b₁, B_Two，
b _Three, ..., b_n} Is obtained.

[0041]

[Equation 1]

However, Q ₁ , Q ₂ , Q ₃ , ..., Q _n
May all have the same value, or may take different values depending on the frequency and region. Further, "xAND1" is the least significant bit (Least Significant) of x.
Bit: LSB).

The obtained sequence of 1-bit values {b ₁ ,
A 1-bit value is acquired from all elements of b ₂ , b ₃ , ..., B _n }. The 1-bit value is obtained, for example, by the following (Equation 2).

[0044]

[Equation 2]

Here, "XOR" is an exclusive OR operator. That is, b ₁ , b ₂ , b ₃ , ..., b
_{If n} is an even number "1", P = 0, and if odd, P = 1. Here, if the 1-bit value of the information to be embedded is the same as P, the process is terminated without embedding.

When the 1-bit value of the embedded information is different from P, the following processing is performed. 1-bit sequence {b ₁ ,
b _{2, b 3,} · · ·, of the b _n}, if any one bit inversion, if P is inverted, to become the same as the 1-bit value of embedded information, reverses which bits decide.

In the present embodiment, in order to determine the bit to be inverted, a series {D ₁ , D ₂ , D ₃ , ... Of the change amount (error amount) of the conversion coefficient value when each value is inverted is determined. ., D _n }
To use. The change amount of the conversion coefficient value can be calculated, for example, by the following (Equation 3).

[0048]

[Equation 3]

However, i% j indicates the remainder when i is divided by j. Further, Min (k, l) indicates the smaller value of k and l.

This series {D ₁ , D ₂ , D ₃ , ..., D
_n } is divided by the series {G ₁ , G ₂ , G ₃ , ..., G _n }, and the minimum value is taken (D _min / G
The transform coefficient values _{a min} as bit _{b min} corresponding to _{mi n)} is inverted _{a min} 'to change f. The sequences {G ₁ , G ₂ , G ₃ , ..., G _n } may all take the same value (in this case, it is the same as no division in this sequence), or the frequency or position of the transform coefficient. The weight may be changed according to the quantization coefficient.

The procedure for changing a _min includes, for example,
The following method is used. Here, k1 and k2 may use arbitrary values.

[0052]

[Equation 4]

The above process is repeated for all bits of information. The image in which the information is embedded is frequency-inverse-transformed by the frequency inverse-transforming means 15, and then the image output terminal 1
It is output from 6.

The digital watermark embedding device side of the image processing system according to this embodiment has been described above. Next, the digital watermark detection device side will be described.
FIG. 3 shows a digital watermark detection apparatus 20 according to this embodiment.
It is explanatory drawing which shows the structure of.

As shown in FIG. 3, the digital watermark detection device 20 includes an image input terminal 21, a frequency conversion means 22, and
The digital watermark detecting means 23 and the information output terminal 24 are provided. The arrows in the figure indicate that images and information are processed in that direction.

(Image Input Terminal 21) From the image input terminal 21, an image in which predetermined information (digital watermark information) is embedded is input. A buffer unit for temporarily accumulating and storing the image may be provided until the process of detecting information from the image is performed. Here, it is assumed that the predetermined information is embedded for a plurality of conversion coefficients of the frequency-converted image.

(Frequency conversion means 22) Frequency conversion means 2
2 performs frequency conversion on the image input from the image input terminal 21.

(Digital Watermark Detection Unit 23) The digital watermark detection unit 23 detects predetermined information from the conversion coefficient of the image whose frequency is converted by the frequency conversion unit 22. The information detecting operation will be described later.

(Information Output Terminal 24) From the information output terminal 24, the predetermined information obtained by the digital watermark detecting means 23 is output as necessary. A buffer unit for temporarily storing and accumulating the predetermined information may be provided until the predetermined information is output.

The digital watermark detection apparatus 20 according to this embodiment is configured as described above. Next, a digital watermark detection method in the digital watermark detection device 20 will be described with reference to FIG.

(Operation of Digital Watermark Detection Device 20) An image is input to the digital watermark detection device 20 from the image input terminal 21. The image is converted by the frequency conversion means 22.
Frequency converted.

The digital watermark detecting means 23 extracts the information embedded on the embedding side from the image input from the image input terminal 21 and outputs it. For extraction, a 1-bit value of information is extracted from each group of conversion coefficients in which information is embedded, for example, by using the following operation for the conversion coefficient string used for extracting 1-bit information.

Transform coefficient sequence to be embedded {a ₁ , a ₂ ,
_{a 3,} ···, a _n relative}, the parameters for implantation _{{Q 1, Q 2, Q} 3, ···, using _{Q n},} from each transform coefficient, 1-bit sequence _{b 1, b ₂ , b ₃ , ...,
Find b _n }.

[0064]

[Equation 5]

However, Q ₁ , Q ₂ , Q ₃ , ..., Q _n
May all have the same value, or may take different values depending on the frequency and region.

A 1-bit value is acquired from all the elements of the acquired sequence of 1-bit values. The 1-bit value is obtained, for example, by the following calculation formula.

[0067]

[Equation 6]

The above operation is performed for all the information bits, and all the information bits are detected and used as information. Information is output from the information output terminal 24.

As described above, according to the present embodiment, on the electronic watermark embedding device side, the number of transform coefficients or the number of pixels that can be used for embedding an image is more sufficient than the number of bits of information to be embedded. If there are many, it is possible to embed a plurality of transform coefficients or pixels for one bit of information. In this way, it is possible to embed information in an image while reducing the adverse effect on the image quality as compared with the conventional watermark embedding.

Furthermore, the electronic watermark detection device side does not need information on the place where the information is embedded (information about which bit has been changed) at the time of information detection. When embedding information in an image, it is not necessary to store header information etc. in the image, so it is not necessary to increase the image data amount more than necessary, and it is speculated that the information is embedded in a malicious tamperer. There is an effect that it is hard to be done.

(Second Embodiment) As a second embodiment of the present invention, another embodiment of a digital watermark embedding method will be described with reference to FIG. The configuration of the digital watermark embedding device and the configuration and operation of the digital watermark detection device are the same as those in the first embodiment, and therefore, duplicated description will be omitted.

In this embodiment, the digital watermark embedding means 14 embeds one bit of information for a plurality of transform coefficients by using the following calculation.

N pieces (n
≧ 2, the conversion coefficient a of n = 12 in the example of FIG. 2)₁, A_Two，
a_Three・ ・ ・ ・ ・ ・ A_nConversion coefficient sequence {a₁, A_Two，
a_Three・ ・ ・ ・ ・ ・ A_n}, N parameters for embedding
Data {Q₁, Q_Two, Q_Three・ ・ ・ ・ ・ ・ Q_n}, The following
From (Equation 7), the 1-bit sequence {b₁, B_Two，
b _Three, ..., b_n} Is obtained.

[0074]

[Equation 7]

However, Q ₁ , Q ₂ , Q ₃ , ..., Q _n
May all have the same value, or may take different values depending on the frequency and region. Further, "xAND1" is the least significant bit (Least Significant) of x.
Bit: LSB).

The obtained sequence of 1-bit values {b ₁ ,
A 1-bit value is acquired from all elements of b ₂ , b ₃ , ..., B _n }. The 1-bit value is obtained, for example, by the following (Equation 8).

[0077]

[Equation 8]

Here, "XOR" is an exclusive OR operator. That is, b ₁ , b ₂ , b ₃ , ..., b
_{If n} is an even number "1", P = 0, and if odd, P = 1. Here, if the 1-bit value of the information to be embedded is the same as P, the process is terminated without embedding. The calculation up to this point is the same as in the first embodiment.

When the 1-bit value of the information to be embedded is different from P, the following processing is performed. 1-bit sequence {b ₁ ,
b _{2, b 3,} · · ·, of the b _n}, if any one bit inversion, P is reversed, it becomes the same as the 1-bit value of embedded information, it is the source of this sequence transform coefficients _{a 1, a 2, a 3} , ···, the conversion coefficient values _{a n,} by changing those taking the maximum value, is inverted.

The inversion is performed by changing the conversion coefficient value a _max to a _max ′ for the conversion coefficient having the maximum value. For example, the following method is used for the procedure of changing the conversion coefficient value a _max . Here, k1 and k2 may use arbitrary values.

[0081]

[Equation 9]

The above process is repeated for all bits of information. The image in which the information is embedded is frequency-inverse-transformed by the frequency inverse-transforming means 15, and then the image output terminal 1
It is output from 6.

As described above, according to the present embodiment, by embedding in the part where the level of the transform coefficient is large,
The relative deterioration amount can be reduced. In this way, it is possible to embed in a portion where the effect of embedding is inconspicuous, and the image quality of the image after embedding is improved.

(Third Embodiment) As a third embodiment of the present invention, another embodiment of the digital watermark embedding method will be described with reference to FIG. The configuration of the digital watermark embedding device and the configuration and operation of the digital watermark detection device are the same as those in the first embodiment, and therefore, duplicated description will be omitted.

In this embodiment, the digital watermark embedding means 14 embeds one bit of information for a plurality of transform coefficients by using the following calculation.

When the transform coefficient string is inversely transformed, the pixels of the image, the area of an arbitrary M × N size, or the same kind of sub-band areas that form the same pixel (these are collectively referred to as an area). In addition, 1 pixel is
It can be regarded as one area. ), Buffers R for accumulating a change (deterioration) due to embedding are prepared by the number of pixels or the number of regions, and initialized by 0, for example. For example, in the case of DCT such as JPEG, 8
A buffer may be prepared for each block for performing DCT such as × 8. When performing wavelet transform such as JPEG2000, a transform coefficient group {n
LL, (n-1) LH, (n-1) HL, (n-1) H
A buffer may be prepared in units of H, ..., 1LH, 1HL, 1HH}. Also, when emphasizing embedding in an arbitrary area, that portion may be initialized with a smaller value than other portions.

N pieces (n
≧ 2, in the example of FIG. 5, n = 12) of the conversion coefficients a ₁ , a ₂ ,
a _3, ···, transform coefficient sequence _{a 1 made of _{a n, a 2,}
For a ₃ , ..., A _n }, using n parameters for embedding {Q ₁ , Q ₂ , Q ₃ , ..., Q _n }, one bit of 1 bit is calculated by the following (Equation 10). Sequence {b ₁ , b ₂ , b
₃ , ..., _Bn } are calculated.

[0088]

[Equation 10]

However, Q ₁ , Q ₂ , Q ₃ , ..., Q _n
May all have the same value, or may take different values depending on the frequency and region. Further, "xAND1" is the least significant bit (Least Significant) of x.
Bit: LSB).

The obtained sequence of 1-bit values {b ₁ ,
A 1-bit value is acquired from all elements of b ₂ , b ₃ , ..., B _n }. The 1-bit value is obtained, for example, by the following (Equation 11).

[0091]

[Equation 11]

Here, "XOR" is an exclusive OR operation.
It is an arithmetic. That is, b₁, B _Two, B_Three, ..., b
_nIf there is an even number of "1" s, P = 0, and if
Then, P = 1. Here, the 1-bit value of the information to be embedded
If is the same as P, the process ends without embedding. This
The calculation up to this point is the same as in the first embodiment.

When the 1-bit value of the embedded information is different from P, the following processing is performed. 1-bit sequence {b ₁ ,
b _{2, b 3,} · · ·, of the b _n}, if any one bit inversion, if P is inverted, to become the same as the 1-bit value of embedded information, reverses which bits decide.

In the present embodiment, in order to determine the bit to be inverted, a series of variations {D ₁ , D ₂ , D ₃ , ..., D _n of the conversion coefficient value when each value is inverted is determined. } Is used. The change amount of the conversion coefficient value can be calculated, for example, by the following (Equation 12).

[0095]

[Equation 12]

However, i% j indicates the remainder when i is divided by j. Further, Min (k, l) indicates the smaller value of k and l.

Transform coefficients to be embedded {a ₁ , a ₂ , a
₃ , ..., a _n } of the position of the area when inversely transformed,
The values of the buffer R are {R ₁ , R ₂ , R ₃ , ...
···, R _n }.

This sequence {D ₁ , D ₂ , D ₃ , ..., D
_n } is divided by the series {G ₁ , G ₂ , G ₃ , ..., G _n }, and is the conversion coefficient used to calculate the same pixel when the inverse conversion is performed. take the value _{(R min + (D min /} G m in)) transform coefficient value _{a min} as bit _{b min} is reversed corresponding to _{a m}
change to _in '. Sequence {G ₁ , G ₂ , G ₃ , ...,
G _n } may all take the same value (in this case,
(This is the same as not performing division in this sequence), and the weight may be changed according to the frequency and position of the transform coefficient, the quantization coefficient, and the like.

The procedure for changing a _min is, for example,
The following method is used. Here, k1 and k2 may use arbitrary values.

[0100]

[Equation 13]

Further, R _min updated as in the following (Equation 14) is reflected in the buffer R as shown in FIG. That is, the same conversion coefficient group {nLL, (n-1) L
H, (n-1) HL, (n-1) HH, ..., 1L
H, 1HL, 1HH} at a low resolution portion (for example, in the conversion coefficient group nHL, the larger n is, the lower resolution), the errors embedded so far are added. Where p
_min may be a fixed value. It may be determined by frequency, position, quantization coefficient, or the like.

[0102]

[Equation 14]

In the example shown in FIG. 5, the conversion coefficient bit-inverted before is the conversion coefficient group 3LH. Therefore, in order to reflect the error of the conversion coefficient changed by embedding at the same position, the conversion coefficient group 2LH The error at the time of embedding the conversion coefficient group 3LH is added to the error level. Then, the conversion coefficient for inverting the value is determined so that the value with the smallest error is inverted.

Buffer R for all bits of information
The processes after the initialization are performed. The image in which the information is embedded is frequency-reversed by the frequency reconversion means 15 and then output from the image output terminal 16.

As described above, according to the present embodiment, the total amount of change is buffered for each region, and the region with the smaller total amount of change is preferentially embedded to concentrate on a specific region. It is possible to prevent the embedding, and it is possible to further reduce the influence on the image quality.

The preferred embodiments of the image processing system, the information embedding device, the information detecting device, the information embedding method, and the information detecting method according to the present invention have been described above with reference to the accompanying drawings. It is not limited to the example. It is obvious to those skilled in the art that various changes or modifications can be conceived within the scope of the technical idea described in the claims, and naturally, these are also within the technical scope of the present invention. It is understood that it belongs.

For example, in the above embodiment, the case where the image is frequency-converted and the predetermined information is embedded in the conversion coefficient has been described, but the present invention is not limited to this, and the pixel of the image is targeted. Alternatively, predetermined information may be embedded. In this case, the digital watermark embedding device can have the configuration shown in FIG. Further, the digital watermark detection device can be configured as shown in FIG.

As shown in FIG. 6, the digital watermark embedding device 30 inputs an image input terminal 31 to which an image to be embedded with predetermined information (digital watermark information) is input, and predetermined information to be embedded in the image. Information input terminal 122
And a digital watermark embedding unit 34 for embedding predetermined information in the pixels of the image, and an image output terminal 36 for outputting an image in which the predetermined information is embedded.
The operation of the digital watermark embedding unit 34 is substantially the same as the operation of the digital watermark embedding unit 14 described in the above embodiment by replacing the conversion coefficient with a pixel.

Further, as shown in FIG. 6, the electronic watermark detecting device 40 detects the predetermined information from the image input terminal 211 to which the image in which the predetermined information is embedded in the pixel is input and the pixel of the image. The digital watermark detecting means 43 and the information output terminal 44 for outputting the detected predetermined information are provided. The operation of the digital watermark detecting means 44 is substantially the same as the operation of the digital watermark detecting means 24 described in the above embodiment by replacing the conversion coefficient with a pixel.

Further, according to the present invention, a digital watermark for alteration detection (fragile) for one frame of a still image or a moving image is used.
e, semi-fragile digital watermark) or copyright protection digital watermark (robust digital watermark).

Further, the above-mentioned three embodiments can be combined with each other. For example, if it is prioritized that the value becomes large, according to the second embodiment, the first embodiment and the third embodiment should be applied to the high-order tens of percent with a large value. Can be considered. If priority is given to the reduction of the error, the second embodiment and the third embodiment should be applied to the upper several tens% of the errors having a large error according to the first embodiment. And so on.

In addition, in the first embodiment, when there are a plurality of conversion coefficients having the smallest error, the conversion coefficient having the largest conversion coefficient value is selected from the conversion coefficients (second conversion coefficient). (Embodiment) can be applied. Similarly, in the second embodiment, when there are a plurality of conversion coefficients having the maximum conversion coefficient value, among those conversion coefficients,
It is possible to apply the method (first embodiment) of selecting the conversion coefficient that minimizes the error.

[0113]

As described above, according to the present invention,
On the information embedding device side, when the number of transform coefficients or pixels that can be used for embedding an image is sufficiently larger than the number of bits of information to be embedded, a plurality of transform coefficients or pixels for 1 bit of information Can be embedded as a target. In this way, it is possible to embed information in an image while reducing the adverse effect on the image quality as compared with the conventional watermark embedding.

Further, the information detecting device side does not need the information of the place where the information is embedded when the information is detected. When embedding information in an image, it is not necessary to store header information etc. in the image, so it is not necessary to increase the image data amount more than necessary, and it is speculated that the information is embedded in a malicious tamperer. There is an effect that it is hard to be done.

[Brief description of drawings]

FIG. 1 is a block diagram showing an example of a digital watermark embedding device.

FIG. 2 is a block diagram showing an example of a digital watermark detection device.

FIG. 3 is an explanatory diagram showing a digital watermark embedding method according to the first and second embodiments.

FIG. 4 is an explanatory diagram showing a digital watermark detection method according to the first and second embodiments.

FIG. 5 is an explanatory diagram showing a digital watermark embedding method according to a third embodiment.

FIG. 6 is a block diagram showing another example of a digital watermark embedding device.

FIG. 7 is a block diagram showing another example of a digital watermark detection device.

[Explanation of symbols]

10 Digital Watermark Embedding Device 11 Image input terminal 12 Information input terminal 13 Frequency conversion means 14 Digital watermark embedding means 15 Frequency inverse conversion means 16 Image output means 20 Digital Watermark Detection Device 21 Image input terminal 22 Frequency conversion means 23 Digital Watermark Detection Means 24 Information output terminal

   ─────────────────────────────────────────────────── ─── Continued front page    F term (reference) 5B057 CE08 CE09 CG05 CG07 DA06                 5C063 AB03 AC01 CA40 DA07 DA13                       DA20 DB09                 5C076 AA14 BA06

Claims (22)

[Claims] 1. An image processing system comprising an information embedding device for embedding predetermined information consisting of one or more bits in an image and an information detecting device for detecting the predetermined information from the image. The device is
Frequency conversion means for frequency-converting the image to generate a conversion coefficient group, and groups of the conversion coefficient group, the number of which is equal to the number of bits of the information, are generated, and each of the information items is targeted to each group. An information embedding means for embedding 1 bit, wherein the information detecting device sets 1 of the information from each group of transform coefficients in which the information is embedded.
An image processing system comprising an information detecting means for detecting bits. 2. An information processing system comprising an information embedding device for embedding predetermined information consisting of one or more bits in an image and an information detecting device for detecting the predetermined information from the image. The device is
The information detecting device is provided with an information embedding means for generating a number of groups equal to the number of bits of the information for the pixels of the image, and embedding each 1 bit of the information for each group. An image processing system comprising: an information detecting means for detecting 1 bit of the information from each group of pixels in which is embedded. 3. An information embedding device for embedding predetermined information consisting of 1 or 2 or more bits into an image, wherein the frequency conversion means for frequency converting the image to generate a conversion coefficient group, and the conversion coefficient group. On the other hand, an information embedding device comprising: an information embedding unit that generates a number of groups equal to the number of bits of the information and embeds each 1 bit of the information for each group. 4. The information embedding means selects, from among the plurality of transform coefficients, a transform coefficient having the smallest change in the transform coefficient value due to embedding 1 bit of the information, and selects 1 bit of the information. Is embedded,
The information embedding device according to claim 3. 5. The information embedding unit selects a conversion coefficient having the largest conversion coefficient value from the plurality of conversion coefficients and embeds 1 bit of the information. 4. The information embedding device according to item 4. 6. The information embedding means stores, for each of the plurality of frequency-transformed transform coefficient groups, a change amount of a transform coefficient value resulting from embedding 1 bit of the information, and stores the changed value according to the accumulated value. 4. A conversion coefficient group for embedding 1 bit of the information is determined according to the above-mentioned method.
4. The information embedding device according to either 4 or 5. 7. An information embedding device for embedding predetermined information consisting of 1 or 2 or more bits in an image, for the pixels of the image, generating groups of a number equal to the number of bits of the information, An information embedding device comprising information embedding means for embedding each 1 bit of the above information for the group 8. The information embedding unit selects a pixel having the smallest change in pixel value due to embedding 1 bit of the information among the plurality of pixels, and embeds 1 bit of the information. The information embedding device according to claim 7, characterized in that. 9. The information embedding means selects the pixel having the largest pixel value from the plurality of pixels and embeds 1 bit of the information, according to claim 7. Information embedding device. 10. The information embedding unit divides the image into a plurality of pixel regions, and for each of the plurality of pixel regions,
The variation of the pixel value due to embedding 1 bit of the information is accumulated, and 1 of the information is accumulated according to the accumulated value.
10. The information embedding device according to claim 7, wherein the pixel region in which the bit is embedded is determined. 11. An information detecting device for detecting predetermined information consisting of one or more bits from an image, and an information detecting means for detecting one bit of the information from each group of transform coefficients in which the information is embedded. An information detection device comprising: 12. An information detecting device for detecting predetermined information consisting of one or more bits from an image, comprising an information detecting means for detecting one bit of the information from each group of pixels in which the information is embedded. An information detection device characterized by being provided. 13. An information embedding method for embedding predetermined information consisting of one or more bits into an image, the step of frequency-converting the image to generate a transform coefficient group, and the transform coefficient group, Generating a number of groups equal to the number of bits of the information and embedding each 1 bit of the information for each group as a target. 14. A conversion coefficient having the smallest change in conversion coefficient value due to embedding 1 bit of the information is selected from among the plurality of conversion coefficients, and 1 bit of the information is embedded. The method for embedding information according to claim 13, wherein 15. The conversion coefficient having the largest conversion coefficient value is selected from the plurality of conversion coefficients, and 1 bit of the information is embedded.
Information embedding method described in. 16. A change amount of a transform coefficient value by embedding 1 bit of the information is accumulated for each of the plurality of frequency-transformed transform coefficient groups, and 1 of the information is stored according to the accumulated value. 16. The information embedding method according to claim 13, characterized in that a transform coefficient group in which bits are embedded is determined. 17. An information embedding method for embedding predetermined information consisting of one or more bits in an image, wherein groups of the number of bits equivalent to the number of bits of the information are generated for each pixel of the image, and An information embedding method comprising the step of embedding each 1 bit of the above information for the group 18. One of the information items among the plurality of pixels.
18. The information embedding method according to claim 17, wherein a pixel having a smallest change in pixel value due to embedding a bit is selected and 1 bit of the information is embedded. 19. The information embedding method according to claim 17, wherein a pixel having the largest pixel value is selected from the plurality of pixels and 1 bit of the information is embedded. 20. partitioning the image into a plurality of pixel regions,
A change in pixel value due to embedding 1 bit of the information is accumulated for each of the plurality of pixel regions, and a pixel region in which 1 bit of the information is embedded is determined according to the accumulated value. The information embedding method according to claim 17, 18, or 19. 21. An information detecting method for detecting predetermined information consisting of one or more bits from an image, including the step of detecting one bit of the information from each group of transform coefficients in which the information is embedded. An information detection method characterized by the following. 22. An information detecting method for detecting predetermined information consisting of one or more bits from an image, including the step of detecting one bit of the information from each group of pixels in which the information is embedded. An information detection method characterized by: