JP2001223882A - Method for detecting deformed processing method of original picture, original picture size detecting method and electronic watermark detecting method for original picture - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a device for conducting an enlargement/reduction deformation processing on an original picture and judging whether deformed picture information from which electronic watermark information cannot be read is formed or not. SOLUTION: The device is provided with a Fourier transformation processing means 3 obtaining an intra-Fourier space picture 20 from an obtained picture 2, a means 4 burying a reference information signal S into the picture 20, an inverse Fourier transformation processing means 5 which transforms the intra-Fourier space picture 20 including the reference information signal S into an intra-picture space picture 30, a size change means 7 which individually changes the size of the picture 30, a means 8 which Fourier-transforms again the respective pictures 30, the detection means 9 of a frequency component value on the reference information signal S of an intra- Fourier space picture 40 and a means 11 arranging the frequency component value in accordance with an enlargement/reduction rate D and judging the presence/absence of a projected component value from a peripheral frequency component value.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting a deformed processing method of an original image, a method for detecting an original image size, and a method for detecting an electronic watermark. However, even if it is erased, deformed, or tampered with by any illegal act, it is necessary to confirm what kind of deformation processing operation has been performed on the obtained inspection image and to receive the confirmed deformation processing operation. A digital watermark information detecting method and apparatus capable of restoring an original image from the obtained image and detecting digital watermark information originally inserted in the original image information from the restored image; It is about.

[0002]

2. Description of the Related Art Still picture digital watermarking
The tal watermark) technology has been conventionally used as a technology for secretly embedding the creator's unique information in a still image and asserting the ownership. Also, in recent years, multimedia information has been widely distributed using the Internet, and the application of digital watermarks has attracted attention. At the same time, watermark information embedded in various images has been deformed, falsified, or erased. As a result, malicious fraudulent acts that attempt to avoid copyright infringement frequently occur, and it is urgently necessary to develop a method for preventing such fraudulent acts.

[0003] In this case, such digital watermark information may be deformed, falsified, or deleted, for example, by enlarging or reducing the original image information, or by rotating and deforming the original image information. It is known that the digital watermark information is easily deleted, deformed or falsified by consciously performing the above operation. Among them, the processing for making the digital watermark easy to be erased or impossible to read is,
This is a transformation operation called reduction.

That is, when an image in which a digital watermark is embedded is enlarged or reduced from the original image size to change the image size of the original image, the digital watermark information is deleted or empty information is added. However, since the digital watermark information is shifted from the position where it has been embedded in advance or is deleted, if the digital watermark is enlarged or reduced in the image, it cannot be detected accurately. is there.

[0005] Above all, by subjecting the above-mentioned original image information to enlargement / reduction deformation processing, the digital watermark information has a characteristic that it is easily erased. In other words, if the image in which the digital watermark information is embedded is subjected to enlargement / reduction deformation processing using an appropriate enlargement / reduction ratio, the electronic watermark information cannot be accurately detected.

That is, when the above-described deformation processing is performed on the original image information as described above, the image information itself often has the same image information between adjacent pixels. However, even if the image quality is slightly reduced, the original image information itself is not completely lost. However, since the digital watermark information has binary values of 0 and 1 arranged at random, However, even if the original digital watermark information becomes unreadable or readable due to the averaging process performed, the digital watermark information often differs from the original digital watermark information.

[0007] For this reason, a person who intends to commit fraudulent acts often employs such means since the image quality itself is in a state where it can be seen and the digital watermark information can be erased. Currently, in response to such fraud,
Frequently used methods employ an affine transformation type linear interpolation method to return an enlarged or reduced image to the original image, or a method of returning a rotated image to the original image using the estimated rotation angle. ing.

However, the strength of the embedded watermark is reduced by the linear interpolation method of the affine transformation, and the watermark may not be detected in some cases. Furthermore, when the original image is not obtained, the size of the original image is not known, so it is difficult to specify the size of the original image from the obtained enlarged or reduced image.

[0009] Therefore, conventionally, there has been no effective and reliable defense means against such misconduct, and therefore, development of such a defense method has been urgently required.

[0010]

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to improve the above-mentioned drawbacks of the prior art, and to provide a result of performing enlargement or reduction processing on original image information including digital watermark information. From the deformed image information in which the digital watermark information has been deleted from the image after the deforming process or has become unreadable, the deformed image information is subjected to an enlargement / reduction deformation process on the original image. A method and apparatus for determining whether or not the original image has been formed, and confirming at what enlargement / reduction ratio the original image has been subjected to the transformed image information, At the same time as providing a method and apparatus for detecting the image size of a digital watermark, and a digital watermark detection method capable of reliably detecting the original digital watermark information embedded in the original image information from the deformed image information. It is intended to.

[0011] Another object of the present invention is to allow a third party to illegally enlarge or reduce original image information, which is a still image whose source is clear, including predetermined digital watermark information. After that, the original digital watermark information is detected from the illegally processed still image information on the market, and the original image size necessary to protect the copyright of the image information from unauthorized infringement is detected. A method and a digital watermark detection method are provided.

[0012]

The present invention basically employs the following technical configuration to achieve the above object. That is, as a first aspect according to the present invention, after embedding a predetermined reference information signal in a predetermined portion in an image in a Fourier space obtained by Fourier-transforming the obtained image to be inspected, the image in the Fourier space is subjected to inverse Fourier transform. To the image in the image space, and then use the plurality of individual enlargement / reduction ratios selected from the enlargement / reduction ratios in the predetermined range to convert the image in the image space into an image of the image in the image space. After individually changing the size, the respective images in the image space in which the respective image sizes have been changed are Fourier-transformed again into Fourier space images, and in each Fourier space image, the image is first preceded by the image. The frequency component of the embedded reference information signal is detected, and the plurality of references in the re-Fourier space image of the plurality of types of image space images whose respective image sizes have been respectively changed. When the frequency components of the information signal are sequentially arranged in accordance with the magnitude of the enlargement / reduction rate, the value of the frequency component at a certain enlargement / reduction rate is more prominent than the value of the surrounding frequency component. When it is recognized that the inspection image is an original image, a method for detecting a deformation processing method of the original image is configured to determine that the inspection image is formed by enlarging / reducing the original image. In a second aspect according to the present invention, a predetermined reference information signal is embedded in a predetermined portion in a Fourier space image obtained by Fourier transforming an obtained image, and then the Fourier space image is subjected to inverse Fourier transform. The image in the image space is returned to the image in the image space, and the image size of the image in the image space is then reduced using a plurality of individual enlargement / reduction ratios selected from the enlargement / reduction ratios within a predetermined range. Change individually And then, at the Fourier space image and without each of the Fourier space image the image space in the image which the image size of the respective is changed by Fourier transform again,
The frequency component of the reference information signal previously embedded in the image is detected, and the frequency component of the reference information signal in the re-Fourier space image of the image in the image space in which the size of each image is changed is calculated as When the value of the frequency component at a certain enlargement / reduction ratio is more prominent than the value of the surrounding frequency component when the pixels are sequentially arranged according to the magnitude of the enlargement / reduction ratio, the value of the frequency component This is an original image size detection method configured to determine that the obtained image has been resized from the original image at the enlargement / reduction ratio corresponding to the frequency component value that is more prominent than the original image.

Further, as a third aspect of the present invention, the reciprocal of the original image size enlargement / reduction ratio of the obtained image obtained in the above-mentioned second aspect of the present invention is represented by the obtained image. This is a digital watermark detection method configured to reproduce an original image by applying the method described above, and detect the presence or absence of watermark information on the original image. Further, as a fourth aspect according to the present invention, there is provided an apparatus for detecting the size of an original image from an obtained image, wherein the apparatus performs a Fourier transform process on the obtained image to convert an image in the first Fourier space into an image. First Fourier transform processing means to be formed, reference information signal embedding means for embedding a predetermined reference information signal at a predetermined position in an image in the Fourier space of the obtained Fourier-transformed image, and the reference information signal being embedded. Inverse Fourier transform processing means for performing an inverse Fourier transform on the image in the Fourier space to convert the image in the image space into an image in the image space, and storing a plurality of types of enlargement / reduction ratios for changing the image size of the image in the image space The desired enlargement / reduction rates are sequentially selected from the enlarged / reduction rate storage means, and the selected plurality of individual enlargement / reduction rates are individually used. Image size changing means for sequentially and individually changing the image sizes of the images in the image space, and performing a Fourier transform process again on each of the images in the image space in which the image size has been changed to convert a plurality of second Fourier space images Second Fourier transform processing means to be formed; frequency component value detection means for detecting a frequency component value related to the reference information signal in the respective second image space image; the respective second image space image Frequency component value arranging means for arranging a plurality of the frequency component values obtained from in a predetermined order in accordance with the enlargement / reduction ratio.
Protruding frequency component value determining means for determining whether or not there is a frequency component value protruding from the surrounding frequency component values; if the presence of the protruding frequency component value is confirmed, An original image size determining means for determining the corresponding enlargement / reduction ratio to be the enlargement / reduction ratio used when the obtained image is resized from the original image It is a detection device.

On the other hand, as a fifth aspect according to the present invention,
A first step of obtaining an image to be inspected from which digital watermark information similar to the original image whose source is known cannot be read, and a step of Fourier-transforming the obtained image to be inspected into a first Fourier space image. A second step, a third step of embedding a predetermined reference information signal in a predetermined portion of the first Fourier space image, an inverse Fourier transform of the Fourier space image in which the reference information signal is embedded, A fourth step of returning to the inside image, enlarging / reducing the image in the image space within a predetermined range.
Fifth step of individually changing the image size of the image in the image space using the plurality of individual enlargement / reduction ratios selected from the reduction ratio, in the image space in which the respective image size has been changed. A sixth step of re-Fourier transforming the image into a second Fourier space image; in each of the Fourier space images, individually separate the frequency component values of the reference information signal previously embedded in the image; A seventh step of detecting, an eighth step of sequentially arranging the detected frequency component values of the individual reference information signals according to the magnitude of the enlargement / reduction ratio, and a seventh step of arranging the frequency component values. The value of the frequency component at a certain enlargement / reduction ratio is
It is determined whether or not the frequency component value protrudes from the value of the surrounding frequency component. If the frequency component value is determined to protrude from the value of the surrounding frequency component, the frequency component value corresponds to the frequency component value. A ninth step of judging the enlargement / reduction ratio to determine that the obtained image has been resized by the enlargement / reduction ratio from the original image. A sixth aspect according to the present invention is to obtain an image to be inspected from which digital watermark information similar to the original image whose source is known cannot be read. A first step, a second step of performing a Fourier transform on the obtained image to be inspected into a first Fourier space image, a predetermined reference information signal to a predetermined portion of the first Fourier space image; Embed Step 3, a fourth step of performing an inverse Fourier transform of the image in the Fourier space in which the reference information signal is embedded to return the image in the image space, selecting the image in the image space from an enlargement / reduction ratio within a predetermined range A fifth step of individually changing the image size of the image in the image space using the plurality of individual enlargement / reduction ratios obtained, and re-Fouriering the image in the image space in which the respective image sizes have been changed. A sixth step of converting into a second Fourier space image;
A seventh step of individually detecting a frequency component value of the reference information signal previously embedded in the image in each of the Fourier space images, a frequency component value of the detected individual reference information signal; An eighth step of sequentially arranging the frequency components in accordance with the magnitude of the enlargement / reduction ratio, wherein the value of the frequency component at a certain enlargement / reduction ratio in the arranged frequency component values is changed to the surrounding frequency component Is determined to be greater than the value of the frequency component, and if the frequency component value is determined to be greater than the value of the surrounding frequency component, the enlargement / reduction ratio corresponding to the frequency component value is determined. To
A ninth step of judging that the obtained image has been resized by the enlargement / reduction ratio from the original image, and applying the reciprocal of the enlargement / reduction ratio obtained in the ninth step to the obtained image And an eleventh step of detecting the presence or absence of watermark information with respect to the original image, thereby performing a digital watermark detection method of the original image. This is a recording medium that records a program.

[0015]

BEST MODE FOR CARRYING OUT THE INVENTION The method for detecting an original image to be deformed, the method for detecting an original image size, and the method for detecting an electronic watermark according to the present invention employ the above-described technical configuration. An unauthorized act such as an image enlargement process or an image reduction process is performed by a third party on the original image information that is a still image including predetermined electronic watermark information, and the electronic watermark information is deleted, or From the illegitimately processed still image information on the market after making it unreadable, what kind of deformation processing operation has been performed on the illegitimately processed still image information for the original image It is easily detected and the commercially available image information is illegally sold, without the consent of the copyright holder, by illegally altering, transforming or deleting the original image information having the predetermined digital watermark information. Because whether or not to have made it possible to identify, become a useful tool in order to protect the copyright from illegal infringement.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an original image deformed processing method detecting apparatus 300 for executing the original image deformed processing detecting method according to the present invention, and an original image size detecting method 100 for executing the original image size detecting method will be described. A specific example of the digital watermark detection device 200 that executes the digital watermark detection method will be described in detail with reference to the drawings.

First, FIG. 1 is a block diagram showing a configuration of a specific example of an original image deformed processing method detecting apparatus 300 that executes the original image deformed processing method detecting method according to the present invention. An apparatus 300 for detecting the size of an original image from an obtained image, wherein a first Fourier transform processing is performed on the obtained image 2 to form a first Fourier space image 20. Means 3,
Reference information signal embedding means 4 for embedding a predetermined reference information signal S at a predetermined position P in the Fourier space image 20 of the obtained Fourier-transformed image, in the Fourier space in which the reference information signal S is embedded. An inverse Fourier transform processing unit 5 for performing an inverse Fourier transform on the image 20 to convert the image 20 into an image 30 in an image space, and a plurality of types of enlargement / reduction ratios D for changing the image size of the image 30 in the image space.
(D1, D2, D3,... Dn), and the desired enlargement / reduction rate D (D1, D2, D3,... Dn) from the enlargement / reduction rate storage means 6 ) Are sequentially selected, and a plurality of selected individual enlargement / reduction rates D are selected.
Image size changing means 7 for individually and sequentially changing the image sizes of the images 30 in the image space by individually using D1, D2, D3,..., Dn; A second Fourier transform of the inner image 30 again to form a plurality of second Fourier space images 40.
A Fourier transform processing means 8, a frequency component value detecting means 9 for detecting a frequency component value related to the reference information signal S in the respective second Fourier space image 40, and the respective second Fourier space image 40, the plurality of frequency component values obtained from the scale factor D (D
1, D2, D3,... Dn), the frequency component value arranging means 10 for arranging in a predetermined order, the frequency protruding from the neighboring frequency component values in the plurality of frequency component values arranged in a series. A protruding frequency component value determining unit 11 that determines whether or not a component value exists; based on an output value output from the protruding frequency component value determining unit 11,
The inspected image is enlarged / reduced to a predetermined original image and is determined to be an enlarged / reduced deformed image.
An apparatus 300 for detecting a method of processing an original image to be deformed, which is constituted by a reduced deformation processing determining means 21 is shown.

Therefore, in this specific example, it is needless to say that it may be determined that the obtained still image 2 has not been subjected to the rotation deformation processing and the enlargement / reduction deformation processing. In the above-described apparatus for detecting a method of processing a deformed original image 300 according to the present invention, the protruding frequency component value judging means 11 determines whether there is a protruding frequency component value from the output of the frequency component value arranging means 10. When determining whether or not the noise component is not present, it is preferable that a noise removing unit 13 that removes a background value, which is a predetermined noise, from the output data of the frequency component value arranging unit 10 is provided.

Further, in the original image deformed processing method detecting apparatus 300 according to the present invention, it is not confirmed what kind of deformation processing operation has been performed on the obtained still image 2 at first. A rotational deformation processing determining means 17 which can determine whether or not the obtained still image 2 is subjected to the rotational deformation processing, which can relatively easily perform the determination. It is also desirable to have.

Here, as a method of identifying whether the obtained image information to be inspected has undergone rotational deformation processing or another deformation processing operation, concretely, there are some known techniques. Among them, a simpler method is, for example, a method in which a still image obtained from the market is subjected to rotational deformation processing by using an "image edge detection method" as shown below. You can detect if there is.

That is, in the image edge detection method, as shown in FIG. 7, the rotation angle of the image can be calculated by extracting the edge of the image using a differential filter. On the other hand, if an edge cannot be detected by the above-described image edge detection method, the still image obtained from the market is enlarged /
It may be estimated that the image has been subjected to the reduction deformation processing or has undergone another deformation processing operation.

Therefore, in this specific example, first, the obtained still image 2 is identified by using the rotation deformation processing discriminating means 17, and if it is determined that the rotation deformation processing has been performed, the notation is displayed. By displaying the result on the means 23,
The subsequent processing may be stopped, and the result of the discrimination by the rotation / transformation processing determining means 17 indicates that the obtained still image 2 is not the one subjected to the rotation / transformation processing, but another transformation including the enlargement / reduction transformation processing. When it is clear that the still image 2 has been processed, it is determined whether the obtained still image 2 has been subjected to the enlargement / reduction deformation processing by using the first Fourier transform processing means 3 and the following means. An operation for identifying whether or not to execute is performed.

On the other hand, in the present invention, the first and the second
Of the Fourier transform means 3 and 8 may be individually provided, or may be configured to be commonly used by the same means. Further, in the apparatus 300 for detecting a method of processing the original image to be deformed in this specific example, it is also possible to display the judgment result of the enlargement / reduction deformation judgment means 21 on an appropriate display means 22. .

In FIG. 1, reference numeral 18 denotes an apparatus for detecting a method of processing the original image to be deformed according to the present invention, and an apparatus 100 for detecting an original image size and an apparatus 200 for detecting a digital watermark described below. It is a central processing control means (CPU) for comprehensively controlling each constituent means. Next, the configuration of a specific example of the original image size detection device 100 and the digital watermark detection device 200 according to the present invention will be described with reference to FIG.

FIG. 1 is a block diagram showing a configuration of one specific example of devices 100 and 200 for executing the original image size detecting method and the digital watermark detecting method according to the present invention. An apparatus 100 for detecting the size of an original image from an image 2, comprising a first Fourier transform processing means 3 for performing a Fourier transform on the obtained image 2 to form a first Fourier space image 20;
Reference information signal embedding means 4 for embedding a predetermined reference information signal S at a predetermined position P in the Fourier space image 20 of the obtained Fourier-transformed image, in the Fourier space in which the reference information signal S is embedded. An inverse Fourier transform processing unit 5 for performing an inverse Fourier transform on the image 20 to convert the image 20 into an image 30 in an image space, and a plurality of types of enlargement / reduction ratios D for changing the image size of the image 30 in the image space.
(D1, D2, D3,... Dn), and the desired enlargement / reduction rate D (D1, D2, D3,... Dn) from the enlargement / reduction rate storage means 6 ) Are sequentially selected, and a plurality of selected individual enlargement / reduction rates D are selected.
Image size changing means 7 for individually and sequentially changing the image sizes of the images 30 in the image space by individually using D1, D2, D3,..., Dn; A second Fourier transform of the inner image 30 again to form a plurality of second Fourier space images 40.
A Fourier transform processing means 8, a frequency component value detecting means 9 for detecting a frequency component value related to the reference information signal S in the respective second Fourier space image 40, and the respective second Fourier space image 40, the plurality of frequency component values obtained from the scale factor D (D
1, D2, D3,... Dn), the frequency component value arranging means 10 for arranging in a predetermined order, the frequency protruding from the neighboring frequency component values in the plurality of frequency component values arranged in a series. A protruding frequency component value determining unit 11 for determining whether or not a component value exists; when the presence of the protruding frequency component value is confirmed, the enlargement / reduction ratio Dn corresponding to the protruding frequency component value is determined. And an original image size determination unit 12 for determining that the obtained image 2 has the enlargement / reduction ratio D used when the size of the obtained image 2 is changed from the original image 1.
It is shown.

In the original image size detecting apparatus 100 according to the present invention, the frequency component value arranging means 1 is used.
When the prominent frequency component value determining means 11 determines whether there is a prominent frequency component value from the output of 0, the background value which is a predetermined noise is removed from the output data of the frequency component value arranging means 10. It is also desirable that a noise removing means 13 is provided.

On the other hand, the digital watermark detecting device 2 according to the present invention
00 is the original image size detection device 100 shown in FIG.
At the same time as the original image size determination means 12, furthermore, in the enlargement processing operation or the reduction processing operation estimated to have been applied to the original image 1 in order to obtain the input image 2. The reciprocal 1 / D of the used enlargement / reduction ratio Dn is applied to the obtained image 2 to reproduce the estimated original image 1 ′, and the estimated image reproduced by the original image reproduction unit 14. A digital watermark detection device 200 configured to detect the presence or absence of digital watermark information from the original image 1 ';
It is shown.

Further, the original image size detecting device 100 and the digital watermark detecting device 200 according to the present invention determine whether or not the obtained still image 2 is a still image formed by enlarging or reducing. It is also desirable to have the inspection state processing means 17 for the image to be determined. on the other hand,
In the present invention, the first and second Fourier transform means 3 and 8 may be individually provided,
Further, they may be configured to be commonly used by the same means.

In the original image size detecting apparatus 100 and the digital watermark detecting apparatus 200 according to the present invention, the digital watermark information detecting means 15 reads the digital watermark from the estimated original image 1 '. In this case, the digital watermark information of the reproduced estimated original image 1 ′ is stored in the original image digital watermark information storage unit 20 in which the digital watermark information of the reference original image 1 whose source is known in advance is stored. It is desirable to provide a digital watermark information match search unit 16 for determining whether or not the stored information is the same or similar, and further to provide a display unit 24 for displaying the search result of the digital watermark information match search unit 16. .

Here, the basic operation principle of the original image size detecting method and the digital watermark detecting method according to the present invention will be described in detail with reference to FIGS. First,
Conventionally developed techniques for embedding digital watermark information in a desired original image are roughly classified into two types: an image space and a Fourier space. That is, in the image space technique, the digital watermark information is directly embedded in the image value.
In the Fourier space technique, an image value is subjected to Fourier transform, and digital watermark information is embedded in the Fourier value of the image.

Each of these methods has advantages. For example, as shown in FIG. 2A, when digital watermark information A having a predetermined frequency value is embedded in the image 2 or 30 in the image space, the image in the image space is subjected to Fourier transform to perform Fourier space conversion. When the inner image 20 is used, FIG.
As shown in (B), the frequency of the embedded digital watermark information A is substantially uniformly distributed in the Fourier space.
The digital watermark information cannot be recognized.

On the other hand, as shown in FIG. 3B, after embedding appropriate digital watermark information B in the Fourier space image 20, the Fourier space image 20 is subjected to an inverse Fourier transform to form an image space image 30. When returned, the energy of the signal of the digital watermark information B (the frequency component value is
As shown in (A), the image is spread over the entire image space, and the quality of the original image is less deteriorated.

On the other hand, the characteristics of the digital watermark information embedded in the Fourier space image 20 as described above will be described below. First, if we consider one-dimensional signals,
The one-dimensional signal has a period of 8P as shown in FIG.
Consider a one-dimensional time series (P: pixel) with a length of 256P. As shown in FIG. 4 (B), when a sequence sampled at 128 points and having an enlargement / reduction ratio of 1.0 (that is, a sequence in which no enlargement / reduction operation has been performed) is subjected to Fourier transform, the frequency becomes as shown in FIG. As shown in FIG.
P / 8T = 16.

That is, FIG. 4C shows a cycle 8P (P: p
3) shows the state of the value of the frequency component when a one-dimensional time series of 256 pixels and a length of 256P is sampled at the original image size. Next, FIGS.
As shown in (C), the same one-dimensional series as described above is enlarged /
The waveform is as shown in FIG. 5B by reducing at a reduction ratio of 0.8 and sampling at 128 points.

When the series is Fourier-transformed, the frequency of the series is larger than 16, as shown in FIG. 5C. Further, as shown in FIG. 5A, the same one-dimensional series as described above is enlarged at a magnification of 1.2, and is sampled at 128 points. It is shown as follows.

Further, it can be seen that the frequency of the sequence after the Fourier transform of the sequence is smaller than 16, as shown in FIG. As is apparent from the above results, when a signal of a certain frequency is enlarged / reduced using a predetermined enlargement / reduction ratio and then resampled, the frequency of the signal becomes higher than the original frequency as shown in FIG. Become smaller,
It turns out that it becomes large.

The unit on the vertical axis in FIG. 7 is the intensity of the frequency component. That is, it can be seen that the position of the watermark signal embedded in the Fourier space deviates from the original position when the image is enlarged or reduced. Next, in the case of considering embedding digital watermark information in a Fourier-transformed image of the Fourier-transformed image in a two-dimensional image, as shown in FIGS. 8A and 8B, FIG. Consider a two-dimensional image.

The images shown in FIGS. 8A and 8B each have an image size of 256p × 256p.
Is a gray image with an image value of 128, each 128p
The block image is divided into four block images each having a size of × 128p, and each block image is subjected to Fourier transform. Then, a total of eight predetermined preset positions P of respective blocks in the Fourier transformed image are obtained. For example, a reference information signal S having a height of 20 (8-bit image) and a thickness of 1 pixel is arranged.

In FIG. 8, the gray symbol at the upper left corner of the image portion is the zero frequency component of the original image. afterwards,
The four block images in which the reference information signal S is embedded are respectively inverse Fourier-transformed to reconstruct an image in the image space, and from the block images in which the reference information signal S is embedded from the image in the image space. 128 × 128
The block in which the reference block is cut out and the image block is again subjected to the Fourier transform so that the position where the reference information signal S is embedded is the original reference information signal S indicated by a gray mark in FIGS. 8A and 8B. Indicates the embedding position P.

However, the image in the image space in which the reference information signal S is embedded has, for example, an enlargement / reduction ratio D of 0.8.
After the re-Fourier transform is performed after performing the reduction process, the frequency component value of the reference information signal S becomes as shown in FIG.
As shown by the black point P ′, the reference information signal S is moved inward from the position P where the reference information signal S was first embedded. Similarly, when the image in the image space in which the reference information signal S is embedded is subjected to a re-Fourier transform after performing an enlargement process in which the enlargement / reduction ratio D is 1.2, for example, the frequency of the reference information signal S is increased. It has been found that the component value moves outside or in the lateral direction from the position P where the reference information signal S was first embedded, as indicated by a black point P ′ in FIG. 8B.

As described above, the position of the watermark signal embedded in the image in the Fourier space of the predetermined image can be shifted laterally or outward and toward the center as the image is enlarged or reduced. Understand. The original image size detection method and the digital watermark detection method according to the present invention detect the image size of the original image by using characteristics related to the behavior of predetermined information embedded in the image in the Fourier space, The digital watermark information embedded in the original image can be detected.

The basic technical configuration of the image in the Fourier space and / or the digital watermark detection method according to the present invention is to detect the original image size using the Fourier space by applying the above principle. It is an encoder / decoder process. In other words, in the encoder operation,
For example, 1) The obtained inspection image is divided into blocks of size 128 × 128. At this time, a portion that is insufficient for the block size is supplemented with zero data.

2) Fourier transform is performed on each of the divided block images to convert a predetermined reference information signal S into a predetermined position P
Embed in place. The intensity of the embedded reference information signal is adjusted and embedded according to the height of the frequency component of the image at the position P. 3) Each block image in which the reference information signal S is embedded is subjected to inverse Fourier transform, and returns to the original block position of the obtained inspection image.

The subsequent decoder operations are as follows: 1) For the inspected image returned to the image in the image space,
When the enlargement / reduction ratio is, for example, 0.5 to 2.0, that is, when the enlargement / reduction ratio is 0.5 to 1.0, the process is a reduction process. In the case of 2.0, it is an enlargement process.

The size of each image to be enlarged / reduced is 128 × 128 pixels. 2) A plurality of block images obtained by using a plurality of different enlargement / reduction ratios, which have been subjected to enlargement or reduction processing, are again subjected to Fourier transform to form images in a Fourier space, and the respective images in the respective Fourier spaces are obtained. Embedding position P of the reference information signal S in the image individually from the image
The frequency component at is detected.

At this time, it is desirable to use an average value for the frequency component value. By integrating the obtained frequency component values or arranging them in descending order or ascending order according to the enlargement / reduction ratio, a graph as shown in FIG. 9 is obtained. 3) A series of graphs of the frequency component values after removing noise components such as the influence of the background (the influence of the integrated value of the original image frequency component at each embedding position) from the graph shown in FIG. In the above, it is determined whether or not there is a frequency component value protruding from the surrounding frequency component value.

The cause of the occurrence of a frequency component value protruding from the surrounding frequency component value on a series of graphs of the frequency component value by the above processing in the present invention is considered as follows. . That is, when the reference information signal is embedded in the image and the image is scaled up or down, the embedded frequency component (reference information signal) is out of the original, so that the image is enlarged and reduced at each size. By converting the size and detecting the frequency component value while embedding the reference information signal, when the image matches the size of the original image, the frequency component value becomes the maximum value.

4) Then, when the existence of the prominent frequency component value can be confirmed, the value of the enlargement / reduction ratio D corresponding to the prominent frequency component value is determined by the obtained inspection image. This is to determine the enlargement / reduction ratio D applied to the original image when the image size change processing is performed from the original image. That is, the image size of the original image that cannot be confirmed and is the source of the obtained inspection image can be determined by the method.

That is, referring to the graph of FIG. 9, the obtained inspected image is enlarged / reduced to its original image.
This means that the image has been enlarged after setting the reduction ratio D to 1.2. Therefore, by performing the reverse operation on the obtained inspection image, that is, in the above specific example, 1/1.
By performing the reduction operation using the enlargement / reduction ratio of 2,
The estimated original image 1 'can be reproduced.

Thereafter, digital watermark information is read out from the reproduced estimated original image 1 ′ by a known method, and compared with the digital watermark information of the original image 1 whose source is known in advance, thereby obtaining the image to be inspected. Can be distinguished from each other. Therefore, by using the original image size detecting method and the digital watermark detecting method according to the present invention, a predetermined original image is improperly subjected to an enlargement operation or a reduction operation, and the original image is The source of the original image can be confirmed for an image in which the embedded digital watermark information has been deleted or cannot be read, so that the image to be inspected incorrectly changes the original image whose source is known in advance. It is possible to determine whether the source image has been processed or the source image has been imitated completely separately from the known original image, and at any stage in the market It is also possible to determine whether the processing operation has been executed.

The apparatus 300 for detecting the method of processing the original image to be deformed and the apparatus 10 for detecting the original image size according to the present invention will be described below.
A specific example of the method for detecting the original image to be deformed, the method for detecting the original image size, and the method for detecting the digital watermark using the digital watermark detection apparatus 200 will be described in more detail. That is, first, a method for detecting a method of processing the original image 1 to be deformed according to the present invention is basically configured such that a predetermined reference information signal S Is embedded in the predetermined part P,
The Fourier space image 20 is inverse Fourier transformed back to the image space image 30, and then the image 3 in the image space 3
0 is a plurality of individual enlargement / reduction rates (D1, D2, D3,... D) selected from the enlargement / reduction rates D within a predetermined range.
n), the image size of the image 30 in the image space is individually changed, and then the image 30 in the image space in which the respective image sizes have been changed is Fourier-transformed again to obtain the image 40 in the Fourier space. None, in each Fourier space image 40, the frequency component of the reference information signal S previously embedded in the image 20 is detected, and the respective image space image 30 whose image size is changed is detected.
The frequency components of the reference information signal S in the re-Fourier space image 40 are represented by the enlargement / reduction rates D1, D2, D
3,... When arranged in order according to the magnitude of Dn, when the value of the frequency component at a certain enlargement / reduction ratio Dn can be recognized as being more prominent than the value of the surrounding frequency component Is a method of detecting a method of processing the original image to be deformed, which is configured to determine that the image to be inspected 2 is formed by enlarging / reducing a predetermined original image. The basic configuration of the original image size detection method according to the present invention is as follows. After embedding a predetermined reference information signal S in a predetermined region P in a Fourier space image 20 obtained by Fourier transforming the obtained inspection image 2, Image 20 in space
Is inverse Fourier transformed back to the image in the image space 30, and then the image in the image space 30 is converted into a plurality of individual enlargement / reduction ratios (D selected from enlargement / reduction ratios D within a predetermined range.
, D2, D3,... Dn), after individually changing the image size of the image 30 in the image space, the image 3 in the image space in which the respective image size is changed.
0 is again Fourier transformed into a Fourier space image 40, and in each Fourier space image 40, the frequency component of the reference information signal S previously embedded in the image 20 is detected, and the size of each image is determined. Has been changed, the re-Fourier space image 40 of the respective image space image 30 has been changed.
When the frequency components of the reference information signal S are sequentially arranged in accordance with the magnitudes of the enlargement / reduction rates D1, D2, D3,..., Dn, the corresponding frequency at a certain enlargement / reduction rate Dn When the value of the component protrudes from the value of the surrounding frequency component, the obtained image 2 is estimated at the enlargement / reduction ratio Dn corresponding to the frequency component value protruding from the value of the frequency component. This is an original image size detection method configured to determine that the size has been changed from the original image 1 '.

In other words, in the above two embodiments of the present invention, for example, when a person who has a copyright on a predetermined image embeds predetermined digital watermark information in the predetermined image and sells it on the market, (In this case, the image that is marketed by embedding the predetermined electronic watermark information in the predetermined image is the original image 1, and the original image 1 is an image whose source is clear.) An image 2 similar or similar to or completely identical to the image 1 whose source is unknown and whose digital watermark information has been deleted or whose digital watermark information cannot be read, Estimated image 1 '
In the case where an image for which the estimated original image 1 ′ cannot be confirmed is commercially available, the image 2 is obtained as the image to be inspected 2, and the image to be inspected 2 is the original image size according to the present invention. Using the detection method and the digital watermark detection method, it is possible to confirm the estimated original image 1 ′ that is the source.

Therefore, in the present invention, first, it is necessary that the original image 1 is an image whose source is known. Further, in the present invention, the obtained image 2 has an image substantially similar to the original image 1, and the digital watermark information has been deleted or the digital watermark information cannot be read. It is an image in a state.

Further, the original image 1 and the obtained image 2 in the present invention are still images. On the other hand, the inspection target image information 2 which is considered to have been subjected to the unauthorized deformation processing in the original image size detection method and the electronic watermark detection method according to the present invention is a still image, and the image information 2 Is image information estimated to have undergone an enlargement / reduction processing operation from estimated image information 1 ′.

Therefore, according to the present invention, as shown in FIG. 1, when inspecting the image 2 to be inspected, whether the image 2 to be inspected is subjected to the enlargement / reduction deformation processing, It is preferable to add a function of determining whether or not the image has been subjected to the rotational deformation process to determine whether or not the image has been subjected to the rotational deformation process, as the image processing state determination unit 17 that can identify whether or not the image has undergone the process.

Specifically, it is desirable to use the configuration as described above. In the present invention, as described above, it is first determined whether or not the obtained image to be inspected 2 has been subjected to rotational deformation processing, and the image to be inspected 2 is subjected to rotational deformation processing. If it can be confirmed that there is, the subsequent processing operation is stopped, but if it can be confirmed that the inspection image 2 has not been subjected to the rotational deformation processing, the subsequent processing operation It is determined whether the inspection image 2 has been subjected to enlargement / reduction deformation processing.

That is, in the present invention, the inspection image 2 is enlarged / removed from the inspection image 2 from which the digital watermark information which seems to have undergone the deformation processing cannot be read.
It is determined whether the image has been subjected to a reduction deformation process or another deformation process other than the rotation deformation process. Further, the enlargement / reduction ratio D used in the present invention.
Although the predetermined range of is not particularly specified,
As a practical range, for example, 0.5 to 2.0 is desirable.

It is desirable that the information on the enlargement / reduction ratio D be stored in the enlargement / reduction ratio storage means 6 in advance. Further, in the present invention, it is desirable to use an average value for the detected frequency component values, and when judging the presence of the peak value of the frequency component values, all the detected frequency component values are used. It is desirable to remove the background value from the component, and it is also desirable to provide a noise removing means 13 for that purpose.

Also, the digital watermark detection method according to the present invention
According to the above-described original image size detection method, what kind of enlargement / reduction ratio D is used to form the inspected image 2 with respect to the original image 1 ′ whose source is unknown is determined. Since it is clear, the reciprocal 1 / D of the obtained enlargement / reduction ratio D is applied to the obtained image 2 to reproduce the estimated original image 1 ′, and the estimated original image 1 ′ is attached with watermark information. It is configured to detect the presence or absence of.

When performing the operation of returning the obtained inspected image 2 to the estimated original image 1 ', for example, a conventionally known affine transformation method or the like can be used. A more specific example of the above-described method of detecting the original image to be deformed, the method of detecting the original image size, and the method of detecting the digital watermark according to the present invention will be described with reference to the flowchart of FIG.

That is, in the original image size detecting method according to the present invention, first, after starting, in step ST-1, the inspection target in which the digital watermark information similar to the original image 1 whose source is known cannot be read. While obtaining the image 2, preferably, in step ST-step ST-101, it is determined whether the inspection image 2 has been subjected to rotational deformation processing. If so, a first step is performed, which says to proceed below.

Next, in step ST-2, a second step of Fourier transforming the obtained image to be inspected 2 into a first Fourier space image 20 is executed, and then in step ST-3. Go to the first image in Fourier space 2
A third step of embedding a predetermined reference information signal S in a predetermined part P of 0 is executed. Thereafter, the process proceeds to step ST-4, where a fourth step of performing an inverse Fourier transform on the Fourier space image 20 in which the reference information signal S is embedded and returning it to the image space image 30 is executed, and then a step ST-5 is performed. In this case, the image 30 in the image space is stored in the enlargement / reduction ratio storage means 6 and the enlargement / reduction ratio D within the predetermined range is stored.
.. Dn selected from (D1, D2, D3,... Dn), the image size of the image 30 in the image space is determined using the respective enlargement / reduction rates D1, D2, D3,. A fifth step of individually changing is executed.

Then, the process shifts to step ST-6 where the image 3 in the image space whose image size is changed is changed.
0 is again Fourier transformed to obtain a second Fourier space image 4
0 is executed, and then the step ST
At -7, a seventh step of individually detecting the frequency component value of the reference information signal S previously embedded in the image 20 in the respective Fourier space images 40 is executed.

Thereafter, in step ST-8, the detected frequency component value of each of the detected reference information signals S is
An eighth step is executed to obtain a waveform as shown in FIG. 9 by sequentially arranging or integrating according to the magnitude of the enlargement / reduction ratio D. Next, the process proceeds to step ST-9, where it is determined whether or not the value of the frequency component at a certain enlargement / reduction ratio Dn in the arrayed frequency component values is more prominent than the value of the surrounding frequency component. And if NO,
It becomes an end, and it is presumed that the inspection image 2 has not been subjected to the rotation deformation processing or the enlargement / reduction deformation processing and has been subjected to other deformation processing operations.

On the other hand, if YES in step ST-9, that is, if it is determined that the frequency component value protrudes beyond the value of the surrounding frequency component, the process proceeds to step ST-90. Then, it is determined that the image to be inspected 2 has been subjected to the enlargement / reduction deformation operation, and if necessary, a ninth step of displaying the result on an appropriate display means 24 is executed, The method of detecting a method of processing the original image to be deformed according to the present invention is executed.

On the other hand, as a specific example of the original image size detection method according to the present invention, as shown in the flowchart of FIG. Until -9, the same process is executed as it is, and the process goes from ST-9 to ST-90 or
Proceeding directly from T-9 to ST-91, the enlargement / reduction ratio Dn corresponding to the frequency component value is calculated from the estimated original image 1 ′ whose acquired inspection image 2 is unknown from the original enlargement / reduction ratio Dn. The step 9a of judging that the size has been changed is executed.

In step ST-9, when judging the presence of the peak value of the frequency component value, it is preferable to use an operation of removing the background value from all the detected frequency components. . The above operation procedure is a basic configuration of the original image size detection method according to the present invention.

In the digital watermark detection method according to the present invention, the same steps are performed from the first step to the ninth step of the above-described original image size detection method. Step ST by the image size detection method
Estimated original image 1 ′ on which the image to be inspected 2 is formed at −91
Since the enlargement / reduction ratio D is determined, the reciprocal of the enlargement / reduction ratio D is applied to the obtained inspection image 2 in step ST-10 using the enlargement / reduction ratio D. A tenth step of reproducing the estimated original image 1 'is executed, and then, in step ST-11, an eleventh step of detecting the presence or absence of watermark information on the unknown source image 1' Is executed.

Thereafter, the process proceeds to step ST-12,
The electronic watermark information obtained from the estimated source image 1 ′ whose source is unknown and the electronic watermark information storage unit 2 of the original image 1 shown in FIG.
6 is compared with the digital watermark information of the original image 1 whose source is clear, and it is determined whether or not the two match. If NO in step ST-12, the obtained estimated source image 1 ′ of the inspected image 2 can be estimated to be an image whose source is imitated regardless of the clear original image 1, and If “YES” in ST-12, the obtained estimated source image 1 ′ of the inspected image 2 is fraudulently falsified by the enlargement / reduction processing operation of the original image 1 whose source is clear, and the digital watermark information is deleted. It can be estimated that it is.

Next, a more specific example of the above-described method of detecting a method of processing the original image to be deformed according to the present invention will be described in detail with reference to FIGS. Now, FIG.
As shown in (A), the pixel value is 128 and the size is 256.
It is assumed that a gray image of × 256 pixels (pixels) is obtained as the image 2 to be inspected. First, as shown in FIG. 11A, the gray image of the image 2 to be inspected is 128 × 128.
Divided into blocks.

Next, as shown in FIG. 11 (B), each block image is subjected to Fourier transform, and a frequency component having an intensity of about 5 units is embedded as a reference information signal in a designated portion (eight locations in this specific example). . That is, the following eight address positions (8, 16), (16, 8),
112, 16), (120, 8), (8, 112),
(16, 120), (112, 120), (120, 1)
12), the frequency component having the predetermined strength is embedded as a reference information signal.

Also, by subjecting each block image to which the reference information signal has been added to inverse Fourier transform, the reference information signal is embedded in the original image. As described above, when the reference information signal is embedded in the Fourier space and returned to the image space by the inverse Fourier transform, the energy of the embedded reference information signal is diffused in the image space. Becomes inconspicuous.

Thereafter, as shown in FIG. 11C, the image (a) in which the reference information signal is embedded is reduced by an η = 1.2 ratio (η: original image size / reduced image size), and the image (a) is reduced. Obtain b). Next, as shown in FIG. 11D, from the reduced image (b), as shown in FIG.
By cutting out a block partial image (c) having a size of 8 × 128 and performing Fourier transform, as shown in FIG. 12, the position of the first embedded reference information signal (black dot) moves inward and (Black triangle point).

The black dot at the upper left corner in FIG. 12 represents the zero frequency component of the original image. Next, as shown in FIG. 11D, the image in which the reference information signal is embedded is enlarged by η = 0.8 ratio (η: original image size / reduced image size) to obtain an image (d). Thereafter, a block partial image (e) having a size of 128 × 128 indicated by a dotted line is cut out from the enlarged original image and subjected to Fourier transform to obtain a position ● ( It turns out that the black dot moves outward and moves to ◆ (black diamond point).

Note that a black dot at the upper left corner in FIG. 12 represents the zero frequency component of the original image. Then, the size of the image obtained in the market is changed in the range of 0.5 to 1.5 as shown in FIG. (In this specific example, the exchange rates of the x-axis and the y-axis are assumed to be the same.) Then, as shown in FIGS. A block partial image (e) having a size of 128 × 128 at the upper left is cut out and subjected to Fourier transform.

Then, the average of the eight frequency components shown in FIG. 11B is calculated for the Fourier-transformed block image, as shown in FIGS. 14, 15 and 16. The average value of the eight frequency components calculated from each image with the enlargement / reduction ratio changed in the range of 0.5 to 1.5 is obtained. Then, the obtained image is enlarged by detecting a point where the maximum value or the frequency component value protrudes from the value of the surrounding frequency component from the graph of the frequency component average value obtained as described above. It is possible to judge whether the image has been subjected to the reduction / enlargement processing or to calculate the enlargement / reduction ratio.

FIG. 14 shows a graph of the average value of the frequency components detected when the obtained image is not enlarged or reduced, and the horizontal axis shows the conversion rate of the image size. FIG. 15 shows a graph of the average value of the frequency components detected when the obtained image is reduced at the rate of 1.2, and the horizontal axis shows the conversion rate of the image size.

FIG. 16 is a graph showing the average value of the frequency components detected when the obtained image is enlarged at a ratio of 0.8, and the horizontal axis shows the conversion ratio of the image size. It is. Since an actual image has more frequency components than the image of the specific example described above, the frequency component graph detected from the position where the reference information signal is embedded is a complicated graph as shown in FIG.

That is, the reason is that the original frequency component of the image is the background value. Therefore, when detecting the peak value of the frequency component value, it is necessary to consider the influence of the background value. Is as described above. In another aspect of the present invention, a first step of obtaining an image to be inspected from which digital watermark information similar to an original image whose source is known cannot be read, and the obtained image to be inspected is subjected to Fourier A second step of converting into a first Fourier space image, a third step of embedding a predetermined reference information signal in a predetermined portion of the first Fourier space image, and embedding the reference information signal. Fourth step of performing an inverse Fourier transform on the image in the Fourier space to return to an image in the image space, and calculating a plurality of individual enlargement / reduction ratios selected from the enlargement / reduction ratios in the predetermined range. A fifth step of individually changing the image size of the image in the image space using the Fourier transform of the image in the image space in which the respective image sizes have been changed, and a second image in the Fourier space; The sixth step to do A seventh step of individually detecting the frequency component value of the reference information signal previously embedded in the image in the respective Fourier space images, the frequency component value of the detected individual reference information signal; To the expansion /
An eighth step of sequentially arranging the frequency components in accordance with the magnitude of the reduction ratio, wherein the value of the frequency component at a certain enlargement / reduction ratio in the arranged frequency component values is larger than the value of the surrounding frequency component. Judgment is made as to whether or not it is protruding, and if it is determined that the frequency component value is protruding from the value of the surrounding frequency component, the obtained image is resized from the original image by the enlargement / reduction ratio. And a ninth step of judging that the original image has been processed.

Further, as still another embodiment of the present invention, a first step of obtaining an image to be inspected from which electronic watermark information similar to the original image whose source is known cannot be read out, A second step of Fourier-transforming the inspection image into a first Fourier space image; a third step of embedding a predetermined reference information signal in a predetermined portion of the first Fourier space image; A fourth step of performing an inverse Fourier transform on the image in the Fourier space in which the signal is embedded and returning the image in the image space, and a plurality of individual enlargements selected from the enlargement / reduction ratio of the image in the image space in a predetermined range A fifth step of individually changing the image size of the image in the image space using the / reduction ratio, and performing a Fourier transform on the image in the image space in which the respective image sizes have been changed again to perform a second Fourier transform. Image in space A sixth step of individually detecting the frequency component value of the reference information signal previously embedded in the image in the respective Fourier space image, and a step of detecting the detected individual reference An eighth step of sequentially arranging the frequency component values of the information signal in accordance with the magnitude of the enlargement / reduction rate, and the value of the frequency component at a certain enlargement / reduction rate in the arranged frequency component values Is determined to be greater than the value of the surrounding frequency component, and if the frequency component value is determined to be greater than the value of the surrounding frequency component, the A ninth step of judging the corresponding enlargement / reduction ratio to be such that the obtained image has been resized by the enlargement / reduction ratio from the original image. A first step of obtaining an image to be inspected in which digital watermark information similar to the original image whose source is known cannot be read out. A second step of Fourier-transforming the obtained image to be inspected into a first Fourier space image, and a third step of embedding a predetermined reference information signal in a predetermined portion of the first Fourier space image A fourth step of performing an inverse Fourier transform on the image in the Fourier space in which the reference information signal is embedded and returning the image in the image space to an image in the image space, a plurality of images selected from the enlargement / reduction ratio within a predetermined range; A fifth step of individually changing the image size of the image in the image space using the individual enlargement / reduction ratios of the image space, and performing a Fourier transform again on the image in the image space in which the respective image size has been changed. The second foo A sixth step of forming an image in the Fourier space; a seventh step of individually detecting the frequency component value of the reference information signal previously embedded in the image in the respective Fourier space image; An eighth step of sequentially arranging the frequency component values of the individual reference information signals in accordance with the magnitude of the enlargement / reduction ratio, wherein the frequency component values of the reference information signal are arranged at a certain enlargement / reduction ratio in the arranged frequency component values. If the value of the frequency component is
It is determined whether or not the frequency component value protrudes from the value of the surrounding frequency component. If the frequency component value is determined to protrude from the value of the surrounding frequency component, the frequency component value corresponds to the frequency component value. A ninth step of judging that the obtained image has been resized from the original image by the enlargement / reduction ratio, and a step of determining the enlargement / reduction ratio obtained in the ninth step. A digital watermark for the original image, comprising: a tenth step of reproducing the original image by applying the reciprocal to the obtained image; and an eleventh step of detecting the presence or absence of watermark information on the original image. This is a recording medium on which a program for executing the detection method is recorded.

[0081]

Since the original image size detection method and the digital watermark detection method according to the present invention employ the above-described technical configuration, the original image size is a still image including predetermined digital watermark information. Detects the original digital watermark information from the illegitimately processed still image information on the market after performing the deformation process by the enlargement / reduction operation on the image information, and illegally infringes the copyright of the image information. It has become possible to provide a method and an apparatus which enable protection from the environment.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the configuration of a specific example of an original image deformation processing method detection device, an original image size detection device, and a digital watermark detection device according to the present invention.

FIG. 2 is a diagram illustrating an example of a state of a frequency of an embedded signal in a case where an image in which a desired signal is embedded in an image in an image space is Fourier-transformed into an image in a Fourier space; .

FIG. 3 is a diagram illustrating an example of a state of a frequency of an embedded signal in a case where an image in which a desired signal is embedded in an image in the Fourier space is inverse Fourier-transformed into an image in the image space; It is.

FIGS. 4A to 4C are diagrams showing frequency positions when signals are cut out from one-dimensional signals with the same image size and subjected to Fourier transform.

FIGS. 5A to 5C are diagrams illustrating frequency positions when a signal is cut out from a one-dimensional signal with a reduced image size and subjected to Fourier transform.

FIGS. 6A to 6C are diagrams showing frequency positions when a signal is cut out from a one-dimensional signal with an enlarged image size and subjected to Fourier transform. FIG.

FIG. 7 is a diagram collectively displaying the frequency positions obtained from FIGS. 4 to 6;

FIGS. 8A and 8B show a state of a change in an embedding position of a two-dimensional signal, that is, when an image including an embedding signal is subjected to enlargement processing and reduction processing and Fourier-transformed. FIG.

FIG. 9 is a graph showing an example of a change in the enlargement / reduction ratio and an average value of frequency components at an embedding position of the embedding signal when an image including an embedding signal is subjected to an enlargement process and a reduction process and subjected to Fourier transform. It is.

FIG. 10 is a flowchart showing a specific example of an execution operation procedure of a method of detecting an original image to be deformed, a method of detecting an original image size, and a method of detecting a digital watermark according to the present invention.

FIG. 11 is a diagram illustrating a method of detecting a deformed processing method of an original image, a method of detecting an original image size, and a method of detecting an electronic watermark according to the present invention; embedding a reference information signal; It is a figure explaining the block image extraction method etc. after a process.

FIG. 12 is a diagram illustrating a change in the position of a reference information signal embedded in an original image by an enlargement / reduction deformation process;

FIG. 13 is a diagram illustrating a state in which enlargement / reduction deformation processing is performed on an obtained original image.

FIG. 14 is a graph of frequency component average values detected when enlargement / reduction processing is not performed on an obtained image.

FIG. 15 is a graph of frequency component averages detected when the obtained image is reduced at a rate of 1.2.

FIG. 16 is a graph of a frequency component average value detected when the obtained image is enlarged at a ratio of 0.8.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Original image whose source is clear 1 '... Original image whose source is unclear, estimated original image 2 ... Inspection image, obtained image 3 ... First Fourier transform processing means 4 ... Reference information signal embedding means 5 ... Inverse Fourier Conversion processing means 6 ... Enlargement / reduction rate storage means 7 ... Image size changing means 8 ... Second Fourier transform processing means 9 ... Frequency component value detection means 10 ... Frequency component value array means, integration means 11 ... Protruding frequency component value judgment Means 12: Original image size determining means 13: Noise removing means 14: Original image reproducing means 15: Digital watermark information detecting means 16 ... Digital watermark information coincidence determining means 17: Processing state determining means of inspected image 18: Central processing means 20 ... First Fourier space image 21 ... Enlargement / reduction deformation processing determination means 22 ... Display means 23 ... Rotation deformation processing determination result display means 24 ... Digital watermark matching result table Means 26 ... electronic watermark information storage unit 30 ... image space image 40 ... second Fourier space image 100 ... original image size detecting device 200 ... the transformation processing method detecting apparatus of the digital watermark detection apparatus 300 ... original image

Claims (27)

[Claims] After embedding a predetermined reference information signal in a predetermined region in a Fourier-space image obtained by Fourier-transforming the obtained image to be inspected, the image in the Fourier space is inverse Fourier-transformed and returned to an image in the image space. Then, after individually changing the image size of the image in the image space using the plurality of individual enlargement / reduction ratios selected from the enlargement / reduction ratios within a predetermined range, The image in the image space in which the respective image sizes have been changed is again Fourier transformed into an image in the Fourier space, and in each image in the Fourier space, the frequency of the reference information signal embedded earlier in the image Detecting the components, the frequency components of the plurality of reference information signals in the re-Fourier space image of the plurality of types of image space images in which the respective image sizes have been changed, If the value of the frequency component at a certain enlargement / reduction ratio can be recognized as being more prominent than the value of the surrounding frequency component when the pixels are sequentially arranged according to the magnitude of the enlargement / reduction ratio, A method for detecting a method of processing a deformed original image, wherein the image to be inspected is determined to have been formed by subjecting a predetermined original image to enlargement / reduction deformation processing. 2. The method according to claim 1, wherein the inspected image is an image that is the same as or approximated to the original image whose predetermined source is known. Method. 3. The method according to claim 1, wherein the original image and the obtained image are still images. 4. The method according to claim 1, wherein the predetermined range of the enlargement / reduction ratio is 0.5 to 2.0. Detection method. 5. A method of determining whether a value of a frequency component is more prominent than a value of a surrounding frequency component, removing a background value from all detected frequency components. A method for detecting a method of processing a deformed original image according to claim 1. 6. Embedding a predetermined reference information signal in a predetermined portion in a Fourier-space image obtained by Fourier-transforming the obtained image to be inspected, and then performing an inverse Fourier transform on the Fourier-space image to return to the image-space image. Then, after individually changing the image size of the image in the image space using the plurality of individual enlargement / reduction ratios selected from the enlargement / reduction ratios within a predetermined range, The image in the image space in which the respective image sizes have been changed is again Fourier transformed into an image in the Fourier space, and in each image in the Fourier space, the frequency of the reference information signal embedded earlier in the image The frequency component of the reference information signal in the re-Fourier space image of the image in the image space in which the image size has been changed is adjusted to the magnitude of the enlargement / reduction ratio. Therefore, when the values of the frequency component at a certain enlargement / reduction ratio protrude more than the values of the surrounding frequency components when they are sequentially arranged, the frequency that protrudes from the value of the relevant frequency component An original image size detection method, comprising: judging that the obtained image has undergone enlargement / reduction processing or reduction / reduction processing from the original image at the enlargement / reduction ratio corresponding to the component value. 7. The original image size detecting method according to claim 6, wherein the inspected image and the original image are images whose sources are known. 8. The original image size detecting method according to claim 6, wherein the obtained inspection image has an image substantially similar to the original image. 9. The original image size detecting method according to claim 6, wherein the original image and the obtained inspection image are still images. 10. The original image size detection method according to claim 6, wherein the predetermined range of the enlargement / reduction ratio is 0.5 to 2.0. 11. The source according to claim 6, wherein a background value is removed from all detected frequency components when determining the presence of the peak value of the frequency component value. Image size detection method. 12. The original image size detecting method according to claim 6, wherein the reciprocal of the original image size enlargement / reduction ratio of the obtained acquired image to be inspected is applied to the acquired image to be inspected to convert the original image. A digital watermark detection method characterized by reproducing and detecting the presence or absence of watermark information on the original image. 13. An apparatus for detecting the size of an original image from an acquired image, wherein the first Fourier transform processing is performed on the acquired image to form a first Fourier space image. Means, reference information signal embedding means for embedding a predetermined reference information signal at a predetermined position with respect to the Fourier space image of the Fourier-transformed obtained image, and reversing the Fourier space image in which the reference information signal is embedded. Inverse Fourier transform processing means for performing Fourier transform processing to convert the image into an image in the image space; enlargement / reduction rate storage means for storing a plurality of types of enlargement / reduction rates for changing the image size of the image in the image space; A desired enlargement / reduction ratio is sequentially selected from the enlargement / reduction ratio storage means, and the selected plurality of individual enlargement / reduction ratios are individually used to obtain an image of the image in the image space. Image size changing means for sequentially changing the size of each of the images, a second Fourier transform for forming a plurality of second Fourier space images by performing a Fourier transform process again on each of the images in the image space having the changed image size Processing means, frequency component value detecting means for detecting a frequency component value related to the reference information signal in each of the second Fourier space images, a plurality of the plurality of the four Fourier space images obtained from the respective second Fourier space images. Frequency component value arranging means for arranging the frequency component values in a predetermined order according to the enlargement / reduction ratio, wherein a plurality of frequency component values arranged in a series form frequency component values protruding from peripheral frequency component values. A protruding frequency component value judging unit for judging whether or not it is present, based on the output value output from the protruding frequency component value judging unit, The original image is constituted by an original image enlargement / reduction processing determining means for determining that the image is formed by performing enlargement / reduction deformation processing on a predetermined original image. Processing method detection device. 14. The original image according to claim 13, wherein said prominent frequency component value determining means includes a noise removing means for removing a background value from said plurality of frequency component values. Processing method detection device. 15. An apparatus for detecting the size of an original image from an obtained image, wherein the first Fourier transform processing is performed on the obtained image to form a first Fourier space image. Means, a reference information signal embedding means for embedding a predetermined reference information signal at a predetermined position with respect to the Fourier space image of the obtained Fourier-transformed image, and reversing the Fourier space image in which the reference information signal is embedded. Inverse Fourier transform processing means for performing Fourier transform processing to convert the image into an image in the image space; enlargement / reduction rate storage means for storing a plurality of types of enlargement / reduction rates for changing the image size of the image in the image space; A desired enlargement / reduction ratio is sequentially selected from the enlargement / reduction ratio storage means, and the selected plurality of individual enlargement / reduction ratios are individually used to obtain an image of the image in the image space. Image size changing means for sequentially changing the size of each of the images, a second Fourier transform for forming a plurality of second Fourier space images by performing a Fourier transform process again on each of the images in the image space having the changed image size Processing means, frequency component value detecting means for detecting a frequency component value related to the reference information signal in each of the second Fourier space images, a plurality of the plurality of the four Fourier space images obtained from the respective second Fourier space images. Frequency component value arranging means for arranging the frequency component values in a predetermined order according to the enlargement / reduction ratio, wherein the frequency component values protruding from the peripheral frequency component values in the plurality of frequency component values arranged in a series are A protruding frequency component value determining means for determining whether or not the protruding frequency component value is present; The corresponding enlargement / reduction ratio is constituted by an original image size determination means for determining that the obtained image is the enlargement / reduction ratio used when the size of the obtained image is changed from the original image. Original image size detection device. 16. The original image size detecting device according to claim 15, wherein said prominent frequency component value determining means includes a noise removing means for removing a background value from said plurality of frequency component values. . 17. The reciprocal of the enlargement / reduction ratio determined by the original image size detection device in addition to the original image size detection device according to claim 15 or 16, applied to the obtained inspection image. And an original image reproducing means for reproducing the original image, and a digital watermark information detecting means for detecting the presence or absence of digital watermark information from the original image reproduced by the original image reproducing means. Digital watermark detection device. 18. A first step of obtaining an image to be inspected from which digital watermark information similar to an original image whose source is known cannot be read out, and performing a Fourier transform on the obtained image to be inspected to obtain a first Fourier space. A second step of forming an internal image, a third step of embedding a predetermined reference information signal in a predetermined portion of the first Fourier space image, and a Fourier space image in which the reference information signal is embedded.
A fourth step of performing an inverse Fourier transform to return the image in the image space to the image in the image space, using the plurality of individual enlargement / reduction ratios selected from the enlargement / reduction ratios in the predetermined range for the image in the image space; A fifth step of individually changing the image size of the in-space image; a sixth step of performing a Fourier transform on the image in the image space in which the respective image sizes have been changed again to obtain a second Fourier space image; A seventh step of individually detecting a frequency component value of the reference information signal previously embedded in the image in each of the Fourier space images; a frequency component value of the detected individual reference information signal; An eighth step of sequentially arranging the frequency components in accordance with the magnitude of the enlargement / reduction rate, and setting the value of the frequency component at a certain enlargement / reduction rate in the arranged frequency component values to the surrounding frequency Component value It is determined whether the frequency component value is higher than the value of the surrounding frequency component. If the frequency component value is higher than the surrounding frequency component value, the image to be inspected is enlarged / reduced to a predetermined original image. A ninth step of judging that the original image has been transformed. 19. A first step of obtaining an inspected image from which digital watermark information similar to an original image whose source is known cannot be read out, and performing a Fourier transform on the obtained inspected image to obtain a first Fourier space. A second step of embedding a predetermined reference information signal in a predetermined part of the first Fourier space image; a Fourier space image having the reference information signal embedded therein;
A fourth step of performing an inverse Fourier transform to return the image in the image space to the image in the image space, using the plurality of individual enlargement / reduction ratios selected from the enlargement / reduction ratios in the predetermined range for the image in the image space; A fifth step of individually changing the image size of the in-space image; a sixth step of performing a Fourier transform on the image in the image space in which the respective image sizes have been changed again to obtain a second Fourier space image; A seventh step of individually detecting a frequency component value of the reference information signal previously embedded in the image in each of the Fourier space images; a frequency component value of the detected individual reference information signal; An eighth step of sequentially arranging the frequency components in accordance with the magnitude of the enlargement / reduction rate, and setting the value of the frequency component at a certain enlargement / reduction rate in the arranged frequency component values to the surrounding frequency Component value It is determined whether or not the frequency component value is more protruding than the value of the surrounding frequency component. A ninth step of judging that the obtained image has been resized from the original image by the enlargement / reduction ratio. 20. The method according to claim 19, wherein the original image is an image whose source is known. 21. The original image size detecting method according to claim 19, wherein the obtained image has an image substantially similar to the original image. 22. The original image size detection method according to claim 19, wherein the original image and the obtained image are still images. 23. The method according to claim 19, wherein the predetermined range of the enlargement / reduction ratio is 0.5 to 2.0.
23. The method for detecting an original image size according to any one of claims 21 to 22. 24. The source according to claim 19, wherein, when judging the presence of the peak value of the frequency component value, the background value is removed from all the detected frequency components. Image size detection method. 25. A first step of obtaining an image to be inspected from which digital watermark information similar to an original image whose source is known cannot be read out, and performing a Fourier transform on the obtained image to be inspected to obtain a first Fourier space. A second step of forming an internal image, a third step of embedding a predetermined reference information signal in a predetermined portion of the first Fourier space image, and a Fourier space image in which the reference information signal is embedded.
A fourth step of performing an inverse Fourier transform to return the image in the image space to the image in the image space, using the plurality of individual enlargement / reduction ratios selected from the enlargement / reduction ratios in the predetermined range for the image in the image space; A fifth step of individually changing the image size of the in-space image; a sixth step of performing a Fourier transform on the image in the image space in which the respective image sizes have been changed again to obtain a second Fourier space image; A seventh step of individually detecting a frequency component value of the reference information signal previously embedded in the image in each of the Fourier space images; a frequency component value of the detected individual reference information signal; An eighth step of sequentially arranging the frequency components in accordance with the magnitude of the enlargement / reduction ratio, wherein the value of the frequency component at a certain enlargement / reduction ratio in the arranged frequency component values is changed to the surrounding frequency component The value of Also, it is determined whether the frequency component value is protruding, and when it is determined that the frequency component value protrudes from the value of the surrounding frequency component, the enlargement / reduction ratio corresponding to the frequency component value is determined. A ninth step of judging that the obtained image has been resized by the enlargement / reduction ratio from the original image, and applying a reciprocal of the enlargement / reduction ratio obtained in the ninth step to the obtained image. A digital watermark detecting method for the original image, comprising: a tenth step of reproducing the original image by using the digital camera and an eleventh step of detecting presence / absence of watermark information on the original image. 26. A first step of obtaining an image to be inspected from which digital watermark information similar to an original image whose source is known cannot be read out, and performing a Fourier transform on the obtained image to be inspected to obtain a first Fourier space. A second step of embedding a predetermined reference information signal in a predetermined part of the first Fourier space image; a Fourier space image having the reference information signal embedded therein;
A fourth step of performing an inverse Fourier transform to return the image in the image space to the image in the image space, using the plurality of individual enlargement / reduction ratios selected from the enlargement / reduction ratios in the predetermined range for the image in the image space; A fifth step of individually changing the image size of the in-space image; a sixth step of performing a Fourier transform on the image in the image space in which the respective image sizes have been changed again to obtain a second Fourier space image; A seventh step of individually detecting a frequency component value of the reference information signal previously embedded in the image in each of the Fourier space images; a frequency component value of the detected individual reference information signal; An eighth step of sequentially arranging the frequency components in accordance with the magnitude of the enlargement / reduction rate, and setting the value of the frequency component at a certain enlargement / reduction rate in the arranged frequency component values to the surrounding frequency Component value It is determined whether or not the frequency component value is higher than the surrounding frequency component value. If the frequency component value is determined to be higher than the surrounding frequency component value, the enlargement / reduction ratio corresponding to the frequency component value is determined. A ninth step of determining that the obtained image has been resized from the original image by the enlargement / reduction ratio; and a program for executing an original image size detection method comprising the following steps: recoding media. 27. A first step of obtaining an inspected image from which digital watermark information similar to an original image whose source is known cannot be read out, and performing a Fourier transform on the obtained inspected image to obtain a first Fourier space. A second step of embedding a predetermined reference information signal in a predetermined part of the first Fourier space image; a Fourier space image having the reference information signal embedded therein;
A fourth step of performing an inverse Fourier transform to return the image in the image space to the image in the image space, using the plurality of individual enlargement / reduction ratios selected from the enlargement / reduction ratios in the predetermined range for the image in the image space; A fifth step of individually changing the image size of the in-space image; a sixth step of performing a Fourier transform on the image in the image space in which the respective image sizes have been changed again to obtain a second Fourier space image; A seventh step of individually detecting a frequency component value of the reference information signal previously embedded in the image in each of the Fourier space images; a frequency component value of the detected individual reference information signal; An eighth step of sequentially arranging the frequency components in accordance with the magnitude of the enlargement / reduction ratio, wherein the value of the frequency component at a certain enlargement / reduction ratio in the arranged frequency component values is changed to the surrounding frequency component The value of It is also determined whether or not the frequency component value protrudes, and when it is determined that the frequency component value protrudes from the value of the surrounding frequency component, the enlargement / reduction ratio corresponding to the frequency component value is determined. A ninth step of judging that the obtained image has been resized by the enlargement / reduction ratio from the original image, and applying a reciprocal of the enlargement / reduction ratio obtained in the ninth step to the obtained image. Program for executing a digital watermark detection method for an original image, comprising: a tenth step of reproducing the original image by using a digital camera, and an eleventh step of detecting the presence or absence of watermark information on the original image. Recording medium on which is recorded.