JP2009253854A - Method, device, and program of embedding digital watermark, and computer readable recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a digital watermark embedding processing for making control appropriately without inverting a relationship between watermark intensity and image quality deterioration while suppressing an unintended change in brightness and color that occurs when a minute signal is subjected to color conversion. <P>SOLUTION: When watermark information is inputted, a watermark pattern is generated, which indicates an amount of change in the pixel value imparted to an original image when watermark information is embedded into the original image by digital watermark embedding, a conversion error caused by the color conversion of the watermark pattern is calculated, the conversion error is diffused by weighting the peripheral pixels, a watermark pattern after being subjected to the conversion is generated, and an image including a watermark is outputted after superimposing the watermark pattern subjected to the change, on the inputted image. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a digital watermark embedding method and apparatus, a program, and a computer-readable recording medium, and more particularly, a digital watermark technique that is a technique for embedding other sub information in a video content so as not to be perceived and reading the sub information. It is about.

  Today, digital watermark technology is used in content copyright protection / management systems, content-related service provision systems, and the like.

  When distributing content such as images and videos, there is a method of using digital watermark technology that embeds other information so that it cannot be perceived in the content for purposes such as content identification / management, copyright protection / management, and related information provision . In particular, in the digital watermark technology for image / video content, a method based on a minute change of a pixel value is common (see, for example, Non-Patent Document 1).

  In addition, there is a method of avoiding a luminance component with high visual sensitivity and finely changing a color component such as a color difference in order to suppress image quality deterioration due to digital watermark embedding (see, for example, Non-Patent Document 2).

  Generally, a pixel value digitally expressed is, for example, a value quantized to 8 bits, that is, 256 values, and digital watermark embedding is a minute change of the pixel value. For example, consider 24-bit color data in which color image data is represented by 8 bits for each of RGB. The conversion from RGB data to luminance / color difference signal is, for example, a conversion formula defined by CCIR601 etc.

等を用いて相互に変換可能である。今、画質劣化を抑えるために視覚感度の高い輝度Ｙを避け、例えば、Ｃｂ成分を微小量△ｃｂだけ変更して電子透かしを埋め込むことを考える。この場合、例えば、△ｃｂ＝±１であるとすると、埋め込みによって生じるＲＧＢの変化量△ｒ、△ｇ、△ｂは（式２）からそれぞれ
△ｒ＝０，
△ｇ＝±0.336，
△ｂ＝±1.732
となるが、ここで量子化による丸め誤差を考えると（四捨五入則を用いるとして）、
△Ｒ＝△Ｇ＝０，
△Ｂ＝±２となる。すなわち、この場合はＣｂ成分の微小変更量はＢ成分にのみ反映される。このとき量子化後の△ｒ，△ｇ，△ｂをＹＣｂＣｒ色空間に変換すると、
△ｙ＝±0.114×2，
△ｃｂ＝±0.511×2，
△ｃｒ＝±0.083×2
となり、本来変更を施さなかったＹ及びＣｒ成分値が変化していることがわかる。特に、輝度成分Ｙが変化することにより、電子透かし埋め込みによる視覚劣化が目立ち易くなるという問題があった。

Etc. can be converted into each other. Now, in order to suppress image quality degradation, it is considered to avoid the luminance Y having high visual sensitivity, for example, to embed a digital watermark by changing the Cb component by a minute amount Δcb. In this case, for example, assuming that Δcb = ± 1, RGB variations Δr, Δg, and Δb caused by embedding are expressed by (r = 0),
△ g = ± 0.336,
Δb = ± 1.732
However, when considering rounding error due to quantization (using the rounding rule),
△ R = △ G = 0
ΔB = ± 2. That is, in this case, the minute change amount of the Cb component is reflected only in the B component. At this time, if Δr, Δg, Δb after quantization is converted into the YCbCr color space,
△ y = ± 0.114 × 2,
△ cb = ± 0.511 × 2,
△ cr = ± 0.083 × 2
Thus, it can be seen that the Y and Cr component values that were originally not changed have changed. In particular, there is a problem that visual deterioration due to digital watermark embedding becomes conspicuous when the luminance component Y changes.

Further, from the conversion formula of (Equation 2), even if a change of | Δcb | <0.5 / 1.732 is given, a rounding error due to quantization causes Δr = Δg = Δb = 0, and the image is completely changed. Since it is not given, it can be seen that the digital watermark is not embedded. In addition, when 0.5 / 1.732 ≦ | Δcb | <1.5 / 1.732, Δr = Δg = 0, Δb = ± 1, etc., and the change amount Δcb due to digital watermark embedding is continuously specified. However, the amount of change of the actual image value becomes discrete due to the quantization. In addition, when Δcb = ± 3,
Δr = 0,
△ g = ± 0.336 × 3
△ b = ± 1.732 × 3
When quantizing,
Δr = 0,
Δg = ± 1,
Δb = ± 5
When the quantized ΔrΔgΔb is converted into the YCbCr color space,
△ y = ± (-0.587 × 1 + 0.114 × 5) = ± 0.017
Therefore, the amount of change is smaller than Δy = ± 0.114 × 2 when Δcb = ± 1. That is, even if the amount of change in the Cb component is reduced, the amount of change in the Y component, which is the main cause of image quality deterioration, may increase due to quantization. As can be seen from the above, in the prior art, it is difficult to finely and accurately control the trade-off balance between image quality degradation and durability by embedding strength, which is an important point in the operation of digital watermark technology. There was a problem.

In summary,
(1) Even if only the color component such as the Cb component is changed while avoiding the Y component in order to suppress the image quality deterioration, the final post-embedding pixel value (RGB value) can only take a positive number. Ingredients fluctuate;
(2) Since the final post-embedding pixel value (RGB value) can only take an integer, the intensity cannot be finely adjusted;
(3) Since the image quality degradation may increase even if the intensity is lowered, the tolerance / image quality trade-off cannot be appropriately controlled by the intensity;
There was a problem.

On the other hand, as a solution to the problem of the influence on the minute change amount when converting the color format of the digital image, there is a method of changing the watermark pattern so that the change amount of the color component other than the quantization error and the target color component becomes small. It has been proposed (see, for example, Patent Document 1).
Nakamura, Yamamoto, Kitahara, Miyatake, Katayama, Anno: "Digital watermark embedding method and apparatus and program" JP 2008-17289 Nakamura, Yamamoto, Kitahara, Miyatake, Katayama, Anno, Sonehara: "Mobile watermarking for video with real-time detection", Vol. 36, No. 4, pp. 426-434 (2007) Takao Nakamura, Hiroshi Ogawa, Atsuki Tomioka, and Youichi Takashima: "Improved Digital Watermark Robustness against Translation and / or Cropping of an Image Area", IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences, Vol.E83-A, No. 1, pp. 68-76 (2000).

  However, it is necessary to set the reference change amount in advance in the implementation of the method of changing the watermark pattern so that the change amount of the color component other than the conventional quantization error and the target color component becomes small. There are countless candidates for the amount of reference change, and the selection method is also described, but there are actually candidates that are difficult to judge whether they are good or bad, so it may not be possible to judge without creating an image. It can happen many times. Also, since it takes time to investigate these settings in advance, it is difficult to say that it is a simple method.

  The present invention has been made in view of the above points, and appropriately suppresses the relationship between the watermark strength and the image quality deterioration without reversing, while suppressing unintended luminance and color changes that occur when color conversion is performed on a minute signal. It is an object of the present invention to provide a digital watermark embedding method and apparatus and program capable of realizing a controllable digital watermark embedding process, and a computer-readable recording medium.

  FIG. 1 is a diagram for explaining the principle of the present invention.

The present invention (Claim 1) is a digital watermark embedding method for embedding a digital watermark in a color image,
In the watermark pattern generating means, when watermark information is input, a watermark pattern generating step (step for generating a watermark pattern representing a change amount of a pixel value given to the original image by embedding the watermark information in the original image with the watermark information) 1) and
A conversion error calculating step (step 2) of calculating a conversion error due to color conversion of the watermark pattern in the conversion error calculating unit of the watermark pattern changing unit;
A conversion error diffusion step (step 3) for generating a converted watermark pattern by weighting and diffusing the conversion error to surrounding pixels in the conversion error diffusion means of the watermark pattern changing means;
The watermark pattern superimposing means performs a watermark pattern weight step (step 4) of superimposing the changed watermark pattern on the input original image and outputting a watermarked image.

Further, the present invention (Claim 2), in the conversion error calculation step (Step 2),
Color-convert the pixel value of the input watermark pattern,
Quantize the color converted value,
Convert the quantized value to the same color space as the input watermark pattern,
A difference between the pixel value of the input watermark pattern and the pixel value of the watermark pattern after color conversion is defined as a conversion error.

In the conversion error calculation step (step 2), the present invention (claim 3)
The error is calculated with a different block size for each color component.

In the conversion error diffusion step (step 3), the present invention (claim 4)
The conversion error calculated in the conversion error calculation step (step 2) is diffused with a different block size for each color component.

  FIG. 2 is a principle configuration diagram of the present invention.

The present invention (Claim 5) is an electronic watermark embedding apparatus for embedding an electronic watermark in a color image,
When the watermark information is input, a watermark pattern generation unit 110 that generates a watermark pattern representing a change amount of a pixel value given to the original image by embedding the watermark information in the original image;
Conversion error calculation means 121 for calculating a conversion error due to color conversion of the watermark pattern;
A conversion error diffusion means 122 for weighting and diffusing the conversion error with respect to surrounding pixels to generate a converted watermark pattern;
Watermark pattern weighting means for superimposing the changed watermark pattern on the input original image and outputting a watermarked image.

Further, the present invention (Claim 6), in the conversion error calculation means 121,
Color-convert the pixel value of the input watermark pattern, quantize the color-converted value, color-convert the quantized value to the same color space as the input watermark pattern, and convert the pixel value of the input watermark pattern Means for converting the difference between the color-converted watermark pattern and the pixel value to a conversion error;

Further, the present invention (Claim 7) is provided in the conversion error calculation means 121,
Means for calculating an error with a different block size for each color component;

Further, according to the present invention (claim 8), in the conversion error diffusion means 122,
Means for diffusing the conversion error calculated by the conversion error calculation means 121 with a different block size for each color component;

  The present invention (Claim 9) is a digital watermark embedding program for causing a computer to function as each means constituting the digital watermark embedding apparatus according to any one of claims 5 to 8.

  As described above, according to the present invention, electronic signals that can be appropriately controlled without reversing the relationship between watermark strength and image quality deterioration while suppressing unintended luminance and color changes that occur when color-converting a minute signal. A watermark embedding process can be realized.

  Also, in the present invention, the image quality when the entire image is viewed as a macro is obtained by error-diffusing the pixel value error when the input watermark pattern is color-converted and the image is restored again by color conversion to surrounding pixels. Can be improved.

  In addition, since the conversion error of the watermark pattern is assigned to the peripheral pixels, when the entire image is viewed as a macro, it is possible to control the intended trade-off between image quality degradation and durability depending on the embedding strength.

  Furthermore, if the error diffusion method is defined, it is not necessary to set parameters in advance, so that extremely simple mounting is possible.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  First, the outline of the present invention will be described.

  In order to embed sub-information in an image by digital watermark technology, it is effective to superimpose a watermark pattern generated from sub-information on color difference information Cb, Cr having lower sensitivity than luminance information Y having high visual sensitivity. It has been known. However, if the image to be embedded is three colors of RGB, even if an attempt is made to insert a watermark pattern only in the color difference information Cb, Cr, since the RGB gradations of the RGB image are discrete, the watermark pattern is changed to Y, Cb, Cr. When converting from RGB to RGB and quantizing, a quantization error occurs, and as a result, the luminance information Y also varies. Therefore, in the present invention, a watermark pattern is created by diffusing the generated error on the watermark pattern using the error diffusion method, the watermark pattern is converted from Y, Cb, Cr to RGB, and the image is superimposed.

  FIG. 3 shows a configuration of a digital watermark embedding apparatus according to an embodiment of the present invention.

  The digital watermark embedding apparatus 100 shown in FIG. 1 includes a watermark pattern generation unit 110, a watermark pattern change unit 120, and a watermark pattern superimposition unit 130.

  When the watermark information is input, the watermark pattern generation unit 110 outputs a watermark pattern having the same size as the image to the watermark pattern change unit 120.

  The watermark pattern changing unit 120 changes the strength of the watermark pattern in which the quantization error is diffused by a method to be described later, and outputs the changed watermark pattern to the watermark pattern superimposing unit 130.

  The watermark pattern superimposing unit 130 superimposes the changed watermark pattern on the image and outputs a watermarked image.

  FIG. 4 shows a configuration of the watermark pattern changing unit in the embodiment of the present invention.

  The watermark pattern changing unit 120 includes a conversion error calculation unit 121 and a conversion error diffusion unit 122.

  The conversion error calculation unit 121 performs color conversion from the color pattern system of the watermark pattern to the color system of the image, quantization, and the color from the color system of the image to the color system of the watermark pattern. A conversion error which is a difference between the watermark pattern obtained by the conversion and the original watermark pattern is calculated and output.

  When the conversion error and the original watermark pattern are input, the conversion error diffusion unit 122 converts the conversion error for each pixel of the original watermark pattern to the color of the peripheral pixel of the pixel and the conversion error at the pixel. A converted watermark pattern is calculated and output by performing error diffusion by adding to the product multiplied by the weight corresponding to the pixel.

  FIG. 5 is a flowchart of processing of the digital watermark embedding apparatus according to the embodiment of the present invention.

  Step 110) When the watermark information is input, the watermark pattern generation unit 110 generates a watermark pattern representing the amount of change in pixel value to be given to the original image by embedding the watermark information in the original image.

  Step 120) In the watermark pattern changing unit 120, when the output changed watermark pattern is viewed as a macro, an error of the watermark embedding target color component due to quantization generated when the input watermark pattern is superimposed on the original image is detected. The changed watermark pattern is generated by changing the input watermark pattern so as to reduce the number.

  Step 130) When the original image is input from the outside in the watermark pattern superimposing unit 130 and the changed watermark pattern is input from the watermark pattern changing unit 120, the changed watermark pattern is superimposed on the original image, and the watermarked image is displayed. Is generated and output.

  Next, the processing of the watermark pattern changing unit in step 120 will be described.

  In the YCbCr color space, the Y component represents the luminance component, and all images of the image are sampled. On the other hand, the Cb and Cr components are color difference components, and the number of samplings in the x and y directions differs depending on the format. Here, the sampling intervals in the x and y directions of the Cb and Cr components are Sx and Sy. In a commonly used YCbCr color space format, the number of samplings in the x and y directions with the Cb component is the same as the number of samplings in the x and y directions with the Cr component. FIG. 6 shows a sampling ratio of a general YcbCr color space and an example of the image. For example, in the case of YUV420, when the Y component is sampled by 2 × 2 blocks, the Cb and Cr components are sampled by 1 × 1 block, so Sx = 2 and Sy = 2.

  FIG. 7 is a flowchart of the process of the watermark pattern changing unit in the embodiment of the present invention.

  Step 121) The watermark pattern changing unit 120 raster scans the input watermark pattern in units of blocks of Sx × Sy size according to the Cb and Cr components.

  Step 122) The Cb and Cr components are the target block itself or the target pixel.

  Step 123) The watermark pattern changing unit 120 updates the target pixel of the Y component in the raster scan order in each block. At this time, in the Cb and Cr components, the current target block itself is the target pixel.

  Step 124) The conversion error calculation unit 121 of the watermark pattern changing unit 120 calculates the Y component conversion error in the scan order of the Y component in the block.

  Step 125) The error of the Y component is diffused to the Y component of the adjacent pixel.

  Step 126) If the process has been performed for all Y component pixels in the block, the process proceeds to step 127. If the process has not been completed for all Y component pixels, the process returns to step 123.

  Step 127) The conversion error calculation unit 121 calculates a conversion error of the block.

  FIG. 8 shows the conversion error calculation method for YUV444 (Sx = 1, Sy = 1). Here, in the case of YUV444, since the block size is 1 × 1, it is not actually necessary to consider the Cb and Cr components separately, and they are handled simultaneously as shown in FIG. FIG. 9 is a flowchart of the process of the conversion error calculation unit in one embodiment of the present invention.

  The numbers in squares shown in FIG. 8 correspond to the numbers in parentheses below and the numbers in parentheses in the flowchart of FIG.

First, the conversion error calculation unit 121 converts each pixel value of YCbCr expressed by a real value into RGB of the real value using the above-described (Equation 1) (1), and then converts the real value to an integer value. Round to (2). After that, the integer-valued bad RGB pixel value is reconverted into a real value Y′Cb′Cr ′ by (Equation 2) (3). The pixel value YCbCr of the watermark pattern input here is different from the re-converted pixel value Y′Cb′Cr ′, and the difference is converted into a conversion error E (E _y , E _Cb , E _Cy ) = ( YY ′, Cb—Cb ′, Cr—Cr ′) (4).

  Step 128) The conversion error diffusion unit 122 acquires the conversion error E from the conversion error calculation unit 121, and diffuses the error to surrounding pixels of the target pixel. In FIG. 8, the error is obtained by Floyd-Steinberg's method in which the error is diffused by weighting 7/16, 3/16, 5/16, and 1/16 times to the right, lower left, lower, and lower right sides of the pixel of interest. Diffusion is performed, but various methods such as Stuki, Sierra, and JajuNi may be used.

  On the other hand, when either Sx or Sy is larger than 1, it is necessary to handle the Y component and the Cb, Cr component separately. FIG. 10 shows the case of YUV420 (Sx = 2, Sy = 2).

  The Y component conversion error is obtained in the order of raster scanning of the Y component in the current block of interest. At this time, the Cb and Cr component values are the same in the target block, and the processing is the same as (1) to (4) in FIGS. A method of using the average value of the Y components in the block and the like when calculating the errors of the Cb and Cr components after diffusing all the errors of the Y components in the block can be considered.

  As an error diffusion method in this case, there is a problem if a method of diffusing an error to a pixel obliquely behind the pixel scanning direction is used. For example, in the Floyd-Steinberg method, it is necessary to diffuse the pixel of interest to the lower left pixel, but for the lower right pixel in the 2 × 2 block, the error from the upper left pixel of the right adjacent block Needs to be spread. However, since the processing is performed in units of blocks of Cb and Cr components, there is a problem in that after the error diffusion processing of the target block is completed, diffusion processing from the pixels is further received, and unintended errors are accumulated.

  Since such a problem occurs when Sy ≧ 2, when YUV420 or YUV410 is used, the error of the target pixel is obliquely diffused only in the three directions of the right, lower, and lower right sides of the target pixel. It is conceivable to use an error diffusion method that does not diffuse errors to the back side.

  Step 129) If the process has been completed for all the pixels, the process is terminated; otherwise, the process proceeds to step 121.

  In the above embodiment, the digital watermark embedding into the Cb component has been described. Of course, the Cr component may be used, and even in other display systems, the block size may be changed for each color component, and error calculation. It is applicable if diffusion is performed.

  In the above embodiment, the digital watermark on the still image has been described. However, when considering the digital watermark on the moving image as in Non-Patent Document 1, each frame image or field image of the moving image is stopped. By considering the present invention and applying the present invention, it can also be realized in a moving image.

  The present invention is not limited to application to the above-described example of the digital watermark method. For example, as in Non-Patent Document 1, a watermark is embedded in each frame of a moving image, or as in Non-Patent Document 2, a watermark pattern is not on a block but on a texture. It doesn't matter.

  Furthermore, in the above embodiment, the Cb component of the YCbCr color system is embedded in order to avoid image quality deterioration. However, this is used as a Cr component or any other color system component. Etc. can be arbitrarily selected.

  It is possible to construct the operation of each component of the digital watermark embedding apparatus 100 in the above embodiment as a program and install it in a computer used as the digital watermark embedding apparatus or distribute the network.

  Further, the constructed program can be stored in a portable storage medium such as a hard disk, a flexible disk, or a CD-ROM, and can be installed or distributed in a computer.

  The present invention is not limited to the above-described embodiment, and various modifications and applications can be made within the scope of the claims.

  The present invention is applicable to a digital watermark embedding technique for still images and moving images.

It is a figure for demonstrating the principle of this invention. It is a principle block diagram of this invention. It is a block diagram of the digital watermark embedding apparatus in one embodiment of this invention. It is a block diagram of the watermark pattern change part in one embodiment of this invention. It is a flowchart of the process of the digital watermark embedding apparatus in one embodiment of this invention. It is a sampling ratio and image example for every general YCbCr color space. It is a flowchart of the process of the watermark pattern change part in one embodiment of this invention. It is a figure for demonstrating the process of the conversion error calculation part and error diffusion part in one embodiment of this invention. It is a flowchart of the process of the conversion error calculation part in one embodiment of this invention. It is a figure for demonstrating the process of the conversion error calculation part and error diffusion part in one embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Digital watermark embedding apparatus 110 Watermark pattern production | generation means, Watermark pattern production | generation part 120 Watermark pattern change part 121 Conversion error calculation means, Conversion error calculation part 122 Conversion error diffusion means, Conversion error diffusion part 130 Watermark pattern superimposition means, Watermark pattern superimposition part

Claims (10)

An electronic watermark embedding method for embedding an electronic watermark in a color image,
In the watermark pattern generation means, when watermark information is input, a watermark pattern generation step for generating a watermark pattern representing a change amount of a pixel value given to the original image by embedding the watermark information in the original image;
A conversion error calculating step of calculating a conversion error due to color conversion of the watermark pattern in the conversion error calculating means of the watermark pattern changing means;
In the conversion error diffusion means of the watermark pattern changing means, a conversion error diffusion step of weighting and diffusing the conversion error with respect to surrounding pixels to generate a converted watermark pattern;
In a watermark pattern superimposing means, a watermark pattern weight step for superimposing the changed watermark pattern on the input original image and outputting a watermarked image; and
An electronic watermark embedding method comprising: In the conversion error calculation step,
Color-convert the pixel value of the input watermark pattern;
Quantize the color converted value,
Color-convert the quantized value into the same color space as the input watermark pattern,
2. The digital watermark embedding method according to claim 1, wherein a difference between a pixel value of the input watermark pattern and a pixel value of the watermark pattern after color conversion is used as a conversion error. In the conversion error calculation step,
3. The digital watermark embedding method according to claim 2, wherein the error is calculated with a different block size for each color component. In the conversion error diffusion step,
The digital watermark embedding method according to claim 1, wherein the conversion error calculated in the conversion error calculation step is diffused with a different block size for each color component. An electronic watermark embedding device for embedding an electronic watermark in a color image,
A watermark pattern generating means for generating a watermark pattern representing a change amount of a pixel value given to the original image by embedding the watermark information in the original image when the watermark information is input;
Conversion error calculation means for calculating a conversion error due to color conversion of the watermark pattern;
A conversion error diffusion means for weighting and diffusing the conversion error with respect to surrounding pixels to generate a converted watermark pattern;
Watermark pattern weighting means for superimposing the changed watermark pattern on the input original image and outputting a watermarked image;
A digital watermark embedding apparatus comprising: The conversion error calculation means includes
The input pixel value of the watermark pattern is color-converted, the color-converted value is quantized, the quantized value is color-converted to the same color space as the input watermark pattern, and the input watermark pattern 6. The digital watermark embedding apparatus according to claim 5, further comprising means for converting a difference between the pixel value and the pixel value of the color-converted watermark pattern as a conversion error. The conversion error calculation means includes
7. The digital watermark embedding apparatus according to claim 6, further comprising means for calculating an error with a different block size for each color component. The conversion error diffusion means includes
6. The digital watermark embedding apparatus according to claim 5, further comprising means for diffusing the conversion error calculated by the conversion error calculation means with a different block size for each color component.   9. A digital watermark embedding program for causing a computer to function as each means constituting the digital watermark embedding apparatus according to claim 5.   A computer-readable recording medium storing the digital watermark embedding program according to claim 9.

Images