JP2005109627A - Method and device for embedding additional information, method and device for detecting additional information and computer program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and an apparatus for embedding additional information into a halftone image such that it can be detected surely even in an environment susceptible to noise incident to output from a printer. <P>SOLUTION: A continuous tone image is inputted (S100) and a noise mask including additional information is created (S110). A threshold matrix is modulated based on that noise mask (S120). The continuous tone image is binarized using a modulated threshold matrix (S130). The noise mask can be made by performing image conversion of a noise image formed of noise signal by orthogonal function expansion, imparting additional information to a noise conversion image thus obtained and then performing reverse image conversion. A binarized image embedded with additional information can be detected by performing image conversion through orthogonal function conversion. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a method and apparatus for embedding additional information in digital data such as image data and audio data, and a method for detecting additional information from digital data in which additional information is embedded, and more specifically, a binarized image. The present invention relates to a method and apparatus for embedding additional information, and a method for detecting additional information.

  In recent years, with the spread of personal computers and the development of information communication technologies such as the Internet, copyrighted works such as music, photos, videos, and documents can be easily created as digital data. Such digitized works are problematic from the viewpoint of copyright protection because they can be easily duplicated and altered. As a technique for solving this problem, a so-called digital watermarking technique in which desired additional information is embedded in digital data such as audio data and image data so as not to be perceptually discerned is drawing attention. This is an attempt to appropriately protect a digitized work by embedding the author's name and other copyright information as a digital watermark.

  As an example of the digital watermark technique, for example, Patent Document 1 below discloses a method of embedding additional information when halftoning an original image. Patent Document 2 below discloses a method of embedding a watermark in a perceptually important component of a frequency spectrum image obtained by Fourier transforming data to be watermarked. Patent Document 3 below discloses a method of performing orthogonal transformation on the entire data to be embedded with watermark information to obtain a transform coefficient, and embedding the watermark information in the transform coefficient. Patent Document 4 below discloses that the real number component and imaginary number component of each coefficient value of the watermark coefficient matrix that is a complex matrix are independently changed from the key information and the digital watermark, and the watermark sequence matrix after the change is subjected to discrete inverse Fourier transform. A method of embedding a digital watermark by adding the obtained watermark pattern to an original image in a tile shape is disclosed.

  By the way, in the method of adding additional information to the real space of image data as in Patent Document 1 below, when reproducing the additional information, it is necessary to refer to the position where the additional information is embedded, and the output resolution and input Processing such as matching the resolution and scaling according to the scale ratio was necessary. For this reason, there is a problem that if the resolution of the image in which the additional information is embedded is changed, the additional information cannot be reproduced.

  Further, as described in Patent Document 2 or 3 below, in the method of embedding additional information by performing Fourier transform or the like on the original image data, even when the same additional information is embedded in a plurality of original images, the Fourier information is added each time. Since it is necessary to perform conversion or the like, there is a problem that it takes time to generate a large amount of images in which the same additional information is embedded. In Patent Document 3 below, since the original image in which the additional information is embedded is limited to digital data, it is difficult to detect information existing in the Fourier space when conversion processing to analog such as printing or scanner is performed. There is also a problem of becoming.

  In Patent Document 4 below, a digital watermark is embedded by independently modulating a real part and an imaginary part in a frequency space. For this reason, when an image with an embedded digital watermark is output by an actual output device, for example, an electrophotographic machine or a printing machine, and the output image is read by using an input device such as a scanner, the image read by the scanner When detecting a digital watermark from the image, the deterioration becomes large, and it may be difficult to accurately reproduce the added information. In addition, when a method of adding a watermark pattern when embedding a digital watermark is used, if there is information on the added pattern, the watermark pattern may be removed by subtracting the information. When using it as a signature, there arises a problem that the signature cannot be obtained.

  By the way, in a display device that can express only binary density values, such as a facsimile, a monochrome printer, or a monochrome digital copier, a halftone image (binarized image) is obtained by performing image processing such as dithering on a continuous tone original image. It is converted and output. There is a demand for a technique for embedding a digital watermark in such a halftone image.

JP 2001-119562 A JP-A-9-191394 JP 2003-32473 A JP 2000-287073 A

  The present invention has been made in order to solve the problems based on the prior art, and an additional information embedding method and an additional information embedding apparatus capable of reliably detecting additional information even when output by an output device. The purpose is to provide.

  It is another object of the present invention to provide an additional information embedding method, an additional information embedding device, and an embedded information detection method that can reliably extract added information even if the output resolution or input resolution is unknown. To do.

  It is another object of the present invention to provide an optimum additional information embedding device which is a display device capable of expressing only binary density values, such as a facsimile, a monochrome printer, and a monochrome digital copier, and which embeds additional information.

In order to achieve the above object, according to the first aspect of the present invention,
An additional information embedding method for generating a binarized image in which additional information is embedded by performing binarization processing using a threshold matrix on an original image,
A noise image is generated using a noise signal, and the additional information is added to the noise conversion image obtained by performing image conversion by orthogonal function expansion on the noise image, and then the noise conversion is performed by performing a reverse conversion to the image conversion. Generating a mask; and
Modulating the threshold matrix based on the noise mask;
And binarizing the original image based on the modulated threshold matrix. The method of embedding additional information is provided.

  In the additional information embedding method according to the first aspect of the present invention, the addition of additional information to the noise-converted image includes components at each pixel position of the additional information image obtained by representing the additional information as an image, and It is preferable to multiply the component of the noise conversion image corresponding to each pixel position.

  In the present invention, the orthogonal function expansion means expansion using a defined orthogonal function in a predetermined section, and the orthogonal function means that the inner product or scalar product of the functions becomes zero. . The orthogonal function expansion is one selected from the group consisting of, for example, discrete cosine transform, Walsh Hadamard transform, discrete Fourier transform, Fourier transform, discrete sine transform, Haar transform, slant transform and Karhunen / Labe transform Is preferred.

In the additional information embedding method of the present invention, the step of modulating the threshold value matrix includes:
When the threshold value at a predetermined coordinate position of the threshold value matrix is T, the noise magnitude of the noise mask at the coordinate position is Δ, and k is a constant,
Preferably, the threshold matrix is modulated by replacing the threshold at a predetermined coordinate position of the threshold matrix with (1 + k · Δ) × T.

  Alternatively, the threshold at a predetermined coordinate of the threshold matrix is replaced with a threshold of a threshold matrix existing at a coordinate position that is separated in a certain direction by a distance obtained by multiplying a noise magnitude of the noise mask at the coordinate position by a constant. It is preferable to modulate the threshold matrix.

  According to a second aspect of the present invention, there is provided a method for detecting additional information from a binary image in which additional information is embedded using the additional information embedding method according to the first aspect of the present invention. There is provided a method for detecting additional information, characterized in that additional information is detected by performing image transformation by orthogonal function expansion on the digitized image.

The method for detecting additional information according to the second aspect of the present invention further includes a step of acquiring key information representing information on orthogonal function expansion used when the additional information is embedded.
It is preferable that image conversion by the orthogonal function expansion is performed on the binarized image based on the acquired key information to visualize the image converted image.

  In the additional information detection method according to the second aspect of the present invention, the key information includes a type of the orthogonal function expansion and a basis function order assignment method in the orthogonal function expansion at the time of image conversion by the orthogonal function expansion. It is preferable that the above information.

According to the third aspect of the present invention, there is provided an additional information embedding device for generating a binary image in which additional information is embedded by performing binarization processing using a threshold matrix on an original image. And
A noise image is generated using a noise signal, and the additional information is added to the noise conversion image obtained by performing image conversion by orthogonal function expansion on the noise image, and then the noise conversion is performed by performing a reverse conversion to the image conversion. Means for generating a mask;
Means for modulating the threshold matrix based on the noise mask;
There is provided an additional information embedding device comprising means for binarizing the original image based on the modulated threshold value matrix.

  In the additional information embedding device according to the third aspect of the present invention, the addition of additional information to the noise-converted image includes components at each pixel position of the additional information image obtained by representing the additional information as an image, It is preferable to multiply the component of the noise conversion image corresponding to the pixel position.

  The orthogonal function expansion is one selected from the group consisting of discrete cosine transform, Walsh Hadamard transform, discrete Fourier transform, Fourier transform, discrete sine transform, Haar transform, slant transform, and Karhunen / Loeve transform. Is preferred.

  In the additional information embedding device according to the third aspect of the present invention, the means for modulating the threshold value matrix sets T as a threshold value at a predetermined coordinate position of the threshold value matrix, and sets the noise magnitude of the noise mask at the coordinate position. It is preferable that the threshold value matrix is modulated by substituting the threshold value at a predetermined coordinate position of the threshold value matrix with (1 + k · Δ) × T, where Δ is a constant and k is a constant.

  Alternatively, the step of modulating the threshold matrix exists at a coordinate position separated in a fixed direction by a distance obtained by multiplying the threshold value at a predetermined coordinate of the threshold matrix by a constant times the magnitude of the noise of the noise mask at the coordinate position. It is preferable to modulate the threshold value matrix by replacing the threshold value matrix with a threshold value.

  According to a fourth aspect of the present invention, there is provided an apparatus for detecting additional information from a binary image in which additional information is embedded by the additional information embedding device according to the third aspect of the present invention, wherein the binary image There is provided an additional information detecting apparatus characterized by having means for performing image conversion by orthogonal function expansion.

  The additional information detecting apparatus according to the fourth aspect of the present invention further includes means for acquiring key information representing information on orthogonal function expansion used when embedding the additional information, based on the acquired key information. It is preferable to have means for performing image conversion by the orthogonal function expansion on the binarized image and visualizing the image converted image.

  In the additional information detection apparatus of the present invention, it is preferable that the key information is a type of the orthogonal function expansion and assignment of a basis function in the orthogonal function expansion to each point in a two-dimensional space.

  According to a fifth aspect of the present invention, there is provided a system comprising an additional information embedding device according to the third aspect of the present invention and an additional information detection device according to the fourth aspect of the present invention.

According to the sixth aspect of the present invention, an addition for causing a computer to execute a step of generating a binarized image in which additional information is embedded by performing binarization processing using a threshold matrix on the original image. An information embedding program,
In the computer,
A noise image is generated using a noise signal, and the additional information is added to the noise conversion image obtained by performing image conversion by orthogonal function expansion on the noise image, and then the noise conversion is performed by performing a reverse conversion to the image conversion. A procedure for performing a step of generating a mask;
Performing a step of modulating the threshold matrix based on the noise mask;
There is provided an additional information embedding program for executing a process including a step of executing a step of binarizing the original image based on the modulated threshold matrix.

  In the additional information embedding method of the present invention, the additional information is binarized because the additional information is given to the noise transformed image obtained by performing image transformation by orthogonal function expansion on the noise image generated using the noise signal. Embedded and distributed throughout the image. Since the embedded additional information is detected by image conversion of the binarized image by orthogonal function expansion, the binary image embedded with the additional information is output using an output device such as a printer. Even if dot variations occur and noise occurs in the binarized image, those variations are averaged and canceled in the binarized image that has been subjected to image conversion by orthogonal function expansion. Therefore, it is possible to detect the additional information with almost no influence of noise. The present invention is most suitable as a method for creating a digital watermark.

  Further, since the additional information embedded in the binarized image by the additional information embedding method of the present invention is distributed throughout the generated binarized image, the additional information is also obtained from a part of the binarized image. It can be reliably detected and restored. That is, additional information can also be detected and reproduced by cutting out a part of the binarized image and subjecting a part of the image to image conversion by orthogonal function expansion.

  Further, in the additional information embedding method of the present invention, additional information is given to the noise conversion image obtained by performing image conversion by orthogonal function expansion on the noise image generated using the noise signal. Even if the input resolution is unknown, the additional information can be reliably detected.

  In the additional information embedding method according to the present invention, the continuous tone image is binarized using the modulated threshold matrix, so that the original continuous tone image is accurately reproduced from the binarized image by back calculation. I can't do it. For this reason, it is impossible to completely remove the additional information. For example, even when the additional information is used as a signature, the signature is prevented from being removed.

  Further, the additional information embedding method of the present invention creates a noise mask based on the additional information, generates a threshold matrix modulated using the noise mask, and uses the modulated threshold matrix to convert the continuous tone image into a continuous tone image. A valuation process is performed. In the present invention, since a noise mask and a modulated threshold matrix can be prepared in advance, it is necessary to perform orthogonal function expansion such as Fourier transform each time the same additional information is embedded in a plurality of original images. In addition, a plurality of original images can be binarized in a short time using a modulation threshold matrix prepared in advance based on the additional information. As described above, according to the additional information embedding device of the present invention, it is possible to shorten the processing time when embedding additional information in a plurality of original images. Moreover, since the same pattern does not appear anywhere in the space by using the noise mask, the periodicity is lost, and the additional information can be dispersed throughout the space.

  Since the additional information detection method of the present invention can perform orthogonal function expansion on a binarized image, as a method for detecting additional information from a binarized image in which information is embedded by the additional information embedding method of the present invention. Is optimal.

  Since the embedding apparatus of the present invention can realize the additional information embedding method of the present invention, a binary image in which additional information having excellent noise resistance is embedded can be created very easily. The embedding device of the present invention is most suitable for next-generation display devices such as a facsimile, a monochrome printer, and a monochrome digital copier capable of embedding additional information.

  Since the additional information detection apparatus of the present invention can acquire key information based on image conversion in the additional information embedding apparatus of the present invention, additional information is obtained from a binary image in which additional information is embedded based on key information. It can be detected reliably. Therefore, a system including the additional information embedding device of the present invention and the additional information detection device of the present invention is optimal for an encryption system using key information as encryption.

The method and apparatus of the present invention will be described below in detail based on preferred embodiments shown in the accompanying drawings.
[Embodiment 1]
FIG. 1 is a block diagram of an image processing system 10 including an additional information embedding device according to the present invention. The image processing system 10 includes an image input device 11, an additional information embedding device 12, and an image output device 13. The image input device 11 can input a continuous tone image in which additional information is to be embedded. The embedding device outputs a halftone image (binarized image) in which additional information is embedded to the continuous tone image output from the image input device by applying the embedding method of the present invention to the image output device as image data. can do. The image output device 13 can output the halftone image output from the additional information embedding device 12.

  The image input device 12 is for creating continuous tone image data and outputting it as image data to the additional information embedding device 12, for example, an image recording device, specifically a color (or monochrome) negative. A film scanner apparatus that reads a color film image such as a film or a color (or monochrome) reversal film with an imaging element such as a CCD element or an imaging device and creates digital image data can be used. In addition, a reflection document scanner device that reads a color reflection document image such as a printed matter or a reflection print image with an image sensor to create digital image data may be used. Alternatively, an image capturing device, specifically, a digital camera, an electronic still camera, or a video camera that directly captures an object with an image sensor or an image capturing device and creates digital image data can be used. Recording medium storing the digital image data, for example, smart media, semiconductor memory such as PC card, magnetic recording medium such as FD or Zip, magneto-optical recording medium such as MO or MD, or CD-ROM or Photo- A driver that drives an optical recording medium such as a CD and reads it as digital image data may be used. Further, a display device such as a CRT monitor or a liquid crystal monitor that reads digital image data and displays a soft copy image, or an image processing that performs image processing of the read or displayed digital image data in whole or in part. A computer such as a PC or WS can also be used.

  The image output device 13 is for outputting a binary image in which the additional information output from the additional information embedding device 12 is embedded. For example, the image output device 13 outputs a hard copy image such as a reflection print image or a reflection original image. Image output devices such as digital photo printers, copiers, electrophotography, laser printers, ink jets, thermal sublimation type, TA, and other digital printers, TVs for displaying soft copy images, CRT monitors, liquid crystal monitors, etc. Examples thereof include a display device and a computer such as a PC or WS.

  The additional information embedding device 12 includes a noise mask generation unit 14, a threshold matrix modulation unit 15, and a binarization processing unit 16. The noise mask generation unit 14 generates a noise image using white noise as a noise signal, performs image conversion by orthogonal function expansion on the noise image, generates a noise conversion image, and generates the noise conversion image from the noise conversion image and the image input device 11. A noise mask can be generated by performing image conversion opposite to the image conversion performed on the noise image on the image obtained by multiplying the corresponding pixels of the input additional information image with each other. The noise mask generated by the noise mask generator 14 is output to the threshold matrix modulator 15.

  The threshold value matrix modulation unit 15 can modulate the threshold value of each component of the known threshold value matrix based on each component of the noise mask generated by the noise mask generation unit 14. The modulated threshold value matrix is output to the binarization processing unit 16. As shown in the figure, the binarization processing unit 16 has an input side connected to the threshold value matrix modulation unit 15 and the image input device 11, and an output side connected to the image output device. The binarization processing unit 16 can acquire an original image to which additional information is added from the image input device 11 and refers to a threshold matrix (hereinafter referred to as a modulation threshold matrix) having components modulated based on a noise mask. Then, the original image can be converted into a binary image in the area modulation type and output to the image output device 13 as binary image data.

  Next, the additional information embedding method of the present invention will be specifically described with reference to FIG. FIG. 2 is a flowchart showing the operation of the embedding device of the present invention. In the following description, the continuous-tone sunflower image shown in FIG. 8 is used as the original image in which the additional information is embedded, and the character “sunflower” shown in FIG. Use. Further, Fourier transform is used as the orthogonal function expansion.

(Step S100)
The data of the continuous tone image of sunflower shown in FIG. 8 is acquired from the image input device of FIG.

(Step S110)
A noise mask is generated in the noise mask generator 14 shown in FIG. Here, the generation of the noise mask will be described with reference to FIG. FIG. 3 is a flowchart of the noise mask generation method of the present embodiment. First, additional information to be embedded in a continuous-tone sunflower image is created as an image (step S301). Here, additional information (hereinafter referred to as an additional information image) on which characters “sunflower” shown in FIG. 9 are drawn is created. The additional information image may be created by inputting a printed matter on which the characters “sunflower” are written from the input device shown in FIG. 1 or using a computer. Next, a normal distribution random image generated in real space is created as white noise (step S302). The created normal distribution random image is subjected to image transformation by Fourier transformation. (Step S303). Thereby, a Fourier spectrum having components of almost the same size from a low frequency to a high frequency in the frequency space is created. Then, filtering processing is performed on the noise converted image using the additional information image (step S304). The filtering process uses a method of obtaining the product of the additional information image and each component of the Fourier spectrum by regarding the additional information image as a filter in the frequency space. The obtained product is noise having a waveform component of the additional information image in the frequency space. That is, filtering processing is performed so that noise is transmitted only through the portion indicated as the character of the additional information image in the frequency space. Further, the obtained product is subjected to inverse Fourier transform (step S305), and an image (noise mask) is created (step S306). FIG. 7 shows the created noise mask. Here, the size of the noise mask is adjusted so as to match the size of the binarized image to be finally created.

(Step S120)
The threshold value of the threshold matrix that is referred to when the continuous tone image is converted into a binary image in the area modulation type is modulated based on the value of each pixel of the noise mask. Here, a dot concentration type dither matrix of 150 lines and 45 degrees is used as the threshold matrix. As a method of modulating the threshold matrix, for example, the coordinate position of the threshold matrix referred to when converting a continuous tone image into a binarized image without using a noise mask is moved to the locus position in a certain direction. It is possible to use a method of referring to threshold values of coordinate positions separated by a distance obtained by multiplying the noise magnitude Δ of the corresponding noise mask by a constant. For example, if the threshold value referred to at a certain pixel position is T, when the threshold value T is modulated, the noise magnitude Δ of the noise mask at the position corresponding to the pixel position is referred to in a certain direction, Reference is made to the threshold value of the threshold value matrix at pixel positions separated by a constant multiple (k times) of the noise magnitude Δ, and this threshold value is used as a new threshold value. The constant k can be adjusted according to the noise value distribution of the noise mask, and the noise value distribution depends on the shape of the additional information image. Here, if the distribution width of k · Δ is increased, it becomes relatively easy to embed additional information and extract additional information. However, since the noise component increases, the image quality of the binarized image may be deteriorated. On the other hand, if the distribution width of the value of k · Δ is reduced, it becomes relatively difficult to embed and extract the additional information, but the image quality of the binarized image is improved. Thus, since the value of k depends on the shape of the additional information image and the amount of noise added to the binarized image, the ease of embedding and extracting additional information and the degree of image quality of the binarized image It is desirable to adjust appropriately according to.

  In addition, as a modulation method of different thresholds, a threshold value that is referred to when converting a continuous tone image to a binarized image without using a noise mask is changed due to the noise magnitude of the noise mask corresponding to the position. It is also possible to apply a method of adding minutes and using this as a new threshold value. For example, if a threshold value referred to at a certain pixel position is T, the noise mask size Δ at the position corresponding to the pixel position is referred to, and a constant multiple (k times) of the noise size Δ is added. The combined threshold value (1 + k · Δ) × T is used as a new threshold value. In this manner, a method of modulating each component of the threshold matrix according to the noise magnitude of the noise matrix can be used.

(Step S130)
Referring to a threshold matrix having a modulated threshold value, a sunflower continuous tone image is binarized into an area modulation type image.
In this way, a binary image in which the additional information image shown in FIG. 9 is embedded in the Fourier space is created. FIG. 10 shows the created binarized image.

  The additional information added to the binarized image shown in FIG. 10 can be easily detected using, for example, an apparatus having a function of Fourier transforming the binarized image. That is, the binarized image shown in FIG. 10 is input to the detection device using the image input device, and the input binarized image is Fourier transformed in the detection device. FIG. 11 shows an image after the Fourier transform of the binarized image. As shown in FIG. 11, it can be seen that the character information of sunflower is reproduced. In FIG. 11, the characters “sunflower” appear in two places with the center of the image as point symmetry. This is due to the nature of the Fourier transform.

  In the above-described embodiment, the Fourier transform is used as the orthogonal function expansion in the generation of the noise mask in S110 of FIG. 2, but the present invention is not limited to this, and an arbitrary orthogonal function expansion can be used. Examples of the orthogonal function expansion include discrete cosine transform (DCT), Walsh-Hadamard transform (WHT), discrete Fourier transform (DFT), and discrete sine transform (DST) employed in JPEG. , Haar transformation, slant transformation, Karhunen / Loeve transformation (KLT), etc. can be used. When detecting the additional information, the same orthogonal function expansion as that used may be used.

[Embodiment 2]
In the present embodiment, in the generation of the noise mask in step S110 of FIG. 2, a method of embedding additional information when the type of orthogonal function expansion and the allocation of each basis function in orthogonal function expansion are used as key information and the additional information. A method of detecting the will be described. The embedding of the additional information is the same as that of the first embodiment except for the generation of the noise mask, and thus the description thereof is omitted.

  FIG. 4 shows a flowchart of the noise mask generation method of the present embodiment. The noise mask generation method will be described below with reference to FIG.

(Step S401)
A normal distribution random image is created as white noise by the same method as in the first embodiment.

(Step S402)
Image transformation by orthogonal function expansion is performed on the normal distribution random image based on the key information. Here, as the orthogonal function expansion, discrete cosine transform, Walsh Hadamard transform, discrete Fourier transform, discrete sine transform, Haar transform, slant transform, Karhunen / Labe transform, etc. can be used. Therefore, using one of these as key information, image transformation of a normal distribution random image is performed using the orthogonal function expansion.

(Step S403)
A noise transformed image obtained by performing image transformation by orthogonal function expansion on a normal distribution random image is visualized on a two-dimensional space based on key information. Here, based on the key information, a process for visualizing a noise transformed image obtained by image transformation based on orthogonal function expansion in a two-dimensional space (hereinafter referred to as a two-dimensional visualization process) will be described in detail.

  First, the horizontal axis of the two-dimensional space is set to the X axis and the vertical axis is set to the Y axis, and the order of the basis function used in the orthogonal function expansion is assigned to each point on the two-dimensional space. Which basis function order is assigned to each point in the two-dimensional space is used as key information. For example, as a typical example, discrete Fourier transform is used as an orthogonal function expansion, the zero frequency component is the origin, fx of the basis function exp (-i * (fx + fy)) is the X axis, and fy is the Y axis Make it correspond. Here, fx and fy indicate spatial frequencies, fx = nπ (x / a), fy = mπ (y / b), n and m indicate orders, and a and b are on the X axis and the Y axis. The range of the image when it is represented. In this case, fx and fy are indexes, and the magnitudes of the frequencies are the magnitudes of the X axis and the Y axis as they are. Then, the noise conversion image is visualized by substituting the absolute value of the coefficient applied to the basis function corresponding to each degree into each point in the two-dimensional space. In this way, an image converted by orthogonal function expansion can be visualized in a two-dimensional space. For example, if luminance, RGB values, gray scale, etc. are used as the absolute values of the basis functions corresponding to the respective orders, a noise-converted image that has been image-converted by orthogonal function expansion can be expressed as a grayscale image in a two-dimensional space. it can. Further, since the additional information component in the frequency space is small, instead of the absolute value of the basis function, the logarithm of the absolute value of the basis function is taken and then displayed in gray scale, and γ correction and image binarization processing are performed appropriately. Thereby, detection of additional information can be further facilitated.

  Here, it is possible to arbitrarily select which basis function order is assigned to each point in the two-dimensional space. As described above, when the discrete Fourier transform is used as the orthogonal function expansion, they may be arranged in ascending order of frequency or may be arranged at random. If such assignment is set in advance as key information, that is, what basis function is assigned to each point in the two-dimensional space is assigned in advance as key information, it is based on this key information. By assigning a basis function to each point in the two-dimensional space, a noise transformed image obtained by image transformation by orthogonal function expansion can be visualized in the two-dimensional space.

(Step S404)
The noise conversion image visualized on the two-dimensional space is filtered using the additional information image. The filtering process is performed by obtaining the product of each pixel of the noise-converted image visualized by the above-described two-dimensional visualization process and the corresponding additional information image pixel.

(Step S405)
An orthogonal expansion opposite to the orthogonal function expansion used in the image conversion is performed on the obtained product. In this way, a noise mask is created.

  By using the noise mask thus obtained, the threshold matrix can be modulated by the same method as in the first embodiment, and the continuous tone image can be binarized using the modulated threshold matrix.

  Next, a method for detecting additional information from a binarized image in which additional information is embedded using the above-described method using an additional information detection system will be described with reference to the drawings. FIG. 5 is a block diagram showing a schematic configuration of the additional information detection system. As illustrated in FIG. 5, the additional information detection system 50 includes an image input device 51, an additional information detection device 52, and an image output device 53. The image input device 51 can input a binary image in which additional information is embedded, and the image output device 53 can output additional information detected by the additional information detection device 52. The image input device 51 and the image output device 53 can be configured using, for example, the same devices as the image input device 11 and the image output device 13 of the image processing system 10 shown in FIG.

  The additional information detection device 52 includes a key information acquisition unit 55, an orthogonal function expansion processing unit 56, and a two-dimensional visualization processing unit 57. The key information acquisition unit 55 can acquire key information necessary for detecting additional information from the binarized image by input from the external input unit 58 or the like. The orthogonal function expansion processing unit 56 can expand the binarized image input from the image input device 51 based on the type of orthogonal function expansion acquired by the key information acquisition unit 55. The two-dimensional visualization processing unit 57 can perform processing for visualizing a binarized image on which an orthogonal function has been developed in a two-dimensional space.

  A method for detecting and reproducing additional information from a binary image in which additional information is embedded based on key information using the additional information detection system 50 will be described with reference to FIG. FIG. 6 is a flowchart showing the operation of the additional information detection system 50.

  First, a binarized image in which additional information is embedded is input from the image input device 51 (step S601). Next, key information is acquired in the key information acquisition unit 55 (step S602). The key information is the type of orthogonal function expansion and which basis function is assigned to each point in the two-dimensional space in the two-dimensional visualization processing unit 57.

  Image conversion of the binarized image is performed using the orthogonal function expansion acquired by the key information acquisition unit 55 (step S603). Then, based on the key information, a two-dimensional visualization process similar to that described above is performed on the binarized image converted by the image conversion based on the orthogonal function expansion to visualize the two-dimensional space (step S604). In the two-dimensional visualization processing unit 57, for example, the horizontal axis of the two-dimensional space is the X axis and the vertical axis is the Y axis, and the order of a predetermined basis function is assigned to each point on the two-dimensional space based on the key information. Perform visualization by substituting the absolute value of the basis function corresponding to each degree to each point in the two-dimensional space. At this time, for example, by using luminance, RGB values, gray scales, and the like as the absolute values of the basis functions corresponding to the respective orders, the additional information is expressed as a grayscale image in the two-dimensional space. In this way, additional information is detected from the binarized image and reproduced (step S605).

  As in this embodiment, information based on the type of image conversion used when embedding additional information in a binarized image is used as key information, and additional information embedded in the binarized image is detected based on the key information. Thus, the reproducing system is suitable as an encryption system.

  In the above embodiment, additional information can be easily extracted by performing image processing such as gamma correction and two-gradation on an image that has been subjected to image conversion by orthogonal function expansion.

  Even if the binarized image in which the additional information is embedded by the embedding method of the present invention is output using an output device such as a printer and then input from an input device such as a scanner or a digital camera, Additional information can be extracted from the input image.

  In the above embodiment, an image on which characters are drawn is used as the additional information, but various patterns such as an image on which a picture, a figure, a design, a pattern, a geometric pattern, or the like is drawn can also be used.

  In the above embodiment, the normal distribution random image generated in the real space is used as the white noise. However, the present invention is not limited to this, and it is only necessary to have a certain intensity in the frequency space. For example, the delta function You may use what was obtained by Fourier-transforming as white noise.

  In the present invention, when the Fourier transform is selected as the orthogonal function expansion, it is preferable that the additional information is formed in a high frequency region around the image region when the additional information is represented by an image. Thereby, the image quality of the binarized image in which the additional information is embedded by the additional information embedding process of the present invention can be improved as compared with the case where the additional information is formed in the low frequency region. However, when a binarized image in which additional information is embedded is output by an output device such as a printer, and the output binarized image is input by an input device such as a scanner, the higher the frequency component, the greater the attenuation. Therefore, it may be difficult to detect the additional information. Therefore, it is preferable to add information according to the system.

In the present invention, the detectability when detecting additional information also depends on the contents of the original image. For example, when an image having many low frequency components and little change in the entire image is used as the original image. Although it is easy to detect additional information, when an image having a fine structure containing a lot of high-frequency components is used as an original image, it may be difficult to detect the additional information. Therefore, the original image is preferably an image having many low-frequency components and little change in the entire image. However, even when an image having a fine structure containing many high-frequency components is used as the original image, the magnitude of noise is large. Since the additional information can be embedded by strengthening, the detection capability is enhanced as a result.
In the present invention, it is preferable that the noise mask is uniform, whereby the binarized image in which the additional information is embedded is less disturbed and the image quality is improved.

  The additional information embedding method and embedding device according to the present invention have been described in detail above. However, the present invention is not limited to the above-described embodiment, and various improvements and modifications may be made without departing from the spirit of the present invention. Good.

It is a block diagram for demonstrating an image processing system provided with the additional information embedding apparatus of this invention. It is a flowchart which shows operation | movement of the additional information embedding apparatus of this invention. It is a flowchart of the production | generation method of the noise mask in Embodiment 1 of this invention. It is a flowchart of the noise mask production | generation method in Embodiment 2 of this invention. It is a block diagram for demonstrating the additional information detection system provided with the additional information detection apparatus of this invention. It is a flowchart for demonstrating operation | movement of the additional information detection system shown in FIG. 3 is a noise mask created in the first embodiment. 2 is a continuous tone image of a sunflower used in Embodiment 1. FIG. It is an image in which sunflower characters used as additional information are drawn. It is a binarized sunflower image in which additional information is embedded by the additional information embedding method of the present invention. This is additional information of the character “sunflower” detected by the additional information detection method of the present invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Image processing system 11, 51 Image input device 12 Additional information embedding device 13, 53 Image output device 14 Noise mask production | generation part 15 Threshold matrix modulation part 16 Binarization processing part 50 Additional information detection system 52 Additional information detection apparatus 55 Key information Acquisition unit 56 Orthogonal function expansion processing unit 57 Two-dimensional visualization processing unit

Claims (17)

An additional information embedding method for generating a binarized image in which additional information is embedded by performing binarization processing using a threshold matrix on an original image,
A noise image is generated using a noise signal, and the additional information is added to the noise conversion image obtained by performing image conversion by orthogonal function expansion on the noise image, and then the noise conversion is performed by performing a reverse conversion to the image conversion. Generating a mask; and
Modulating the threshold matrix based on the noise mask;
An additional information embedding method comprising: binarizing the original image based on the modulated threshold matrix.   The addition of additional information to the noise-converted image is performed by multiplying the component of each pixel position of the additional information image obtained by representing the additional information as an image and the component of the noise-converted image corresponding to each pixel position. The additional information embedding method according to claim 1, wherein:   The orthogonal function expansion is one selected from the group consisting of discrete cosine transform, Walsh-Hadamard transform, discrete Fourier transform, Fourier transform, discrete sine transform, Haar transform, slant transform, and Karhunen / Labe transform The additional information embedding method according to claim 1 or 2. Modulating the threshold matrix comprises:
When the threshold value at a predetermined coordinate position of the threshold value matrix is T, the noise magnitude of the noise mask at the coordinate position is Δ, and k is a constant,
The additional information according to claim 1, wherein the threshold value matrix is modulated by replacing a threshold value at a predetermined coordinate position of the threshold value matrix with (1 + k · Δ) × T. Embed method.   The step of modulating the threshold value matrix includes the threshold value matrix existing at a coordinate position separated in a fixed direction by a distance obtained by multiplying a threshold value at a predetermined coordinate of the threshold value matrix by a constant multiple of the noise magnitude of the noise mask at the coordinate position. The additional information embedding method according to claim 1, wherein the threshold value matrix is modulated by substituting the threshold value.   A method for detecting the additional information from a binarized image in which additional information is embedded by the method according to any one of claims 1 to 5, wherein image conversion by orthogonal function expansion is performed on the binarized image. A method for detecting additional information, characterized in that the additional information is detected by performing. Further, the method includes a step of obtaining key information representing information on orthogonal function expansion used when the additional information is embedded,
7. The detection of additional information according to claim 6, wherein image conversion by the orthogonal function expansion is performed on the binarized image based on the acquired key information, and the image converted image is visualized. Method.   The said key information is the kind of said orthogonal function expansion | deployment, and the information of the allocation method of the order of the basis function in the said orthogonal function expansion | deployment at the time of the image transformation by the said orthogonal function expansion | deployment, It is characterized by the above-mentioned. To detect additional information. An additional information embedding device for generating a binarized image in which additional information is embedded by performing binarization processing using a threshold matrix on an original image,
A noise image is generated using a noise signal, and the additional information is added to the noise conversion image obtained by performing image conversion by orthogonal function expansion on the noise image, and then the noise conversion is performed by performing a reverse conversion to the image conversion. Means for generating a mask;
Means for modulating the threshold matrix based on the noise mask;
Means for binarizing the original image based on the modulated threshold value matrix.   The addition of additional information to the noise-converted image is performed by multiplying the component of each pixel position of the additional information image obtained by representing the additional information as an image and the component of the noise-converted image corresponding to each pixel position. The additional information embedding device according to claim 9, wherein   The orthogonal function expansion is one selected from the group consisting of discrete cosine transform, Walsh-Hadamard transform, discrete Fourier transform, Fourier transform, discrete sine transform, Haar transform, slant transform, and Karhunen / Labe transform The additional information embedding device according to claim 9 or 10. The means for modulating the threshold matrix comprises:
When the threshold value at a predetermined coordinate position of the threshold value matrix is T, the noise magnitude of the noise mask at the coordinate position is Δ, and k is a constant,
The additional information according to claim 9, wherein the threshold value matrix is modulated by replacing a threshold value at a predetermined coordinate position of the threshold value matrix with (1 + k · Δ) × T. Implantation device.   The step of modulating the threshold value matrix includes the threshold value matrix existing at a coordinate position separated in a fixed direction by a distance obtained by multiplying a threshold value at a predetermined coordinate of the threshold value matrix by a constant multiple of the noise magnitude of the noise mask at the coordinate position. The additional information embedding device according to claim 9, wherein the threshold value matrix is modulated by substituting the threshold value.   An apparatus for detecting additional information from a binarized image in which additional information is embedded by the apparatus according to any one of claims 9 to 13, wherein image conversion is performed on the binarized image by orthogonal function expansion. An additional information detection apparatus comprising: means. Furthermore, means for obtaining key information representing information of orthogonal function expansion used when embedding the additional information,
The addition according to claim 14, further comprising means for performing image conversion by the orthogonal function expansion on the binarized image based on the acquired key information and visualizing the image converted image. Information detection method. The additional information embedding device according to claim 9,
A system comprising the additional information detection device according to claim 14. An additional information embedding program for causing a computer to execute a step of generating a binarized image in which additional information is embedded by performing binarization processing using a threshold matrix on an original image,
In the computer,
A noise image is generated using a noise signal, and the additional information is added to the noise conversion image obtained by performing image conversion by orthogonal function expansion on the noise image, and then the noise conversion is performed by performing a reverse conversion to the image conversion. A procedure for performing a step of generating a mask;
Performing a step of modulating the threshold matrix based on the noise mask;
An additional information embedding program for executing a process including a step of executing a step of binarizing the original image based on the modulated threshold value matrix.

Images