JP5322999B2 - Digital watermark embedding device and digital watermark detection device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a digital watermark embedding apparatus and a digital watermark detecting apparatus that simultaneously realize improvement of tolerance to segmentation and suppression of deterioration in image. <P>SOLUTION: In a case where 0s are embedded as digital watermark bits, when a difference &Delta;<SB POS="POST">0</SB>between a function value f<SB POS="POST">v</SB>and a representative value KV<SB POS="POST">0</SB>of 0s is equal to or greater than a threshold value Th, a bit embedding process that edits pixel values of pixels forming an image that is subjected to bit embedding so that the difference &Delta;<SB POS="POST">0</SB>is less than the threshold value Th is performed, while in a case where 1s are embedded as digital watermark bits, when a difference &Delta;<SB POS="POST">1</SB>between a function value f<SB POS="POST">v</SB>and a representative value KV<SB POS="POST">1</SB>of 1s is equal to or greater than the threshold value Th, a bit embedding process that edits the pixel values of the pixels forming the image that is subjected to bit embedding so that the difference &Delta;<SB POS="POST">1</SB>is less than the threshold value Th is performed. <P>COPYRIGHT: (C)2012,JPO&amp;INPIT

Description

  The present invention provides, for example, a digital watermark embedding device that embeds a digital watermark having high resistance to noise mixing and image compression, and a digital watermark that detects a digital watermark embedded by the digital watermark embedding device. The present invention relates to a detection device.

For example, Patent Document 1 below discloses a digital watermark embedding device that embeds a digital watermark having high resistance to noise mixing and image compression in an image.
This digital watermark embedding apparatus is configured to assign each bit constituting digital watermark data to a subset of pixels constituting the original image.
Then, among the representative value representing the bit 0 and the representative value representing the bit to be embedded in the representative value representing the bit 1 arranged at equal intervals, it is calculated from the pixel values of the pixel group constituting the subset. The digital value is embedded by editing the pixel value of the pixel group constituting the subset so that the representative value closer to the function value is selected and the function value matches the representative value. .
Note that image quality degradation is suppressed by selecting and editing pixels that are considered to have little effect on image quality due to image characteristics from the pixel groups that make up the subset. Yes.

Japanese Patent No. 4271018 (paragraph number [0012])

  Since the conventional digital watermark embedding apparatus is configured as described above, if a pixel set corresponding to a certain bit is provided in a local area, among N bits constituting the digital watermark data, Even if some bits are cut off, if bits that have been cut off are detected, it may be possible to restore the N bits constituting the digital watermark data by a technique such as an error correction code. If it is recognized that the local area has a great influence on the image quality by editing the pixel value, the image quality is greatly deteriorated. On the other hand, if pixel sets corresponding to a certain bit are scattered throughout the image, the number of bits that can be cut off is greatly reduced, making it difficult to detect digital watermark data. For this reason, even if a part of the image is cut out, there is a problem that it is impossible to achieve both improvement in characteristics (cutting resistance) capable of detecting digital watermark data and suppression of image quality deterioration.

The present invention has been made to solve the above-described problems, and an object of the present invention is to provide a digital watermark embedding apparatus capable of simultaneously improving cut-off resistance and suppressing image quality deterioration.
Another object of the present invention is to provide a digital watermark detection apparatus capable of detecting a digital watermark embedded by the digital watermark embedding apparatus.

  In the digital watermark embedding device according to the present invention, when the bit embedding unit embeds 0 as a digital watermark bit, the difference between the function value obtained by the function value calculating unit and the 0 representative value set by the representative value setting unit is If the threshold value is greater than or equal to the threshold value, a bit embedding process is performed to edit the pixel value of each pixel constituting the input image so that the difference is less than the threshold value. If the difference between the function value obtained by the means and one representative value set by the representative value setting means is greater than or equal to the threshold, the pixel value of each pixel constituting the input image is edited so that the difference is less than the threshold. If the digital watermark bit is not embedded, the difference between the function value obtained by the function value calculation means and the null representative value set by the representative value setting means is greater than or equal to the threshold value. If, in which so as to implement the process of editing the pixel value of each pixel constituting the input image so that the difference is less than the threshold value.

  According to the present invention, when the bit embedding unit embeds 0 as a digital watermark bit, if the difference between the function value obtained by the function value calculating unit and the 0 representative value set by the representative value setting unit is greater than or equal to the threshold value When the bit embedding process for editing the pixel value of each pixel constituting the input image is performed so that the difference is less than the threshold value and 1 is embedded as the digital watermark bit, the function obtained by the function value calculation means Bit embedding processing for editing the pixel value of each pixel constituting the input image so that if the difference between the value and one representative value set by the representative value setting means is greater than or equal to the threshold, the difference is less than the threshold When the digital watermark bit is not embedded, if the difference between the function value obtained by the function value calculating means and the null representative value set by the representative value setting means is greater than or equal to the threshold value, the difference is the threshold value. Become constitute an input image is so configured as to implement the process of editing the pixel value of each pixel as below, there is an effect that it is possible to realize cut the suppression of resistance increase and image quality degradation at the same time.

1 is a configuration diagram showing a digital watermark embedding device according to Embodiment 1 of the present invention. FIG. It is a flowchart which shows the processing content of the digital watermark embedding apparatus by Embodiment 1 of this invention. It is explanatory drawing which shows an input image, a data embedding object image, a bit embedding object image, and an area where pixel values are added / subtracted by the function value calculation unit 3. 0 is an explanatory diagram showing a setting example of the representative values _KV 0, 1 representative value KV ₁ and null representative value _{KV null.} It is a block diagram which shows the digital watermark detection apparatus by Embodiment 2 of this invention. It is a flowchart which shows the processing content of the digital watermark detection apparatus by Embodiment 2 of this invention. It is explanatory drawing which shows the example of a production | generation of the input image of a digital watermark detection apparatus.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing a digital watermark embedding apparatus according to Embodiment 1 of the present invention.
In FIG. 1, the data embedding target image extraction unit 1 divides an input image that is an uncompressed luminance image to acquire one or more local region images (partial images of the input image). Are sequentially selected from among the partial images (hereinafter referred to as “data embedding target image”), and the data embedding target image is sent to the bit embedding target image extracting unit 2. Perform the output process.

The bit embedding target image extracting unit 2 divides the data embedding target image output from the data embedding target image extracting unit 1 to generate one or more local region images (partial images of the data embedding target image). ) To select partial images (hereinafter referred to as “bit embedding target images”) in which digital watermark bits are embedded from one or more partial images, and the bit embedding target images are selected as functions. A process of outputting to the value calculation unit 3, the image quality degradation degree evaluation unit 5, and the bit embedding unit 7 is performed.
The data embedding target image extracting unit 1 and the bit embedding target image extracting unit 2 constitute an image dividing unit, and the data embedding target image extracting unit 1 and the bit embedding target image extracting unit 2 are partially included. Image selection means is configured.

The function value calculation unit 3 performs an operation of a function for adding and subtracting the pixel value of each pixel constituting the bit embedding target image selected by the bit embedding target image extracting unit 2 to calculate the pixel value of each pixel. A process of obtaining a function value f _v indicating the addition / subtraction result is performed. The function value calculation unit 3 constitutes function value calculation means.
Representative value setting unit 4 in the function value f _v possible values of the function calculated by the function value calculation unit 3 is executed, the value of the representative representing the bit 0 is 0 representative value KV 0 _(e.g., KV ₀ = −750, −250, 500, 1000), the representative value representing bit 1 is 1 representative value KV ₁ (for example, KV ₁ = −1000, −500, 250, 750), and the representative value representing null. Is set as a null representative value KV _null (for example, KV _null = 0), and has a built-in memory for storing the 0 representative value KV ₀ , the 1 representative value KV ₁ and the null representative value KV _null . The representative value setting unit 4 constitutes a representative value setting unit.
The example of FIG. 1 shows an example in which the representative value setting unit 4 is provided separately from the bit embedding unit 7, but even if the bit embedding unit 7 incorporates the function of the representative value setting unit 4. Good.

The image quality degradation degree evaluation unit 5 executes a function calculation that quantifies the degradation degree of image quality that occurs when the pixel value of each pixel constituting the bit embedding target image is edited by the bit embedding unit 7. Then, a process of outputting a numerical value f _deg indicating the image quality degradation degree to the bit embedding unit 7 is performed. The image quality deterioration degree evaluation unit 5 constitutes an image quality deterioration degree calculation unit.
However, when the numerical value f _deg indicating the image quality deterioration level output from the image quality deterioration level evaluation unit 5 indicates deterioration exceeding the allowable range, the function value f _v obtained by the function value calculation unit 3 is changed to the null representative value KV _null . A function for which the calculation is performed by the image quality degradation degree evaluation unit 5 and a function for which the calculation is performed by the function value calculation unit 3 are defined so as to approach each other.

  The embedding target bit extracting unit 6 inputs, for example, 48-bit digital watermark data to be embedded in the data embedding target image output from the data embedding target image extracting unit 1, and from each bit constituting the digital watermark data A process of extracting digital watermark bits to be embedded in the bit embedding target image and outputting the digital watermark bits to the bit embedding unit 7 is performed. The embedding target bit extracting unit 6 constitutes a digital watermark bit extracting unit.

When the digital watermark bit output from the embedding target bit extraction unit 6 is bit 0, the bit embedding unit 7 has a numerical value f _deg indicating the image quality degradation level output from the image quality degradation level evaluation unit 5 within an allowable range. It represents deterioration, and the difference Δ ₀ (Δ ₀ = | f _v −KV ₀₎ between the function value f _v obtained by the function value calculation unit 3 and the 0 representative value KV ₀ stored in the memory of the representative value setting unit 4 |) is a threshold value Th (e.g., if Th = 10) or more, as the difference delta ₀ is less than the threshold value Th, constitute a bit embedding target image selected by the bit embedding target image extracting unit 2 A bit embedding process for editing the pixel value of each pixel is executed.
In addition, when the digital watermark bit output from the embedding target bit extraction unit 6 is bit 1, the bit embedding unit 7 has a numerical value f _deg indicating the image quality degradation level output from the image quality degradation level evaluation unit 5 within an allowable range. The difference Δ ₁ (Δ ₁ = | f _v − between the function value f _v obtained by the function value calculator 3 and the one representative value KV ₁ stored in the memory of the representative value setting unit 4. If KV ₁ |) is equal to or greater than the threshold value Th, the difference Δ ₁ of each pixel constituting the bit embedding target image selected by the bit embedding target image extracting unit 2 is set so as to be less than the threshold value Th. A bit embedding process for editing the pixel value is performed.

When the bit embedding unit 7 does not embed the digital watermark bit output from the embedding target bit extracting unit 6, the numerical value f _deg indicating the image quality degradation level output from the image quality degradation level evaluation unit 5 is within the allowable range. This represents the deterioration, and the difference Δ _null (Δ _null = | f _v −KV _null) between the function value f _v obtained by the function value calculation unit 3 and the null representative value KV _null stored in the memory of the representative value setting unit 4 If |) is equal to or greater than the threshold Th, the pixel value of each pixel constituting the bit embedding target image selected by the bit embedding target image extracting unit 2 so that the difference Δ _null is less than the threshold Th. Perform the process of editing.
However, if the numerical value f _deg indicating the image quality degradation level output from the image quality degradation level evaluation unit 5 indicates degradation exceeding an allowable range, the bit embedding unit 7 determines that the difference Δ ₀ , the difference Δ ₁ , and the difference Δ _null are Even if the threshold value Th is equal to or higher than the threshold Th, the bit embedding process for editing the pixel value of each pixel constituting the bit embedding target image is not performed. The bit embedding unit 7 constitutes a bit embedding unit.
In the first embodiment, when the numerical value f _deg indicating the degree of image quality degradation indicates degradation within an allowable range, processing is performed in which bit 0 is embedded, bit 1 is embedded, or digital watermark is not embedded. As shown, the bit 0 or 1 can always be embedded when the numerical value f _deg indicating the degree of image quality deterioration indicates deterioration within an allowable range.

The bit embedding target image combining unit 8 combines all the bit embedding target images that have been subjected to the bit embedding processing by the bit embedding unit 7, thereby outputting the data embedding target image extracting unit 1. Processing for generating a data embedding target image after embedding processing in which digital watermark data is embedded in the embedding target image is performed.
The data embedding target image combining unit 9 combines all the data embedding target images generated by the bit embedding target image combining unit 8, so that the digital watermark data is embedded in the input image. A process for generating an input image is performed.
The bit embedding target image combining unit 8 and the data embedding target image combining unit 9 constitute image combining means.

In the example of FIG. 1, a data embedding target image extracting unit 1, a bit embedding target image extracting unit 2, a function value calculating unit 3, a representative value setting unit 4, and an image quality degradation degree evaluation, which are components of the digital watermark embedding device. Each of the unit 5, the embedding target bit extracting unit 6, the bit embedding unit 7, the bit embedding target image combining unit 8 and the data embedding target image combining unit 9 has dedicated hardware (for example, a CPU mounted thereon It is assumed that the device is composed of a semiconductor integrated circuit or a one-chip microcomputer. However, when the digital watermark embedding device is composed of a computer, the data embedding target image extracting unit 1, the bit embedding target Image extraction unit 2, function value calculation unit 3, representative value setting unit 4, image quality degradation degree evaluation unit 5, embedding target bit extraction unit 6, bit embedding unit 7, bit embedding target image combining unit 8 and data embedding Processing contents of the target image combining unit 9 Stores programs that are mentioned in the memory of the computer, may execute a program that the CPU of the computer is stored in the memory.
FIG. 2 is a flowchart showing the processing contents of the digital watermark embedding apparatus according to Embodiment 1 of the present invention.

Next, the operation will be described.
FIG. 3 is an explanatory diagram showing an input image, a data embedding target image, a bit embedding target image, and an area where pixel values are added / subtracted by the function value calculation unit 3.
When an image as shown in FIG. 3 is input, the data embedding target image extraction unit 1 divides the input image and acquires one or more local region images (partial images of the input image).
Then, the data embedding target image extracting unit 1 sequentially selects partial images in which the digital watermark data is embedded from one or more partial images, and uses the partial images as data embedding target images. It outputs to the image extraction part 2 (step ST1).

In the input image shown in FIG. 3, the horizontal pixel is W _in and the vertical pixel is H _in .
In the first embodiment, it is assumed that the horizontal pixel W _in of the input image is 500 pixels and the vertical pixel H _in is 300 pixels. The input image is a rectangle having a horizontal pixel W _data of 192 pixels and a vertical pixel H _data of 144 pixels. It is assumed that each partial image is selected in order of raster scanning from the upper left of the input image.

In the example of FIG. 3, four partial images having a horizontal pixel W _data of 192 pixels and a vertical pixel H _data of 144 pixels are generated, and the horizontal pixel W _data existing at a position in contact with the right side of the input image is 116 pixels. Two partial images having a vertical pixel H _data of 144 pixels are generated.
In addition, two partial images having a horizontal pixel W _data of 192 pixels and a vertical pixel H _data of 12 pixels existing at a position in contact with the lower side of the input image are generated and exist at the lower right position of the input image. One partial image in which the horizontal pixel W _data is 116 pixels and the vertical pixel H _data is 12 pixels is generated.
As a result, a total of nine partial images are generated, the partial image existing at the upper left position of the input image is first selected as the data embedding target image, and the remaining eight partial images are data in the raster scan order. The images are sequentially selected as images to be embedded.

When the bit embedding target image extracting unit 2 receives the data embedding target image as shown in FIG. 3 from the data embedding target image extracting unit 1, the bit embedding target image extracting unit 2 divides the data embedding target image into one or more local embedding target images. An image of a specific region (a partial image of the data embedding target image) is acquired.
Then, the bit embedding target image extraction unit 2 sequentially selects partial images in which the digital watermark bits are embedded from one or more partial images, and sets the partial images as bit embedding target images. 3. Output to the image quality degradation evaluation unit 5 and the bit embedding unit 7 (step ST2).

In the example of FIG. 3, the data embedding target image is divided into a square area in which the horizontal pixel W _bit is 24 pixels and the vertical pixel H _bit is 24 pixels, so that the horizontal pixel W _bit is 24 pixels at most and the vertical pixel H _{bit is} A maximum of 48 partial images having a maximum of 24 pixels are generated.
The partial image existing at the upper left position of the data embedding target image is first selected as the bit embedding target image, and the remaining partial images are sequentially selected as the bit embedding target images in the raster scan order.

When the 48-bit digital watermark data to be embedded in the data embedding target image output from the data embedding target image extracting unit 1 is input, the embedding target bit extracting unit 6 receives from each bit constituting the digital watermark data. The digital watermark bits embedded in the bit embedding target image are extracted, and the digital watermark bits are output to the bit embedding unit 7 (step ST3).
In the example of FIG. 3, each bit embedding target image included in the data embedding target image is 24x + 1 to 24x + 24 pixels (0 ≦ x ≦ 7) in the horizontal direction from the left end of the data embedding target image. And from the upper end of the data embedding target image in the vertical direction at positions 24y + 1 to 24y + 24 (0 ≦ y ≦ 5).

Therefore, for example, if the digital watermark bit embedded in each bit embedding target image included in the data embedding target image is the (x + 8y + 1) th bit of the digital watermark data, for example, the first pixel to the 24th pixel in the horizontal direction The bit embedding target image existing at the position of (x = 0) and at the position of the first pixel to the 24th pixel (y = 0) in the vertical direction includes the first bit of the digital watermark data. (= 0 + 8 × 0 + 1) is assigned.
Further, for example, a bit that exists in the position of the 49th pixel to the 72nd pixel (x = 2) in the horizontal direction and exists in the position of the 73rd pixel to the 96th pixel (y = 3) in the vertical direction. The 27th (= 2 + 8 × 3 + 1) bit of the digital watermark data is assigned to the embedding target image.

When the bit embedding target image extracting unit 2 selects a bit embedding target image, the function value calculation unit 3 performs a function operation for adding and subtracting the pixel value of each pixel constituting the bit embedding target image. Then, the function value f _v indicating the addition / subtraction result of the pixel value of each pixel is calculated, and the function value f _v is output to the bit embedding unit 7 (step ST4).
In the example of FIG. 3, among the pixels constituting the bit embedding target image, an area where pixels shown in dark gray are gathered is an addition target area, and an area where pixels shown in light gray are gathered. The function value f _v is calculated by subtracting the pixel values of all the pixels in the subtraction target area while adding the pixel values of all the pixels in the subtraction target area.

That is, if the pixel value of the x-th pixel (1 ≦ x ≦ 24) from the left of the bit embedding target image and the y-th pixel (1 ≦ y ≦ 24) from the top are f _bit (x, y), the function value f _v is calculated as in the following equation (1).
f _v
= G (1,13) + g (3,1) + g (4,20) + g (7,7)
+ G (8,12) + g (16,21) + g (17,9) + g (17,16)
+ G (20, 2) + g (21, 21)
-G (1,9) -g (3,5) -g (8,16) -g (8,20)
-G (11,7) -g (12,21) -g (13,16) -g (16,2)
-G (21, 9) -g (21, 17)
(1)

G (x, y) at f _v in the formula (1) is expressed as the following formula (2) or formula (3).
g (x, y)
= F _bit (x, y) + f _bit (x, y + 1)
+ F _bit (x, y + 2) + f _bit (x, y + 3)
+ F _bit (x + 1, y) + f _bit (x + 1, y + 1)
+ F _bit (x + 1, y + 2) + f _bit (x + 1, y + 3)
+ F _bit (x + 2, y) + f _bit (x + 2, y + 1)
+ F _bit (x + 2, y + 2) + f _bit (x + 2, y + 3)
+ F _bit (x + 3, y) + f _bit (x + 3, y + 1)
+ F _bit (x + 3, y + 2) + f _bit (x + 3, y + 3) (2)

g (x, y)
= (F _bit (x, y) + f _bit (x, y + 1)
+ F _bit (x, y + 2) + f _bit (x, y + 3)
+ F _bit (x + 1, y) + f _bit (x + 1, y + 1)
+ F _bit (x + 1, y + 2) + f _bit (x + 1, y + 3)
+ F _bit (x + 2, y) + f _bit (x + 2, y + 1)
+ F _bit (x + 2, y + 2) + f _bit (x + 2, y + 3)
+ F _bit (x + 3, y) + f _bit (x + 3, y + 1)
+ F _bit (x + 3, y + 2) + f _bit (x + 3, y + 3)) / 16 (3)

However, since the horizontal size or vertical size of the bit embedding target image is less than 24, and there is a nonexistent pixel, the pixel value f _bit (x, y) of such a pixel is set to “0” and the expression ( 1) shall apply.

As described above, the function value calculation unit 3 performs a function operation for adding and subtracting the pixel value of each pixel constituting the bit embedding target image, and the function value f _v of the function has a possible value. The range is finite, and among the possible values of the function value f _v of the function, the representative value representing bit 0 is 0 representative value KV ₀ , the representative value representing bit 1 is 1 representative value KV ₁ , null A representative value representing (null) is preset in the representative value setting unit 4 as a null representative value KV _null .
FIG. 4 is an explanatory diagram showing a setting example of 0 representative value KV ₀ , 1 representative value KV ₁ and null representative value KV _null .
In the example of FIG. 4, the 0 representative value KV ₀ is “−750”, “−250”, “500”, “1000”, and the 1 representative value KV ₁ is “−1000”, “−500”, “250”, “750” and the null representative value KV _null are set to “0”.

When the bit embedding target image extraction unit 2 selects the bit embedding target image, the image quality degradation degree evaluation unit 5 determines the pixel value of each pixel constituting the bit embedding target image by the bit embedding unit 7 described later. by performing the operation of a function that quantifies the degree of deterioration of image quality caused when edited to calculate a numerical value f _deg indicating the image quality degradation degree, bit embedding section 7 a number f _deg indicating the image quality degradation degree (Step ST5).
For example, if the pixel value of the x-th pixel (1 ≦ x ≦ 24) from the left and the y-th pixel (1 ≦ y ≦ 24) from the top of the bit embedding target image is f _bit (x, y), the image quality degradation degree The numerical value f _deg indicating is calculated as in the following equation (4).
f _deg
= _{Σx∈ [1,6] Σy∈ [1,6]} | f _bit (4x, 4y) −f _bit (4x−2,4y−2) |
(4)
However, since the horizontal size or vertical size of the bit embedding target image is less than 24, and there is a nonexistent pixel, the pixel value f _bit (x, y) of such a pixel is set to “0” and the expression ( 4) shall apply.

When the numerical value f _deg indicating the degree of image quality degradation is calculated as in Expression (4), when the pixel value changes drastically in the bit embedding target image, the numerical value f _deg indicating the degree of image quality deterioration increases, and the pixel value In the case where the change of is moderate, the numerical value f _deg indicating the degree of image quality deterioration becomes small.
In general, it is known that a monotonous image is prominent in image quality degradation when editing pixel values. When a numerical value f _deg indicating an image quality degradation degree is defined as in Expression (4), a numerical value indicating the image quality degradation degree is defined. A smaller f _deg means that image quality deterioration is more conspicuous.
However, when the numerical value f _deg indicating the image quality deterioration level output from the image quality deterioration level evaluation unit 5 indicates deterioration exceeding the allowable range, the function value f _v obtained by the function value calculation unit 3 is changed to the null representative value KV _null . A function that is calculated by the image quality degradation degree evaluation unit 5 and a function that is calculated by the function value calculation unit 3 are defined so as to approach each other (f _v ≈KV _null = 0).
For example, the numerical value f _deg indicating the degree of image quality degradation is defined to be a small value in the case of a bit embedding target image with small variations in pixel values.

The bit embedding unit 7 indicates that the numerical value f _deg indicating the image quality deterioration level output from the image quality deterioration level evaluation unit 5 represents deterioration within an allowable range (the numerical value f _deg is _equal to or greater than a preset threshold S). In the case of embedding the digital watermark bit in the bit embedding target image, it is determined that there is no significant image quality degradation, and the embedding target bit is selected for the bit embedding target image selected by the bit embedding target image extracting unit 2. A bit embedding process for embedding digital watermark bits output from the extraction unit 6 is performed (step ST6).
On the other hand, if the numerical value f _deg indicating the image quality deterioration level output from the image quality deterioration level evaluation unit 5 represents deterioration exceeding the allowable range (when the numerical value f _deg is less than the preset threshold value S), the electronic It is determined that when the watermark bit is embedded in the bit embedding target image, the image quality is greatly deteriorated, and the bit embedding process of the digital watermark bit is performed on the bit embedding target image selected by the bit embedding target image extraction unit 2. Instead, the bit embedding target image is directly output to the bit embedding target image combining unit 8 as a bit embedding target image after the bit embedding process.
Note that the threshold value S can be arbitrarily set by the user according to the degree to which image quality deterioration is allowed.

Hereinafter, a specific bit embedding process of the bit embedding unit 7 in the case where the numerical value f _deg indicating the image quality degradation level output from the image quality degradation level evaluation unit 5 represents degradation within an allowable range will be described.
(1) When the digital watermark bit output from the embedding target bit extraction unit 6 is bit 0 The bit embedding unit 7 is a function obtained by the function value calculation unit 3 as shown in the following equation (5). A difference Δ ₀ between the value f _v and the 0 representative value KV ₀ stored in the memory of the representative value setting unit 4 is calculated.
Δ ₀ = | f _v −KV ₀ | (5)
However, 0 the representative value KV _0, as shown in FIG. 4, since they are a plurality set (e.g., -750, -250,500,1000), from among the plurality of 0 representative value KV _0, the function The 0 representative value KV ₀ that is closest to the value f _v is selected to calculate equation (5).
For example, if the function value f _v is “740”, the 0 representative value KV ₀ of “500” is selected, and if the function value f _v is “760”, the 0 representative value KV ₀ of “1000” is selected. Then, equation (5) is calculated.

When calculating the equation (5), the bit embedding unit 7 compares the difference Δ ₀ with a preset threshold Th (for example, Th = 10), and if the difference Δ ₀ is equal to or greater than the threshold Th (Δ ₀ ≧ Th), so the difference delta ₀ is less than the threshold value Th, the bit edit the pixel value of each pixel constituting the bit embedding target image selected by the bit embedding target image extracting unit 2 embedding Implementation process.
For example, if the 0 representative value KV ₀ of “500” is selected, the pixel value of each pixel constituting the bit embedding target image is edited so that the function value f _v becomes “500”. Alternatively, if the threshold value Th is “10”, the pixel value of each pixel constituting the bit embedding target image is edited so that the function value f _v falls within the range of 490 <f _v <510. .
If the difference Δ ₀ is less than the threshold Th (Δ ₀ <Th), in the digital watermark detection apparatus described later without editing the pixel value of each pixel constituting the bit embedding target image, Since the digital watermark bit of bit 0 can be accurately detected from the bit embedding target image, the pixel value of each pixel constituting the bit embedding target image is not edited.
Here, if it is less than the difference delta ₀ is the threshold value Th (Δ _{0 <Th),} an example is shown not to edit the pixel value of each pixel constituting the bit embedding target image, the difference delta ₀ is the threshold Even if it is less than Th, it is possible to implement such that the pixel value is edited so that the function value f _v is as close to the 0 representative value KV ₀ as possible.

(2) When the digital watermark bit output from the embedding target bit extraction unit 6 is bit 1 The bit embedding unit 7 is a function obtained by the function value calculation unit 3 as shown in the following equation (6). A difference Δ ₁ between the value f _v and one representative value KV ₁ stored in the memory of the representative value setting unit 4 is calculated.
Δ ₁ = | f _v −KV ₁ | (6)
However, since one representative value KV ₁ is set as shown in FIG. 4 (for example, −1000, −500, 250, 750), a function is selected from the plurality of one representative value KV _1. One representative value KV ₁ closest to the value f _v is selected, and the formula (6) is calculated.
For example, if the function value f _v is “510”, one representative value KV ₁ of “750” is selected, and if the function value f _v is “490”, one representative value KV ₁ of “250” is selected. Then, Equation (6) is calculated.

When calculating the equation (6), the bit embedding unit 7 compares the difference Δ ₁ with a preset threshold Th (for example, Th = 10), and if the difference Δ ₁ is equal to or greater than the threshold Th (Δ ₁ ≧ Th), as the difference delta ₁ is less than the threshold value Th, the bit edit the pixel value of each pixel constituting the bit embedding target image selected by the bit embedding target image extracting unit 2 embedding Implementation process.
For example, if the function value f _v is “320”, the pixel value of each pixel constituting the bit embedding target image is edited so that the function value f _v becomes “250”. Alternatively, if the threshold Th is “10”, the pixel value of each pixel constituting the bit embedding target image is edited so that the function value f _v falls within the range of 240 <f _v <260. .

However, when the digital watermark bit is bit 1, when editing the pixel value of each pixel constituting the bit embedding target image, the pixel that affects the function value f _v (the darkness in the bit embedding target image). The pixels shown in gray (pixels in the addition target area) or the pixels shown in light gray (pixels in the subtraction target area)) are edited.
For example, among the pixels which affects the function value f _v, edits to increase or decrease in order from the upper left pixel of the pixel value by one.
That is, when the function value f _v is increased to approach one representative value KV ₁ , the pixel value of the pixel shown in dark gray (the pixel in the addition target area) is incremented by 1 and is shown in light gray. The process of subtracting 1 from the pixel value of the current pixel (the pixel in the subtraction target area) is repeatedly performed.
On the other hand, when the function value f _v is decreased to approach one representative value KV ₁ , the pixel value of the pixel indicated by dark gray is decremented by 1, and the pixel value of the pixel indicated by light gray is reduced to 1. Repeat the process of adding.

Even without editing the difference delta ₁ is less than the threshold value Th (Δ _{1 <Th),} the pixel value of each pixel constituting the bit embedding target image, the digital watermark detection apparatus described later, Since the digital watermark bit of bit 1 can be accurately detected from the bit embedding target image, the pixel value of each pixel constituting the bit embedding target image is not edited.
Here, if it is less than the difference delta ₁ is the threshold Th (Δ _{1 <Th),} an example is shown not to edit the pixel value of each pixel constituting the bit embedding target image, the difference delta ₁ is the threshold Even if it is less than Th, it is possible to implement so that the pixel value is edited so that the function value f _v is as close to the one representative value KV ₁ as possible.

(3) When the digital watermark bit output from the embedding target bit extracting unit 6 is not embedded The bit embedding unit 7 obtains the function value f obtained by the function value calculating unit 3 as shown in the following equation (7). A difference Δ _null between _v and the null representative value KV _null stored in the memory of the representative value setting unit 4 is calculated.
Δ _null = | f _v −KV _null | (7)
When calculating the equation (7), the bit embedding unit 7 compares the difference Δ _null with a preset threshold Th (for example, Th = 10), and if the difference Δ _null is equal to or greater than the threshold Th (Δ _null ≧ Th), and a process of editing the pixel value of each pixel constituting the bit embedding target image selected by the bit embedding target image extracting unit 2 so that the difference Δ _null is less than the threshold Th. carry out.

For example, if the function value f _v is “80”, the pixel value of each pixel constituting the bit embedding target image is edited so that the function value f _v becomes “0”. Alternatively, if the threshold Th is “10”, the pixel value of each pixel constituting the bit embedding target image is edited so that the function value f _v falls within the range of −10 <f _v <10. To do.
If the difference Δ _null is less than the threshold Th (Δ _null <Th), the digital watermark detection device described later can edit the pixel value of each pixel constituting the bit embedding target image. Since null can be accurately detected from the bit embedding target image, the pixel value of each pixel constituting the bit embedding target image is not edited.

The bit embedding unit 7 indicates that the numerical value f _deg indicating the degree of image quality deterioration output from the image quality deterioration degree evaluation unit 5 represents deterioration exceeding the allowable range (the numerical value f _deg satisfies the preset threshold value S). If not), as described above, the bit embedding process of the digital watermark bit for the bit embedding target image is not performed.
However, in the bit embedding target image after the bit embedding processing that is output from the bit embedding unit 7 to the bit embedding target image combining unit 8, the function value f _v matches the null representative value KV _null or , obtained it is necessary to have become null representative value KV _null value close, as described above, when the numerical value f _deg indicating the image quality degradation degree represents degradation unacceptable, the function value calculation unit 3 The function value f _v is approximated to the null representative value KV _null (f _v ≈KV _null = 0), and the function is calculated by the image quality degradation degree evaluation unit 5 and the function value calculation unit 3 performs the calculation. Therefore, when the numerical value f _deg indicating the image quality deterioration level indicates deterioration exceeding an allowable range, it is not necessary to perform bit embedding processing.

Note that, in a certain bit embedding target image, when the numerical value f _deg indicating the degree of image quality degradation is small (when the predetermined threshold value S is not reached), the function value f _v obtained by the function value calculation unit 3 is a null representative. The reason for approaching the value KV _null is that the numerical value f _deg indicating the degree of image quality degradation is defined to be a small value in the case of a bit embedding target image with a small variation in pixel value. This is because the function value f _v takes a value close to “0”.
That is, in the bit embedding target image shown in FIG. 3, the addition target area and the subtraction target area for calculating the function value f _v are necessarily adjacent to each other, and the number of pixels in the addition target area and the subtraction target area Since the same number of pixels is included, when the function value _fv is calculated, the sum of the pixel values in the addition target area and the sum of the pixel values in the subtraction target area act so as to cancel each other.
For this reason, the function value f _v is close to “0”. As an extreme example, in a completely monotonous bit embedding target image having the same pixel value, it can be understood that both the function value f _v and the numerical value f _deg indicating the image quality degradation level are “0”. Easy.

In steps ST2 to ST6, all the bit embedding target images are selected by the bit embedding target image extracting unit 2, and all the bit embedding target images after the bit embedding processing are selected from the bit embedding unit 7. The process is repeated until it is output to the bit embedding target image combining unit 8 (step ST7).
Each time the bit embedding target image combining unit 8 receives the bit embedding target image after the bit embedding processing from the bit embedding unit 7, it arranges the bit embedding target images in the raster scan order, By combining the images, a data embedding target image after embedding processing in which digital watermark data is embedded in the data embedding target image output from the data embedding target image extracting unit 1 is generated (step ST8). .

In the processes from step ST1 to ST8, all the data embedding target images are selected by the data embedding target image extracting unit 1, and all the data embedding targets after the embedding processing are selected from the bit embedding target image combining unit 8. The process is repeated until the image is output to the data embedding target image combining unit 9 (step ST9).
Each time the data embedding target image combining unit 9 receives the data embedding target image after the embedding process from the bit embedding target image combining unit 8, the data embedding target image is arranged in the raster scan order, and all the data embedding target images are arranged. By combining the embedding target images, an input image after embedding processing in which digital watermark data is embedded in the input image is generated (step ST10).

As is apparent from the above, according to the first embodiment, when the bit embedding unit 7 embeds 0 as a digital watermark bit, the function value f _v obtained by the function value calculating unit 3 and the representative value setting unit 4 If the difference Δ _{0 of} the 0 representative value KV ₀ set by the above is greater than or equal to the threshold Th, the pixel value of each pixel constituting the bit embedding target image is edited so that the difference Δ ₀ is less than the threshold Th When a bit embedding process is performed and 1 is embedded as a digital watermark bit, the difference Δ ₁ between the function value f _v obtained by the function value calculation unit 3 and the one representative value KV ₁ set by the representative value setting unit 4 is if the threshold value Th or more, conducted bit embedding processing to edit the pixel value of each pixel constituting the bit embedding target image as the difference delta ₁ is less than the threshold value Th, watermarked bit If not, calculate function value If the difference Δ _null between the function value f _v obtained by the unit 3 and the null representative value KV _null set by the representative value setting unit 4 is equal to or larger than the threshold Th, the bit is embedded so that the difference Δ _null is less than the threshold Th. Since the processing for editing the pixel value of each pixel constituting the embedding target image is performed, there is an effect that it is possible to simultaneously realize the improvement of the clipping resistance and the suppression of the image quality deterioration.

  That is, in the first embodiment, the data embedding target image extracting unit 1 divides an input image at equal intervals, and repeatedly embeds 48-bit digital watermark data in a partial image of 192 pixels wide and 144 pixels high. It is out. Therefore, even if an input image in which digital watermark data is embedded is cut and edited, if an image of 383 pixels in width and 287 pixels in length remains, the entire data embedding target image is always included in the image. Therefore, compared to the conventional digital watermark embedding device that embeds each bit of the digital watermark data in the entire input image, the detectability of the digital watermark data when the input image after the embedding process is cut out (cut off) Resistance).

  Further, in the first embodiment, the bit embedding target image extracting unit 2 divides the data embedding target image selected by the data embedding target image extracting unit 1 at equal intervals, and is 24 pixels wide and 24 pixels long. 48 digital watermark bits different from each other are embedded in 48 partial images. Therefore, even if an input image in which a digital watermark bit is embedded is cut and edited, if an image with a width of 215 pixels and a length of 167 pixels remains, it is always necessary to have 48 different watermark bits in the image. Since the corresponding bit embedding target image is included, compared with the conventional digital watermark embedding device that embeds 48 digital watermark bits in the entire input image, the electronic when the input image after embedding processing is cut out It is possible to increase the detectability (cutout resistance) of watermark bits.

According to the first embodiment, the embedding target bit extracting unit 6 extracts the embedding target digital watermark bits from each bit constituting the 48-bit digital watermark data, and the digital watermark bits are bit-embedded. Since it is configured to output to the embedding unit 7, if 48 digital watermark bits are detected, the embedded digital watermark data can be restored.
In other words, compared with a conventional digital watermark embedding device that embeds digital watermark data in the entire input image, the detectability (cut-off resistance) of the digital watermark data when the input image after the embedding process is cut is increased. Can do.

Further, according to the first embodiment, the function value obtained by the function value calculation unit 3 when the numerical value f _deg indicating the image quality degradation level output from the image quality degradation level evaluation unit 5 represents degradation exceeding an allowable range. f _v to approach the null representative value KV _null, the function calculated by the image quality degradation degree evaluation unit 5 is executed, and functions calculated by the function value calculation unit 3 is performed is constituted as defined That is, that is, when the numerical value f _deg indicating the image quality deterioration level output from the image quality deterioration level evaluation unit 5 indicates deterioration exceeding the allowable range, the pixel value of each pixel constituting the bit embedding target image is edited. Even if there is no function, a function that is calculated by the image quality degradation degree evaluation unit 5 and a function value calculation unit 3 so that the function value f _v is likely to be close to the null representative value KV _null. The operation is Since it defines a function that is, (if the number f _deg showing the image quality degradation degree is a value close to 0) bit embedding target image, may involve a large image quality degradation when changing the pixel values, the bit A state without embedding can be expressed without editing the pixel value of each pixel constituting the embedding target image.
For this reason, even if a bit embedding target image, which has a high risk of image quality degradation due to editing of pixel values, is not edited at all, the digital watermark detecting device side described later includes a digital watermark bit in the bit embedding target image. It is possible to determine that there is a high possibility that it is not embedded. Therefore, it is possible to suppress adverse effects on detection while suppressing image quality deterioration.

In the first embodiment, the data embedding target image extracting unit 1 divides the input image and sequentially selects the data embedding target image of 192 pixels in width and 144 pixels in length. However, the data embedding target images of other sizes may be sequentially selected.
For example, a data embedding target image having horizontal 96 pixels and vertical 144 pixels may be selected.
In this case, if the size of the bit embedding target image selected by the bit embedding target image extracting unit 2 is, for example, horizontal 24 pixels and vertical 12 pixels, 48 bits in one data embedding target image. Since an embedding target image can be prepared, 48-bit digital watermark data can be embedded.

In addition, the size of the data embedding target image and the bit embedding target image can be reduced.
When the size of the data embedding target image is reduced, the area where the same bit is embedded is frequently repeated. Therefore, even if the input image after embedding processing is cut out to a small size, there is a possibility that all bits of the digital watermark data can be detected. Can be increased.
As described above, the data embedding target image extraction unit 1 can be configured to select a data embedding target image of an arbitrary size.

In the first embodiment, the bit embedding target image extracting unit 2 divides the data embedding target image and sequentially selects the bit embedding target image of 24 pixels wide and 24 pixels high. Alternatively, the division size of the data embedding target image may be changed, and bit embedding target images of other sizes may be sequentially selected.
For example, a bit embedding target image having a width of 12 pixels and a height of 24 pixels may be selected.
In this case, when the data embedding target image selected by the data embedding target image extracting unit 1 is 192 pixels wide and 144 pixels long, 96 bit embedding target images are included in one data embedding target image. Since the data can be prepared, the number of bits of the digital watermark data to be embedded in the data embedding target image can be 96 bits.

If the size of the bit embedding target image is reduced, the amount of digital watermark data that can be embedded can be increased.
As described above, the bit embedding target image extracting unit 2 can be configured to select a bit embedding target image of an arbitrary size.
Further, the number of bits of the digital watermark data, the size of the data embedding target image, and the size of the bit embedding target image can be arbitrarily set as long as they are appropriately set.

  In the first embodiment, the embedding target bit extracting unit 6 is located in the position of 24x + 1 to 24x + 24 pixels from the left end in the horizontal direction and 24y + 1 pixels from the upper end in the vertical direction in the data embedding target image. Although it has been shown that the x + 8y + 1 bit (0 ≦ x ≦ 7, 0 ≦ y ≦ 5) of the digital watermark data is assigned to the bit embedding target image at the position of the 24th + 24th pixel, the above bit embedding is performed. Other bits of the digital watermark data may be assigned to the target image.

For example, each bit of the digital watermark data to be assigned to the bit embedding target image may be reversed, or each bit of the digital watermark data may be assigned based on a mapping rule generated by a pseudo random number. .
Alternatively, 48-bit digital watermark data may be encrypted to generate 48-bit encrypted data, and each bit of the encrypted data may be assigned to the bit embedding target image.
As described above, the embedding target bit extraction unit 6 determines an electronic watermark bit to be embedded from each bit of the digital watermark data and assigns it to an embedding destination bit embedding target image. Can do.

In the first embodiment, the image quality deterioration degree evaluation unit 5 calculates the numerical value f _deg indicating the image quality deterioration degree by calculating Expression (4) as a function for digitizing the image quality deterioration degree. However, any function that quantifies the degree of image quality deterioration may be used, and the numerical value f _deg indicating the degree of image quality deterioration may be calculated by performing calculations other than Equation (4).
For example, a function that calculates the standard deviation of all pixel values of the bit embedding target image can be used.
The standard deviation becomes a positive number close to “0” when the pixel values are nearly uniform, and becomes a large positive number when the pixel values take various values. Therefore, the standard deviation can be used as an index representing the degree of image quality degradation that occurs when the digital watermark bits are embedded.
When the standard deviation of all the pixel values of the bit embedding target image is calculated as the numerical value f _deg indicating the degree of image quality deterioration, the smaller the standard deviation, the easier the image quality deterioration is noticeable.
Here, the numerical value f _deg indicating the degree of image quality deterioration is shown for calculating the standard deviation of all pixel values of the bit embedding target image, but the larger the numerical value f _deg indicating the degree of image quality deterioration, the more noticeable the image quality deterioration is. You may use the numerical formula which means that it is easy.
As described above, the image quality deterioration degree evaluation unit 5 can be configured to calculate the image quality deterioration degree by an arbitrary mathematical formula that estimates the degree to which the image quality deteriorates when the pixel value is changed.

In the first embodiment, the function value calculation unit 3 calculates Expression (1) as a function for adding and subtracting the pixel value of each pixel constituting the bit embedding target image, thereby calculating each pixel. The function value f _v indicating the addition / subtraction result of the pixel value has been described. However, any function that adds / subtracts the pixel value of each pixel may be used, and by performing calculation other than Expression (1), the pixel of each pixel is calculated. A function value f _v indicating the result of addition / subtraction of values may be calculated.
However, in order for the digital watermark embedding apparatus of FIG. 1 to function effectively, if the numerical value f _deg indicating the image quality degradation level output from the image quality degradation level evaluation unit 5 represents degradation exceeding an allowable range, It is necessary to satisfy the property that the value f _v is close to a specific value (the specific value may not necessarily be “0” or one type) with a high probability.

For example, when the numerical value f _deg indicating the degree of image quality deterioration is less than the preset threshold value S, the function value f _v has a characteristic of being concentrated near “100” or “−100” with high probability. The function can be used in the function value calculation unit 3.
In this case, in order for the digital watermark embedding apparatus of FIG. 1 to function effectively, the representative value setting unit 4 sets the null representative value KV _null to “100” and “−100”, which is larger than “100”. Value, a value smaller than “−100”, and a value between “−100” and “100”, the 0 representative value KV so that the intervals between the 0 representative value KV ₀ and the 1 representative value KV ₁ are substantially equal. _It is necessary to set ₀ and 1 representative value KV ₁ .
In this way, in cooperation with the bit embedding unit 7 and the image quality degradation level evaluation unit 5, if the representative value setting and the function setting for evaluating the image quality degradation level are appropriately set, the function value calculation unit 3 As a function for calculating the value f _v , an arbitrary mathematical formula can be used.

  In the first embodiment, the data embedding target image extracting unit 1 and the bit embedding target image extracting unit 2 select partial images in the raster scan order. However, the partial images are selected in any order. Good. Alternatively, a plurality of partial images may be selected at the same time, and the subsequent processing unit may execute processing on the plurality of partial images in parallel.

Further, in the first embodiment, and when a function value calculation unit 3 calculates a function value f _v, when the image quality degradation evaluation unit 5 calculates a numerical value f _deg showing the image quality degradation degree is used in the calculation In the case where a pixel does not exist, the calculation is performed by evaluating the pixel value of the pixel as “0”. This is when the horizontal or vertical size of the bit embedding target image is not a multiple of 24 pixels. Is equivalent to padding the right or lower side with pixels having a pixel value of “0” so as to be a multiple of 24 pixels.
In this case, the right and lower sides of the bit embedding target image output from the bit embedding target image extraction unit 2 are padded as necessary, and the bit embedding after embedding processing output from the bit embedding unit 7 is performed. This is realized by deleting the padding portion of the target image.

As a result, the function value calculation unit 3, the image quality degradation degree evaluation unit 5 and the bit embedding unit 7 used when the vertical and horizontal are multiples of 24 pixels are converted into a bit embedding target image whose horizontal size or vertical size is less than 24 pixels. Can also be used.
In addition to this, it is conceivable that the bit embedding target image whose horizontal size or vertical size is less than 24 pixels is excluded from the digital watermark bit embedding target. In this case, even in the digital watermark detection apparatus, the bit embedding target image whose horizontal size or vertical size is less than 24 pixels is removed from the digital watermark bit embedding target, and the detection process is performed without any problem. be able to.

In the first embodiment, the input image is an uncompressed luminance image (uncompressed monochrome image). However, the present invention is not limited to this. For example, the input image may be an uncompressed color image. Good.
When the input image is an uncompressed color image and the uncompressed color image is in the YCbCr format, the digital watermark embedding apparatus of FIG. 1 may be applied to one component of Y, Cb, and Cr.
If the uncompressed color image is in RGB format, the digital watermark embedding apparatus of FIG. 1 may be applied to one component of R, G, and B.
In the first embodiment, digital watermark data can be embedded in a compressed image (encoded image). In this case, it is only necessary to uncompress (decode) and convert it to an uncompressed image (decoded image), and then embed the digital watermark data by the above procedure.

Embodiment 2. FIG.
FIG. 5 is a block diagram showing a digital watermark detection apparatus according to Embodiment 2 of the present invention.
5, the data detection target image extraction unit 11 has the same function as the data embedding target image extraction unit 1 of FIG. 1, and an input in which digital watermark data is embedded by the digital watermark embedding device of FIG. 1. The image is divided to acquire one or more local region images (partial images of the input image), and from the one or more partial images, a partial image (hereinafter referred to as “data”) is detected. The detection target image is selected in order, and the data detection target image is output to the bit detection target image extraction unit 12.

The bit detection target image extraction unit 12 has the same function as the bit embedding target image extraction unit 2 in FIG. 1, and divides the data detection target image output from the data detection target image extraction unit 11 into one The above-described local region image (partial image of the data detection target image) is acquired, and a partial image (hereinafter referred to as “bit detection target image”) for detecting a digital watermark bit from one or more partial images. ) Are selected in order, and the bit detection target image is output to the function value calculation unit 13 and the bit detection unit 15.
The data detection target image extraction unit 11 and the bit detection target image extraction unit 12 constitute an image dividing unit, and the data detection target image extraction unit 11 and the bit detection target image extraction unit 12 include a partial image selection unit. It is configured.

The function value calculation unit 13 has the same function as the function value calculation unit 3 in FIG. 1, and adds or subtracts the pixel value of each pixel constituting the bit detection target image selected by the bit detection target image extraction unit 12. The process of obtaining the function value f _v indicating the addition / subtraction result of the pixel value of each pixel is executed by executing the calculation of the function. The function value calculation unit 13 constitutes a function value calculation unit.

Similar to the representative value setting unit 4 in FIG. 1, the representative value setting unit 14 represents a representative value representing bit 0 among the values that can be taken by the function value f _v of the function to be calculated by the function value calculation unit 13. Is a representative value KV ₀ (eg, KV ₀ = −750, −250, 500, 1000), and a representative value representing bit 1 is 1 representative value KV ₁ (eg, KV ₁ = −1000, −500, 250, 750), a representative value representing null is set as a null representative value KV _null (for example, KV _null = 0), and 0 representative value KV ₀ , 1 representative value KV ₁ and null representative value KV _null are stored. Built-in memory. The representative value setting unit 14 constitutes a representative value setting unit.
In the example of FIG. 5, an example in which the representative value setting unit 14 is provided separately from the bit detection unit 15 is shown, but the bit detection unit 15 may incorporate the function of the representative value setting unit 14.

Of the 0 representative value KV ₀ , the 1 representative value KV _1, and the null representative value KV _null set by the representative value setting unit 14, the bit detection unit 15 has the function value f _v obtained most by the function value calculation unit 13. If a representative value is selected and the representative value is 0 representative value KV ₀ , a detection result indicating that bit 0 is embedded as a digital watermark bit is output to the digital watermark data generation unit 16, and the representative value is displayed. Is a representative value KV ₁ , a detection result indicating that bit 1 is embedded as a digital watermark bit is output to the digital watermark data generation unit 16, and if the representative value is a null representative value KV _null , A process of outputting a detection result indicating that the digital watermark bit is not embedded to the digital watermark data generation unit 16 is performed. The bit detector 15 constitutes a digital watermark bit detector.

  The digital watermark data generation unit 16 aggregates the detection results relating to all the partial images output from the bit detection unit 15 and performs processing for generating digital watermark data from the aggregation results of the detection results. The digital watermark data generation unit 16 constitutes digital watermark data generation means.

In the example of FIG. 5, the data detection target image extraction unit 11, the bit detection target image extraction unit 12, the function value calculation unit 13, the representative value setting unit 14, the bit detection unit 15, and the digital watermark that are components of the digital watermark detection apparatus It is assumed that each of the data generation units 16 is configured by dedicated hardware (for example, a semiconductor integrated circuit on which a CPU is mounted, or a one-chip microcomputer). , The processing contents of the data detection target image extraction unit 11, the bit detection target image extraction unit 12, the function value calculation unit 13, the representative value setting unit 14, the bit detection unit 15, and the digital watermark data generation unit 16 are described. Stored in the memory of the computer, and the CPU of the computer executes the program stored in the memory. It may be so.
FIG. 6 is a flowchart showing the processing contents of the digital watermark detection apparatus according to Embodiment 2 of the present invention.

Next, the operation will be described.
In general, there is a possibility that the input image is edited, such as clipping or enlargement / reduction, after the digital watermark data is embedded. The input image handled by the digital watermark detection apparatus of the second embodiment is Assume that the image is an uncompressed luminance image that has not been subjected to editing such as clipping or enlargement / reduction after the digital watermark data has been embedded by the digital watermark embedding apparatus of FIG.

The data detection target image extraction unit 11 receives an image in which digital watermark data is embedded by the digital watermark embedding apparatus in FIG. 1 and divides the input image to input one or more local region images (input). Acquire a partial image).
Then, the data detection target image extraction unit 11 sequentially selects partial images for detecting digital watermark data from one or more partial images, and uses the partial images as data detection target images to extract a bit detection target image. It outputs to the part 12 (step ST21).
In the example of FIG. 3, an input image in which the horizontal pixel is W _in and the vertical pixel is H _in is divided into rectangular regions (partial images) in which the horizontal pixel W _data is 192 pixels and the vertical pixel H _data is 144 pixels. Partial images are selected one by one in the raster scan order from the upper left of the input image.

Assuming that the horizontal pixel W _in of the input image is 500 pixels and the vertical pixel H _in is 300 pixels, four partial images having the horizontal pixel W _data of 192 pixels and the vertical pixel H _data of 144 pixels are generated. Two partial images having 116 pixels for the horizontal pixel W _{data and} 144 pixels for the vertical pixel H _data existing at the position in contact with the right side are generated.
In addition, two partial images having a horizontal pixel W _data of 192 pixels and a vertical pixel H _data of 12 pixels existing at a position in contact with the lower side of the input image are generated and exist at the lower right position of the input image. One partial image in which the horizontal pixel W _data is 116 pixels and the vertical pixel H _data is 12 pixels is generated.
As a result, a total of nine partial images are generated, the partial image existing at the upper left position of the input image is first selected as the data detection target image, and the remaining eight partial images are detected in the raster scan order. The target images are sequentially selected.

When receiving the data detection target image as shown in FIG. 3 from the data detection target image extraction unit 11, the bit detection target image extraction unit 12 divides the data detection target image to obtain one or more local regions. An image (a partial image of the data detection target image) is acquired.
Then, the bit detection target image extraction unit 12 sequentially selects partial images for detecting the digital watermark bits from one or more partial images, and sets the partial images as the bit detection target images. And it outputs to the bit detection part 15 (step ST22).

In the example of FIG. 3, the data detection target image is divided into square regions having a horizontal pixel W _bit of 24 pixels and a vertical pixel H _bit of 24 pixels, so that the horizontal pixel W _bit is 24 pixels at the maximum and the vertical pixel H _bit is A maximum of 48 partial images of maximum 24 pixels are generated.
The partial image existing at the upper left position of the data detection target image is first selected as the bit detection target image, and the remaining partial images are sequentially selected as the bit detection target image in the raster scan order.

When the bit detection target image extraction unit 12 selects a bit detection target image, the function value calculation unit 13 performs a function calculation for adding and subtracting the pixel value of each pixel constituting the bit detection target image. A function value f _v indicating the addition / subtraction result of the pixel value of the pixel is calculated, and the function value f _v is output to the bit detection unit 15 (step ST23).
In the example of FIG. 3, among the pixels constituting the bit detection target image, an area where dark gray pixels are gathered is an addition target area, and an area where light gray pixels are gathered is subtracted. The function value f _v is calculated by subtracting the pixel values of all the pixels in the subtraction target area while adding the pixel values of all the pixels in the addition target area as the target area.

That is, if the pixel value of the x-th pixel (1 ≦ x ≦ 24) from the left of the bit detection target image and the y-th pixel (1 ≦ y ≦ 24) from the top is f _bit (x, y), the function value f _v Is calculated as shown in equation (1) above.
However, since the horizontal size or vertical size of the bit detection target image is less than 24 and there is a nonexistent pixel, the pixel value f _bit (x, y) of such a pixel is set to “0” and the expression (1 ) Shall apply.

As described above, the function value calculation unit 13 performs the calculation of the function for adding and subtracting the pixel value of each pixel constituting the bit detection target image. The function value f _v of the function can take a range of values. Is a finite, among the possible values of the function value f _v of the function, the representative value representing bit 0 is 0 representative value KV ₀ , the representative value representing bit 1 is 1 representative value KV ₁ , null ( A representative value representing (null) is preset in the representative value setting unit 14 as a null representative value KV _null .
In the example of FIG. 4, the 0 representative value KV ₀ is “−750”, “−250”, “500”, “1000”, and the 1 representative value KV ₁ is “−1000”, “−500”, “250”, “750” and the null representative value KV _null are set to “0”.
The representative value setting unit 14 0 representative value _KV 0 is set to 1 representative value KV ₁ and null representative value _{KV null} is 0 representative value _KV 0, 1 representative set to the representative value setting unit 4 of FIG. 1 It is the same as the value KV ₁ and the null representative value KV _null .

When the function value calculation unit 13 calculates the function value f _v , the bit detection unit 15 includes the 0 representative value KV ₀ , 1 representative value KV _1, and null representative value KV _null set by the representative value setting unit 14. The representative value closest to the function value f _v is selected.
For example, if the function value f _v is “376”, 0 representative value KV ₀ is selected, and if the function value f _v is “374”, 1 representative value KV ₁ is selected.
If the function value f _v is “−126”, the 0 representative value KV ₀ is selected, and if the function value f _v is “−124”, the null representative value KV _null is selected.
If the function value f _v is “126”, one representative value KV ₁ is selected, and if the function value f _v is “124”, the null representative value KV _null is selected.

When the function value f _v is just between the 0 representative value KV ₀ and the 1 representative value KV ₁ (for example, f _v = 375), the function value f _v is equal to the 0 representative value KV ₀ and the null representative value KV _null . When it is just in the middle (for example, f _v = −125), when the function value f _v is just in the middle of one representative value KV ₁ and the null representative value KV _null (for example, f _v = 125), which representative value It is assumed that whether to select is set in advance.

When the bit detection unit 15 selects a representative value closest to the function value f _v , the bit detection unit 15 outputs a detection result corresponding to the representative value to the digital watermark data generation unit 16 (step ST24).
That is, when the 0 representative value KV ₀ is selected as the representative value closest to the function value f _v , the bit detection unit 15 sets “0” as a detection result indicating that bit 0 is embedded as a digital watermark bit. The detected amount is output to the digital watermark data generation unit 16.
When the bit detection unit 15 selects one representative value KV ₁ as the representative value closest to the function value f _v , the detection result indicating that bit 1 is embedded as a digital watermark bit is “1”. The detected amount is output to the digital watermark data generation unit 16.
When the null representative value KV _null is selected as the representative value closest to the function value f _v , the bit detection unit 15 sets the detection amount of “null” as the detection result indicating that no digital watermark bit is embedded. The data is output to the data generation unit 16.

As shown in FIG. 4, when the 0 representative value KV ₀ , the 1 representative value KV ₁ and the null representative value KV _null are set, the vicinity of each representative value on the number line of the function value f _v is When a digital watermark bit is detected, a range of function values for discriminating between bit 0, bit 1, and null is set.
As a result, not only an image that has not been edited at all after embedding a digital watermark bit, but also an image in which a small amount of noise is added to change the pixel value, the function value exceeds the boundary of the range. it can be detected precisely as f _v does not change significantly.
Therefore, it is possible to provide resistance to noise added by image compression or digital / analog conversion.

In the processes from step ST22 to ST24, all the bit detection target images are selected by the bit detection target image extracting unit 12, and the detection result of the digital watermark bits for all the bit detection target images from the bit detection unit 15 is digital watermark data. This is repeated until output to the generation unit 16 (step ST25).
In the processes from step ST21 to ST25, all the data detection target images are selected by the data detection target image extraction unit 11, and the electronic processing for all the bit detection target images included in all the data detection target images from the bit detection unit 15 is performed. This is repeated until the watermark bit detection result is output to the digital watermark data generation unit 16 (step ST26).

When the digital watermark data generation unit 16 receives the detection results of the digital watermark bits for all the bit detection target images from the bit detection unit 15, the digital watermark data generation unit 16 totals all the detection results and obtains the digital watermark data from the total result of the detection results. Generate (step ST27).
Specifically, digital watermark data is generated as follows.

In the digital watermark embedding device of FIG. 1, the embedding target bit extracting unit 6 applies each bit constituting 48-bit digital watermark data to each bit embedding target image in the data embedding target image. , Embedded in raster scan order.
Since the bit embedding target image is 24 pixels wide and 24 pixels long and the data embedding target image is 192 pixels wide and 144 pixels long, the digital watermark embedding device in FIG. 1 starts from the left end of the input image. It exists at the position of 24x + 1 pixel to 24x + 24 pixel (0 ≦ x ≦ 7) in the horizontal direction, and at the position of 24y + 1 pixel to 24y + 24 pixel (0 ≦ y ≦ 5) in the vertical direction from the upper end of the input image. [(X mod 8) +8 (y mod 6) +1] bits of the digital watermark data are embedded in the existing square area.

For this reason, in the digital watermark data generation unit 16, for each bit constituting the digital watermark data, the detection result of the digital watermark bit related to the bit detection target image at the position where the bit is embedded (detection amount = 0, 1 or null), the detection amount of “0” and the detection amount of “1” are totaled. If the number of detection amounts of “0” is larger than the number of detection amounts of “1”, The digital watermark bit embedded in the bit detection target image is determined to be bit 0.
On the other hand, if the number of detection amounts of “0” is not greater than the number of detection amounts of “1”, the digital watermark bit embedded in the bit detection target image is determined to be bit 1.
When the digital watermark data generating unit 16 determines the digital watermark bits embedded in all the bit embedding target images in the data embedding target image as described above, the digital watermark data generating unit 16 concatenates the final digital watermark bits. Digital watermark data is generated, and the digital watermark data is output.

As is apparent from the above, according to the second embodiment, the bit detection unit 15 sets the 0 representative value KV ₀ , 1 representative value KV _1, and null representative value KV _null set by the representative value setting unit 14. Among them, the representative value closest to the function value f _v obtained by the function value calculation unit 13 is selected, and if the representative value is 0 representative value KV ₀ , it indicates that bit 0 is embedded as a digital watermark bit. When the representative value is 1 representative value KV ₁ , the detection result indicating that bit 1 is embedded as the digital watermark bit is output to the digital watermark data generation unit 16. output to, if the representative value is null representative value KV _null, structure detection results indicating that the watermark bit is not embedded to output to the electronic watermark data generator 16 Since the present embodiment offers an advantage of being able to detect the electronic watermark embedded by the electronic watermark embedding apparatus in FIG.

That is, in the second embodiment, the bit detection unit 15 embeds not only a detection result indicating that 0 bit is embedded and a detection result indicating that 1 bit is embedded, but also a digital watermark bit. In the case where the watermark embedding image cannot be embedded unless the image quality of the bit embedding is accompanied by a large image quality degradation, the detection result indicating that the bit is not embedded can be output. It is not necessary to embed digital watermark bits by accepting image quality degradation (if the bit detection unit 15 cannot output a detection result indicating that no digital watermark bits are embedded, image quality degradation is accepted. Therefore, it is necessary to embed a digital watermark bit), and image quality deterioration can be suppressed.
In addition, the bit detection unit 15 adds up the detection amount of “0” and the detection amount of “1” among the detection results (detection amount = 0, 1, or null) of the digital watermark bit to detect the bit. Since the digital watermark bits embedded in the target image are finally determined, the detection performance of the digital watermark can be improved.

In the second embodiment, the data detection target image extraction unit 11 recognizes a data detection target image that is repeated in the input image, and the bit detection target image extraction unit 12 regularly includes the data detection target image. The digital watermark data can be detected from a plurality of bit detection target images in which the same digital watermark bits are embedded. And since the final detection result is estimated by the digital watermark data generation unit 16 summing up the detection amounts indicating the detection results related to the plurality of bit detection target images, the detection results related to some of the bit detection target images Even if there is an error, there is a high possibility that a correct detection result is finally obtained from the detection results of other bit detection target images.
As described above, the correct detection result is estimated for all the bits constituting the digital watermark data based on the detection results of the plurality of bit detection target images, so that correct digital watermark data can be obtained. Increases nature.

In the second embodiment, the bit detection unit 15 outputs a detection amount of “0”, “1”, or “null” as the detection result of the digital watermark bit, and the digital watermark data generation unit 16 However, although the number of detected amounts of “0” and the number of detected amounts of “1” for the same bit is totaled to finally determine the digital watermark bit, other configurations are possible.
For example, when the bit detection unit 15 outputs the detection amount, if the representative value closest to the function value f _v is 0 representative value KV ₀ , the function value f _v and 0 together with the detection amount of “0”. The absolute value of the difference Δ ₀ of the representative value KV ₀ is output.
Further, if the function value f _v closest representative value 1 representative value KV _1, "1" along with outputting the detected amount of, the function value f _v and 1 absolute difference delta ₁ of the representative value KV ₁ Output the value.
Further, if the closest representative value and the function value _{f v} is a null representative value _{KV null,} along with the detected amount of "null", the absolute value of the function value _{f v} and null representative value _{KV null} of the difference delta _null Output.

Electronic watermark data generator 16, the detection result for the same bit, the sum of the absolute value of the difference delta ₀ output together when the detected amount of "0" from the bit detector 15 is outputted, "1" detection amount is compared with the sum of the absolute value of the difference delta ₁ outputted together when it is output.
If the sum of the absolute values of the difference Δ ₀ is smaller than the sum of the absolute values of the difference Δ ₁ , the digital watermark data generation unit 16 determines that the digital watermark bit embedded in the bit detection target image is bit 0. .
On the other hand, if the sum of the absolute values of the difference Δ ₀ is larger than the sum of the absolute values of the difference Δ ₁ , the digital watermark bit embedded in the bit detection target image is determined to be bit 1.
In this case, even if the number of detection amounts “0” and the number of detection amounts “1” output from the bit detection unit 15 are the same, the detection result indicating the function value f _v close to the representative value is obtained. Since the evaluation is focused on, the detection performance can be improved.

As another configuration, the bit detection unit 15 outputs the function value f _v calculated by the function value calculation unit 13 to the digital watermark data generation unit 16 as a detection amount, and the digital watermark data generation unit 16 outputs the function value f _v . The electronic watermark data may be generated by evaluation.
As described above, the bit detection unit 15 and the digital watermark data generation unit 16 can have any configuration as long as the digital watermark data generation unit 16 appropriately processes the detection amount output from the bit detection unit 15. .

  In the second embodiment, the configuration of the data detection target image extraction unit 11, the bit detection target image extraction unit 12, and the function value calculation unit 13 includes the data embedding target image extraction unit 1 in the digital watermark embedding apparatus of FIG. Although the same configurations as those of the bit embedding target image extracting unit 2 and the function value calculating unit 3 have been shown, any one corresponding to the digital watermark data embedding method in the digital watermark embedding device of FIG. The configuration of the data embedding target image extraction unit 1, the bit embedding target image extraction unit 2, and the function value calculation unit 3 may be different.

In the second embodiment, the input image is an uncompressed luminance image (uncompressed monochrome image). However, the present invention is not limited to this. For example, the input image may be an uncompressed color image. Good.
If the input image is an uncompressed color image in the YCbCr format and the digital watermark embedding apparatus in FIG. 1 embeds digital watermark data in the Cr component of the uncompressed color image, the electronic watermark in FIG. The watermark detection apparatus extracts the Cr component of the uncompressed color image in the YCbCr format and performs the processing shown in the second embodiment, so that the digital watermark data can be detected appropriately.

In addition, when the obtained image is in the compressed RGB format, the Cr component of the uncompressed color image in the YCbCr format can be extracted by releasing the compression and converting the RGB format into the YCbCr format. After all, by executing the processing shown in the second embodiment, the digital watermark data can be detected appropriately.
In this way, the digital watermark detection device may be any configuration as long as it corresponds to the input image of the digital watermark embedding device assumed by the digital watermark.

Embodiment 3 FIG.
In the second embodiment, it has been described that the input image has not been edited such as cutting or scaling after the digital watermark data is embedded. However, the input image has been edited such as cutting, scaling, or rotation. Even if the image is an image, if appropriate preprocessing is performed to generate an input image of the digital watermark detection apparatus, the digital watermark data can be accurately detected as in the second embodiment.

FIG. 7 is an explanatory diagram showing an example of generation of an input image of the digital watermark detection apparatus.
FIG. 7 illustrates an example in which a part of an image in which digital watermark data is embedded is cut out and an input image of the digital watermark detection apparatus is generated from the part of the cut out image.
That is, the input image of the digital watermark detection apparatus is generated by padding the cut image with a single color with respect to the image that remains without being cut.

In addition to cropping the input image, even an image that has been edited, such as scaling or rotation, is corrected to the original image position by image alignment technology, and then the lost image area is padded with a single color. By doing so, an input image of the digital watermark detection apparatus can be generated.
A number of techniques for aligning an image whose position has shifted from the original image by editing such as cropping, scaling, and rotation have already been disclosed, and image processing professionals can use these techniques in FIG. It is easy to generate an input image to be given to the digital watermark detection apparatus.
Further, correction can be made while visually comparing the original image and the acquired image without using an advanced image alignment technique.

In the digital watermark detection apparatus of FIG. 5, when a padded input image is given, the bit detection unit 15 detects null (no embedding) from a lost area portion (area where padding has been performed). become.
This is because the padding is a monotone color and the function value calculation unit 13 calculates a function value f _v that is very close to the null representative value KV _null representing null.
In addition, any one of bit 0, bit 1, and null can be detected from the image portion that has not been lost as described above.

Finally, the digital watermark data generation unit 16 adds up the detection amounts “0”, “1”, and “null” to generate digital watermark data.
Alternatively, the lost region portion is excluded from the detection target, and the bit detection unit 15 calculates the detection amount only from the image portion that is not lost, and the digital watermark data generation unit 16 calculates the digital watermark data from these detection amounts. Generate.
In this way, by correcting the input image of the digital watermark detection apparatus in advance by the image alignment technique, even if the image has been subjected to processing such as clipping, enlargement / reduction, and rotation, the digital watermark data can be stored. Can be detected.

  1 data embedding target image extracting unit (image dividing unit, partial image selecting unit), 2 bit embedding target image extracting unit (image dividing unit, partial image selecting unit), 3 function value calculating unit (function value calculating unit), 4 representative value setting unit (representative value setting unit), 5 image quality deterioration degree evaluating unit (image quality deterioration degree calculating unit), 6 embedding target bit extracting unit (digital watermark bit extracting unit), 7 bit embedding unit (bit embedding) Means), 8-bit embedding target image combining unit (image combining unit), 9 data embedding target image combining unit (image combining unit), 11 data detection target image extracting unit (image dividing unit, partial image selecting unit), 12 Bit detection target image extraction unit (image division unit, partial image selection unit), 13 function value calculation unit (function value calculation unit), 14 representative value setting unit (representative value setting unit), 15 bit detection unit (digital watermark bit detection) hand Stage), 16 a digital watermark data generation unit (digital watermark data generation means).

Claims (7)

  Function value calculation means for obtaining a function value indicating a result of addition / subtraction of the pixel value of each pixel by executing a calculation of a function for adding / subtracting the pixel value of each pixel constituting the input image, and a value that the function value can take A representative value setting means for setting a representative value representing bit 0 as 0 representative value, a representative value representing bit 1 as 1 representative value, and a representative value representing null as a null representative value, and a digital watermark bit If the difference between the function value obtained by the function value calculation means and the zero representative value set by the representative value setting means is greater than or equal to a threshold value, the input is performed so that the difference is less than the threshold value. When bit embedding processing for editing the pixel value of each pixel constituting the image is performed and 1 is embedded as a digital watermark bit, the function value obtained by the function value calculating means and the representative value setting means are set. One If the difference between the table values is equal to or greater than the threshold value, bit embedding processing is performed to edit the pixel value of each pixel constituting the input image so that the difference is less than the threshold value, and the digital watermark bits are embedded. If not, if the difference between the function value obtained by the function value calculating means and the null representative value set by the representative value setting means is greater than or equal to a threshold value, the input image is configured so that the difference is less than the threshold value. A digital watermark embedding apparatus comprising: a bit embedding unit that performs a process of editing a pixel value of each pixel being performed. Image quality deterioration degree calculating means for executing a calculation of a function for digitizing the degree of deterioration of image quality that occurs when the pixel value of each pixel constituting the input image is edited, and outputting a numerical value indicating the degree of deterioration; Provided,
Even if the difference between the function value obtained by the function value calculating means and the representative value set by the representative value setting means is greater than or equal to the threshold value, the bit embedding means is capable of maintaining the numerical value output from the image quality degradation degree calculating means. Bit embedding processing is performed only when it indicates degradation within the range, and if the numerical value output from the image quality degradation degree calculating means indicates degradation exceeding an allowable range, bit embedding processing is not performed. The digital watermark embedding apparatus according to claim 1, wherein:   An image dividing unit that divides an input image to obtain a partial image that is an image of one or more local regions, and an electronic watermark bit is selected from one or more partial images obtained by the image dividing unit. A partial image selection unit that sequentially selects partial images to be embedded, and a function of adding / subtracting the pixel value of each pixel that constitutes the partial image selected by the partial image selection unit, so that the pixel value of each pixel A function value calculating means for obtaining a function value indicating the result of addition and subtraction, and among the values that can be taken by the function value, a representative value representing bit 0 is represented as 0 representative value, a representative value representing bit 1 is represented as 1 representative value, and Representative value setting means for setting a representative value representing null as a null representative value, and calculation of a function that quantifies the degree of image quality degradation that occurs when the pixel value of each pixel constituting the partial image is edited To show the degree of deterioration In the case of embedding 0 as the digital watermark bit and the image quality deterioration degree calculating means for outputting a value, the numerical value output from the image quality deterioration degree calculating means represents deterioration within an allowable range, and is obtained by the function value calculating means. If the difference between the function value and the 0 representative value set by the representative value setting means is greater than or equal to the threshold value, the pixel value of each pixel constituting the partial image is edited so that the difference is less than the threshold value. When bit embedding processing is performed and 1 is embedded as a digital watermark bit, the numerical value output from the image quality deterioration degree calculating means represents deterioration within an allowable range, and the function value obtained by the function value calculating means If the difference of one representative value set by the representative value setting means is greater than or equal to a threshold value, the pixel value of each pixel constituting the partial image is edited so that the difference is less than the threshold value. When the bit embedding process is performed and the digital watermark bit is not embedded, the numerical value output from the image quality deterioration degree calculating unit represents deterioration within an allowable range, and the function value obtained by the function value calculating unit and the above If the difference between the null representative values set by the representative value setting means is greater than or equal to the threshold value, a process of editing the pixel value of each pixel constituting the partial image is performed so that the difference is less than the threshold value. A digital watermark embedding apparatus comprising bit embedding means and image combining means for combining all partial images for which bit embedding processing has been completed by the bit embedding means.   When the numerical value output from the image quality deterioration degree calculating means represents deterioration exceeding the allowable range, the image quality deterioration degree calculating means performs an operation so that the function value obtained by the function value calculating means approaches the null representative value. 4. The digital watermark embedding apparatus according to claim 2, wherein a function to be calculated and a function to be operated by the function value calculation means are defined.   5. The digital watermark bit extracting means for extracting a digital watermark bit to be embedded from the digital watermark data and outputting the digital watermark bit to a bit embedding means. The electronic watermark embedding device according to any one of the preceding claims.   Function value calculation means for performing a function operation for adding / subtracting the pixel value of each pixel constituting the input image in which the digital watermark data is embedded to obtain a function value indicating the addition / subtraction result of the pixel value of each pixel; Among the possible values of the function value, a representative value representing bit 0 is set as 0 representative value, a representative value representing bit 1 is set as 1 representative value, and a representative value representing null is set as a null representative value. A representative value closest to the function value obtained by the function value calculating means is selected from the value setting means and the 0 representative value, 1 representative value, and null representative value set by the representative value setting means; If the representative value is 0 representative value, a detection result indicating that bit 0 is embedded as a digital watermark bit is output. If the representative value is 1 representative value, bit 1 is embedded as a digital watermark bit. Detected result indicating Outputs, if the representative value is null representative value, the electronic watermark detecting apparatus and an electronic watermark bit detecting means for outputting a detection result indicating that the watermark bit is not embedded.   Image dividing means for dividing an input image in which digital watermark data is embedded to obtain a partial image that is an image of one or more local areas, and one or more partial images obtained by the image dividing means A partial image selection means for sequentially selecting a partial image for detecting a digital watermark bit, and a function for adding and subtracting the pixel value of each pixel constituting the partial image selected by the partial image selection means. And a function value calculating means for obtaining a function value indicating the result of addition / subtraction of the pixel value of each pixel, and among the values that the function value can take, a representative value representing bit 0 is 0 representative value, and bit 1 is set A representative value setting means for setting a representative value to be represented as one representative value and a representative value representing null as a null representative value, and a zero representative value, one representative value, and a null representative value set by the representative value setting means. In the above function value calculation The representative value closest to the function value obtained by the stage is selected, and if the representative value is 0 representative value, a detection result indicating that bit 0 is embedded as a digital watermark bit is output, and the representative value is If the representative value is 1, a detection result indicating that bit 1 is embedded as a digital watermark bit is output. If the representative value is a null representative value, detection indicating that the digital watermark bit is not embedded. Watermark bit detection means for outputting the result, and watermark data generation for counting the detection results of all the partial images output from the watermark bit detection means and generating the watermark data from the result of the detection results And a digital watermark detection apparatus.

Images