JP4922965B2 - Digital watermark generation apparatus, digital watermark generation method, digital watermark generation program, digital watermark detection apparatus, and digital watermark detection program - Google Patents

Description

  The present invention relates to a digital watermark generation device, a digital watermark generation method, a digital watermark generation program, a digital watermark detection device, and a digital watermark detection program. For example, the present invention relates to a method for embedding a signal pattern in orthogonal space using orthogonal transformation.

In recent years, research on digital watermarks in which watermark information is embedded in image data (digital images) has been actively conducted.
When image data in which watermark information is embedded is printed, the watermark is superimposed on the image and printed, but the watermark is printed in a color and gradation that does not impair the visibility of the user.
The watermark can be detected by taking an image with a camera having a digital watermark detection function. Thus, additional information can be embedded in an image by using a digital watermark.

Various methods for embedding watermark information in image data have been proposed, and one of them uses orthogonal transformation. In this method, watermark information is embedded in an orthogonal space different from the image space.
Examples of the orthogonal transform include a discrete Fourier transform, a discrete cosine transform, and a wavelet transform.

  Of these orthogonal transforms, a space composed of transform coefficients after discrete Fourier transform is also called a frequency space, and a space composed of transform coefficients of other transforms is not called a frequency space but is broadly called a frequency space. In some cases, these are hereinafter collectively referred to as frequency space.

More specifically, when watermark information is defined in a frequency space and converted into an image space, a watermark image corresponding to the watermark information defined in the frequency space is generated. When this watermark image is inversely transformed into the frequency space, the watermark information is restored, whereby the watermark information can be detected.
Here, if the watermark information in the frequency space is set to be a watermark image that is inconspicuous or negligible for human vision when converted into the image space, the designability or visibility of the image to be embedded is impaired. Without having to embed a watermark image.
The digital watermark embedding method using this frequency space has a feature that it has higher tolerance (ie, robustness) than other embedding methods.

As a technique using the watermark information in the frequency space as described above, there is “the digital watermark insertion method, the digital watermark detection method, the digital watermark insertion device, and the digital watermark detection device” of the following Patent Document 1.
JP 2006-287700 A

  This technique sets watermark information by arranging binarized signals in a frequency space in a predetermined pattern, performs inverse discrete Fourier transform on the watermark information, and targets the watermark image generated as a result. Is to be embedded. When the watermark image embedded in the image is detected and inversely converted, the original signal pattern can be restored.

Further, as an application example using this technique, there is “printing apparatus, printing method, and printing program” of the following Patent Document 2.
JP 2006-130801 A

As shown in FIG. 9, in this technique, a signal pattern 16 in which signals are distributed in a frequency space 1 stretched between an imaginary axis and a real axis is defined.
Here, black circles in the figure indicate that signals are arranged, and white circles indicate that signals are not arranged. That is, the signal pattern 16 defines binary information indicating whether the signal is “present” or “not present” at a predetermined position in the frequency space 1.

Next, the signal pattern 16 is subjected to inverse discrete Fourier transform to generate a watermark pattern 2 in the image space.
If necessary, this is binarized to generate a binarized watermark pattern 3, which is subjected to gradation adjustment or the like and superimposed on the image for printing.
In the example shown in the figure, the binary watermark pattern 3 is added to or subtracted from the character data 5 "Tomorrow is a special sale day" and printed.

When this print is taken with a camera and discrete Fourier transform is performed, a spectrum image 1a is obtained in the frequency space 1, and the signal pattern 6a is restored.
For simplification, in the spectrum image 1a, the spectrum of the original image is omitted, and only the signal pattern 6a is shown.

If this signal pattern is associated with a specific URL (Uniform Resource Locators) or the like, the user can be guided to the website of the URL.
Using such a mechanism, for example, a digital watermark is embedded in a poster, and when the user takes a picture with the camera of the mobile phone, the mobile phone analyzes the digital watermark of the poster, and the content related to the poster A service such as connecting to a website that provides the service can be provided.

For example, when developing a business in which an electronic watermark associated with an identification number (ID) or URL of a specific content is embedded in an image, a large number of signal patterns are required.
For that purpose, in the algorithm based on the binary method of “when there is” or “when there is no signal”, it is necessary to increase the arrangement density of the signals arranged in the frequency space and increase the signal pattern. When the density is increased, there is a problem that it becomes difficult to detect a signal pattern because it is easily affected by adjacent signals.

  Accordingly, an object of the present invention is to increase the amount of information that can be provided by digital watermarking.

According to the first aspect of the present invention, position specifying means for specifying a position in the frequency space , signal selecting means for selecting a plurality of types of signals having different feature values, and a signal for arranging the selected signal at the specified position Digital watermark generation, comprising: arrangement means; and conversion means for converting a pattern formed by the arranged signals into an image space , wherein the feature amount is an area occupied by the signal in a frequency space. Providing the device.
According to the second aspect of the present invention, position specifying means for specifying a position in the frequency space , signal selecting means for selecting a plurality of types of signals having different feature quantities, and a signal for arranging the selected signal at the specified position A digital watermark comprising: arrangement means; and conversion means for converting a pattern formed by the arranged signals into an image space , wherein the feature amount is a shape formed by the signal in a frequency space. A generating device is provided .
According to a third aspect of the present invention, there is provided the digital watermark generating apparatus according to the first or second aspect , wherein the signal is composed of a set of binary signals.
According to a fourth aspect of the invention, there is provided pattern embedding means for embedding the pattern converted into the image space in an image, and image output means for outputting an image in which the pattern is embedded. to provide an electronic watermark generating apparatus according to any one of claims of from claim 1 to claim 3.
According to the fifth aspect of the present invention, in the computer including the position specifying means, the signal selecting means, the signal arranging means, and the converting means, the position specifying step of specifying the position in the frequency space by the position specifying means. A signal selection step of selecting a plurality of types of signals having different feature quantities by the signal selection unit, a signal arrangement step of arranging the selected signal at the designated position by the signal arrangement unit, and the conversion unit in a conversion step of converting the pattern formed by the arrangement signals the image space, consists, the feature quantity, the digital watermark generation method, wherein the signal is the area occupied in frequency space Provide .
According to a sixth aspect of the present invention, in the computer comprising the position specifying means, the signal selecting means, the signal arranging means, and the converting means, the position specifying step for specifying the position in the frequency space by the position specifying means. A signal selection step of selecting a plurality of types of signals having different feature quantities by the signal selection unit, a signal arrangement step of arranging the selected signal at the designated position by the signal arrangement unit, and the conversion unit And a conversion step of converting the pattern formed by the arranged signals into an image space , wherein the feature amount is a shape formed by the signal in a frequency space. Provide .
In the invention according to claim 7, a position designation function for designating a position in a frequency space , a signal selection function for selecting a plurality of types of signals having different feature values, and a signal for arranging the selected signal at the designated position An electronic watermark generation program for realizing a placement function and a conversion function for converting a pattern formed by the placed signals into an image space by a computer , wherein the feature amount is an area occupied by the signal in a frequency space. There is provided a digital watermark generation program characterized by a certain feature .
In the invention according to claim 8, a position designation function for designating a position in a frequency space , a signal selection function for selecting a plurality of types of signals having different feature values, and a signal for arranging the selected signal at the designated position A digital watermark generation program that realizes an arrangement function and a conversion function for converting a pattern formed by the arranged signals into an image space by a computer , wherein the feature amount is a shape formed by the signal in a frequency space A digital watermark generation program characterized by the above is provided .
According to the ninth aspect of the present invention, image data acquisition means for acquiring image data in which a pattern defined in the frequency space is embedded, and the pattern embedded in the acquired image data is converted from the image space to the frequency space. Pattern restoration means for restoring the pattern in the frequency space, and detection means for detecting the position of the signal arranged in the restored pattern and the feature quantity of the signal , wherein the feature quantity is Provided is a digital watermark detection apparatus characterized in that a signal occupies an area in a frequency space .
According to a tenth aspect of the present invention, image data acquisition means for acquiring image data in which a pattern defined in a frequency space is embedded, and a pattern embedded in the acquired image data is converted from the image space to the frequency space. Pattern restoration means for restoring the pattern in the frequency space, and detection means for detecting the position of the signal arranged in the restored pattern and the feature quantity of the signal , wherein the feature quantity is There is provided a digital watermark detection apparatus characterized in that a signal has a shape formed in a frequency space .
According to the invention of claim 11, an image data acquisition function for acquiring image data in which a pattern defined in a frequency space is embedded, and a pattern embedded in the acquired image data is converted from an image space to a frequency space. A digital watermark detection program that implements a pattern restoration function for restoring a pattern in a frequency space and a detection function for detecting a signal position and a signal feature amount arranged in the restored pattern by a computer. The feature amount is an area occupied by the signal in a frequency space, and provides a digital watermark detection program.
In the invention described in claim 12, the image data acquisition function for acquiring the image data in which the pattern defined in the frequency space is embedded, and the pattern embedded in the acquired image data is converted from the image space to the frequency space. A digital watermark detection program that implements a pattern restoration function for restoring a pattern in a frequency space and a detection function for detecting a signal position and a signal feature amount arranged in the restored pattern by a computer. The feature value is a shape formed by the signal in a frequency space, and provides a digital watermark detection program.

  According to the present invention, it is possible to increase the amount of information that can be provided by digital watermarking by multi-leveling a signal to be arranged in a frequency space, and use a binary signal when embedding a certain amount of information. The target image size can be set smaller than in the case where there is a problem.

(1) Outline of Embodiments A signal pattern is generated by distributing multi-valued signals in a frequency space, and an inverse discrete Fourier transform is performed on the signal pattern to generate a watermark pattern in an image space. Here, multi-leveling means ternization or higher.
Signal multi-leveling is performed by collecting binary signals, which are signals that can be detected robustly, into a lump, and generating signals having different characteristics such as shape and area. Thereby, a robust multilevel signal can be realized.

  These multi-valued signals are converted into a watermark pattern, and then converted into a frequency space by performing discrete Fourier transform, the original signal pattern is restored, but the restored individual signals are, for example, the original signal Corresponding to the feature, the maximum amplitude value is different or the integrated value of the amplitude is different, so that they can be distinguished from each other.

As described above, in the present embodiment, the feature value of the original signal can be recognized using the maximum value of the amplitude at the time of restoration, and signals having different values can be distinguished.
The amount of the characteristic value of the signal measured with respect to a multi-valued signal differs depending on the individual signal, and can be determined by experiments or the like.
In other words, when the original signal is restored, the signal is formed in a shape and area such that an identifiable feature amount is generated.

When the signal itself is multi-valued in this way, the amount of information that can be embedded in the frequency space increases dramatically.
As an example, when the first type signal and the second type signal are arranged in the frequency space as two signals having different feature quantities, in the signal pattern, “when there is a first type signal”, “second” Three values of “when there is a seed signal” and “when there is no signal” are possible, and the number of definable signal patterns increases dramatically.

(2) Details of Embodiment FIG. 1A is a diagram for explaining an example of a signal pattern of a signal according to the present embodiment.
The frequency space 1 is a space stretched between an orthogonal imaginary axis (vertical axis) and a real axis (horizontal axis), and a signal pattern is formed by arranging signals concentrically (that is, by giving a frequency component). Has been. When this signal pattern is subjected to inverse discrete Fourier transform, a pattern-like watermark pattern in the image space is obtained.

As an example, the signal pattern is generated by placing signals at predetermined signal positions on concentric circles 10 to 40 centered on the origin.
The signal arranged in the outermost circle 10 is for detecting “an image enlargement / reduction ratio depending on the distance between the subject and the camera” that occurs when an image in which a watermark pattern is embedded is taken by the camera.
On the circle 10, the signal positions 10a,..., The signal positions 10e,.
Later, when the signal pattern is restored from the watermark pattern, the enlargement / reduction ratio by photographing is corrected using the signal on the circle 10.

The signal arranged in the circle 20 positioned second from the outermost side is for detecting the “rotation angle of the camera in the direction parallel to the subject” that occurs when the camera shoots an image in which the watermark pattern is embedded. Is.
On the circle 20, the signal positions 20 a,..., 20 c,.
Later, when the signal pattern is restored from the watermark pattern, the rotation angle by photographing is corrected using the signal on the circle 20.

The signals arranged in the circles 30 and 40 are for defining watermark information, and a signal pattern is set by the signals arranged in these.
On the circle 30, the signal positions 30a,..., The signal positions 30c,.
In this way, 12 signal positions are designated on the circle 30, but in this frequency space 1, there is a characteristic that the signals are distributed point-symmetrically. Therefore, the signal pattern is defined by 6 signal positions. Is done.

On the circle 40, the signal positions 40a, 40b,... Are specified to be evenly distributed as positions for arranging signals.
As described above, eight signal positions are specified on the circle 30, but in this frequency space 1, there is a characteristic that the signals are distributed point-symmetrically. Therefore, the signal pattern is defined by the four signal positions. Is done.
For example, content IDs and website URLs can be associated with signal patterns distributed in the circles 30 and 40.

FIG. 1B is a diagram showing signals arranged at signal positions on the circles 30 and 40.
For example, a first type signal and a second type signal having different feature amounts such as area and shape are prepared as signals arranged at the signal positions.
In FIGS. 1A and 1B, the first type signal is represented by a black circle, the second type signal is represented by a black square, and the case where no signal is arranged is represented by a white circle.

  As described above, by using the first type signal and the second type signal as the signals arranged at the signal positions, conventionally, only two values of “when there is a signal” and “when there is no signal” can be taken at the signal position. However, in the present embodiment, the three values “when there is a first type signal”, “when there is a second type signal”, and “when there is no signal” can be taken.

In this embodiment, at the stage of restoring and detecting the signal pattern, the signal components are arranged in the descending order of the signal intensity (feature amount), and the signal components are determined based on relative values.
When this method is used, in consideration of the fact that the signal is distributed point-symmetrically, in the case of the circle 30, the second type signal is distributed at two signal positions of the six signal positions, The degree of freedom is maximized when the first type signal is distributed to two other signal positions.
Therefore, the signal pattern of the circle 30 can represent ₆ C ₂ × ₄ C ₂ × ₂ C ₂ = 90 types of signal patterns.

On the other hand, in the case of binary, since signals are distributed to three signal positions out of six signal positions, ₆ C ₃ = 20 kinds of signal patterns are possible.
Thus, it can be seen that even when signals are distributed at the same signal position, many signal patterns can be realized by multi-leveling the signals.
In the above example, the signals are distributed in the circles 30 and 40. For example, when a circle having 24 signal positions is provided outside the circle 30, a signal pattern is formed by 12 signal positions. Since it can be defined, it is possible to embed an information amount approximately 1000 times that of the binary case.

Next, the multi-value signal will be described.
In general, in digital signal processing, the amount of information can be increased by multi-leveling the signal. However, in the field of digital watermarking, no technology has been developed for increasing the amount of information by multi-level signal processing. It was. This is due to the following reason.

In other words, attempts to multilevel signals are generally made by setting the signal strength in multiple stages, but in the case of digital watermarking, noise components caused by the content of the image in which the watermark pattern is embedded This is because the signal intensity cannot be properly detected with respect to the noise component in the input processing, and sufficiently robust embedding cannot be expected.
For this reason, it is considered difficult to increase the strength of the embedded signal by making the binary signal itself multi-valued.

Therefore, in this embodiment, a multilevel signal is expressed using a set of binary signals that can be detected robustly.
As a result, the signal intensity can be made multi-level even though binary signals that can be detected robustly are handled.
Specifically, a new signal is formed by a block made up of a set of binary signals, and different signals are mixed and distributed in the frequency space, thereby achieving multilevel signal processing.

As shown in FIG. 2A, the conventional signal distribution is a signal in which one signal unit of a binary signal is distributed at signal positions, or a plurality of signal units are aggregated as shown in FIG. 2B. Units have been distributed in signal positions.
On the other hand, in the present embodiment, for example, as shown in FIG. 2C, a signal by one signal unit and a signal unit that is a lump formed by aggregating a plurality of one signal unit into a predetermined shape are mixed. Multi-leveling is planned.

In FIG. 2C, if a signal consisting of one signal unit of a binary signal is a first type signal, a signal consisting of a set of one signal unit is a second type signal.
For example, as shown in FIG. 2D, various shapes such as a cross shape, a rectangular shape, or a donut shape are conceivable as the shape of the signal composed of the set.
Here, what is important is that these multi-valued signals can be distinguished when the watermark pattern is restored and detected after the watermark pattern is embedded in the image.

Next, signal shape examples will be described with reference to FIG.
In this embodiment, one pixel is defined as one signal unit of a binary signal, and as an example, the signal shape is defined by pixels arranged in 5 × 5 (7 × 7 in the case of FIG. 3E).
Among these pixels, black pixels correspond to one signal unit forming a signal. As described above, in this embodiment, the frequency space is divided into one signal unit by pixels, and the shape of the signal is set by a set of the one signal unit.

FIG. 3A shows a signal shape defined by one central signal unit.
FIG. 3B shows a signal shape defined by collecting all signal units of 5 × 5.
FIG. 3 (c) defines the signal shape by gathering one signal unit in a cross shape.
FIG. 3D shows a signal shape defined by aggregating one signal unit so as to be as close to a circular shape as possible.

Of the signal shapes exemplified above, any signal shape that can be distinguished when the signal pattern is restored can be selected and combined to form a multi-value signal.
For example, when the signal of FIG. 3A and the signal of FIG. 3B can be distinguished, for example, the signal of FIG. 3A is the first type signal and the signal of FIG. It can be a seed signal.
Furthermore, when the signal of FIG. 3C can be distinguished, this is adopted as the third type signal, and the signal pattern can be set by a quaternary signal.

  FIG. 3E shows an example in which the signal shape is set by 7 × 7 pixels. In this example, a donut-shaped signal is formed by collecting one signal unit of the outer periphery and the inner periphery.

The inventor of the present application has experimented with various signal shapes. When the signal shape is a square, if the number of signal units on one side is an even number (for example, 6 × 6), the maximum signal strength in the local region is obtained. Stable signal detection can be performed by using the value as a feature amount. Further, when the number of signal units on one side is an odd number (for example, 5 × 5), the total value of the intensity in the local region is also stable. The result that the signal can be detected is obtained.
Further, when the maximum value is used, one signal consisting of one pixel results from a large change (reduction) in the strength of the watermark signal in the process of superimposing an image using one signal unit signal consisting of a plurality of pixels. The unit signal has a higher intensity.
Further, when using the total value, it has been experimentally verified that the signal intensity appears in a larger area in the single signal unit signal composed of a plurality of pixels, so that the total intensity is larger than that in the single signal unit signal.
When this total value is used, a detection result that can detect a stable signal has been experimentally obtained when a signal configuration excluding the central pixel in the region of one signal unit signal composed of a plurality of pixels is obtained. .

FIG. 4 is a diagram for explaining a hardware configuration of the information processing apparatus 50 according to the present embodiment.
The information processing apparatus 50 includes a CPU (Central Processing Unit) 51, a RAM (Random Access Memory) 52, a ROM (Read Only Memory) 53, an input device 54, a display device 55, a printing device 56, a communication control device 57, and a storage device 58. Each functional unit such as a storage medium driving device 59 and an input / output I / F (interface) 60 is connected by a bus line.

The CPU 51 is a central processing unit that performs various types of arithmetic processing, information processing, and control of each component constituting the information processing apparatus 50 according to a predetermined program.
More specifically, for example, the CPU 51 performs signal arrangement in the frequency space, generates a watermark pattern by performing inverse discrete Fourier transform on the signal pattern of the signal, embeds the watermark pattern in the image data, or The watermark pattern is detected from the image, discrete Fourier transform is performed, and the signal pattern is restored.

The ROM 53 is a read-only memory that stores a basic program and data for operating the information processing apparatus 50.
The RAM 52 is a readable / writable memory that provides a working memory for the CPU 51 to operate.

The input device 54 has an operation device such as a keyboard and a mouse, for example, and is used when the user operates the information processing apparatus 50.
The display device 55 includes a display device such as a liquid crystal display, for example, and displays various screens when the user operates the information processing device 50, for example.

The communication control device 57 is a device that connects the information processing device 50 to a communication network such as the Internet. The information processing device 50 can communicate with various server devices and terminal devices via the communication control device 57.
The printing device 56 includes, for example, a printer such as a laser printer, an ink jet printer, or a thermal transfer printer, and can print an image in which a digital watermark is embedded or other images.
An input / output I / F (interface) 60 is an interface for connecting to various external devices. For example, a scanner or a digital camera is connected to the information processing device 50, and an image is processed using these. 50 can be configured.

The storage medium drive device 59 is a functional unit that drives the attached removable storage medium and reads and writes data.
Examples of the readable / writable storage medium include an optical disk, a magneto-optical disk, and a magnetic disk. Image data can be read from the storage medium, and image data and other data can be written in the case of a writable storage medium. it can.

The storage device 58 is a large-capacity readable / writable storage device configured by, for example, a hard disk.
The storage device 58 includes a program storage unit 61 that stores programs and a data storage unit 62 that stores data.

In the program storage unit 61, an OS (Operating System) which is basic software for operating the information processing apparatus 50, a signal pattern is generated, a signal pattern is converted, and the converted watermark pattern is converted into image data. A watermark embedding program for causing the CPU 51 to perform a function of embedding in a digital watermark, a digital watermark detection program for causing the CPU 51 to exhibit a function of detecting a watermark pattern from image data, inversely transforming it, and restoring a signal pattern Is remembered.
The data storage unit 62 stores, for example, a signal pattern, a watermark pattern obtained by converting the signal pattern, and other data.

FIG. 5 is a diagram showing an example of the signal pattern generation screen 71.
The signal pattern generation screen 71 is displayed on the display of the display device 55 (FIG. 4) and assists the user in generating a signal pattern.
On the left side of the screen, a frequency space for arranging signals is displayed. In circles 10 and 20, signals for detecting an enlargement / reduction ratio and a rotation angle are arranged by default.

In the circle 30 and the circle 40, by default, all signal positions are “no signal”, that is, a white circle, and the user arranges signals.
Since the signal pattern in the frequency space has the property of being point-symmetric with respect to the origin, the first quadrant and the fourth quadrant are input.

In the signal column 73, the first type signal and the second type signal are displayed. When the user operates the pointer 72 with the mouse, the signal column 73 can be dragged and dropped from the signal column 73 to desired signal positions of the circles 30 and 40. It is like that.
Thus, the user can arrange desired signals at the signal positions of the circles 30 and 40 by operating the mouse.

The conversion button 74 is a button for generating a watermark pattern by performing inverse discrete Fourier transform on the signal pattern set in the frequency space.
When the user clicks the conversion button 74 after arranging the signal, the information processing apparatus 50 performs a conversion process.

Next, a procedure for using the digital watermark will be described. The usage procedure is roughly as follows.
(1) A signal pattern is set by distributing multilevel signals in a frequency space.
(2) A watermark pattern is generated by performing inverse discrete Fourier transform on the set signal pattern.
(3) By synthesizing the generated watermark pattern with an image to be embedded (original image), the watermark pattern is embedded in the image.
(4) Print an image in which a watermark pattern is embedded.
The above is a process for generating an image with an embedded digital watermark, and is performed using a digital watermark embedding program.

(5) Take a printed image with a digital camera or the like.
(6) A watermark pattern is extracted from the captured image.
(7) The signal pattern is restored by subjecting the extracted watermark pattern to discrete Fourier transform.
The above is the process of restoring the embedded digital watermark, which is performed using a digital watermark detection program.

Next, the process described above will be described in more detail using a flowchart.
FIG. 6 is a flowchart for explaining the procedure of the digital watermark embedding process performed by the information processing apparatus 50. The following processing is performed by the CPU 51 according to the digital watermark embedding program.
Of the following steps, Steps 5 to 25 are processing in the frequency space, and Steps 30 to 40 are processing in the image space.

  First, for example, the information processing apparatus 50 receives the arrangement of the signal pattern for detecting the enlargement / reduction ratio in the frequency space on the signal pattern generation screen 71 (FIG. 5) (step 5), and further, the signal pattern for detecting the rotation angle. Is accepted (step 10). These signal patterns are, for example, signals arranged in circles 10 and 20 in FIG. 1A, and are set by default, for example.

Next, the information processing apparatus 50 receives, for example, the arrangement of the signal pattern for watermark information in the frequency space on the signal pattern generation screen 71 (step 15).
This signal pattern is, for example, a signal arranged in the circle 30 and the circle 40 in FIG. 1A. For example, the user operates the information processing apparatus 50 and multi-values at predetermined signal positions on these circles. A desired pattern is generated by arranging these signals.

As described above, the information processing apparatus 50 includes position specifying means for specifying a position (a predetermined position on a circle) at which a signal is arranged in the frequency space.
In addition, the information processing apparatus 50 includes a signal selection unit that selects a plurality of types of signals having different feature values by multi-leveling, for example, by accepting an operation from a user, and a position specified in a frequency space. Are provided with signal arrangement means for arranging the selected signals.

After receiving the signal pattern arrangement as described above, for example, when the user clicks the conversion button 74 on the signal pattern generation screen 71, the information processing apparatus 50 generates a signal pattern with the received signal arrangement. (Step 20), this is subjected to inverse discrete Fourier transform (Step 25).
As described above, the information processing apparatus 50 includes conversion means for converting a pattern formed by signals arranged in the frequency space into the image space.

At this time, the size of the watermark pattern output by the inverse transformation is the same as the number of samples of the input signal pattern.
By this inverse transformation, image data having a pattern like the watermark pattern 2 shown in the conventional example of FIG. 9 is generated.
Although a negative value is also calculated for the watermark pattern obtained by the inverse transformation, the information processing apparatus 50 performs an offset process so as to eliminate the negative portion as an example.

Next, the information processing apparatus 50 binarizes the information using, for example, a predetermined threshold (step 30). For example, zero can be used as a threshold value for a non-offset watermark pattern, and the offset amount can be used as a threshold value for an offset watermark pattern.
Then, watermark pattern image data such as the binary watermark pattern 3 shown in the conventional example of FIG. 9 is generated. The binarized watermark pattern is also offset so that the negative part disappears.
The information processing apparatus 50 stores the image data of the watermark pattern or the binarized watermark pattern generated as described above as a file, for example, in the data storage unit 62 (FIG. 4).

Next, the user operates the information processing apparatus 50 to select original image data to be embedded with the digital watermark from the image data stored in the data storage unit 62.
Then, the information processing apparatus 50 adds or subtracts the binarized watermark pattern to the original image data (step 35), and embeds the binarized watermark pattern in the image data.

The embedding is performed by matching the sizes of the original image and the watermark pattern by repeating the watermark pattern until it becomes the same size as the original image.
As described above, since the watermark pattern is repeatedly embedded in the original image, even if a part of the original image is lost, the signal pattern can be restored from other parts, and a robust watermark can be realized. it can.
As described above, the information processing apparatus 50 includes pattern embedding means for embedding a pattern (watermark pattern) converted into an image space into an image (original image).

The information processing apparatus 50 outputs the image data in which the binarized watermark pattern 3 is embedded in this way to a file or prints it with a printer (step 40).
In the present embodiment, a binarized watermark pattern may be embedded in the target image, but a watermark pattern that is not binarized may be embedded.
As described above, the information processing apparatus 50 includes image output means for outputting an image in which a pattern (watermark pattern) is embedded.

As described above, the information processing apparatus 50 generates a signal pattern from a multi-valued signal in a frequency space, generates a watermark pattern in an image space by converting the signal pattern, and generates a watermark pattern for image data. Can be embedded.
Note that the information processing apparatus 50 can transmit and provide the generated signal pattern, watermark pattern, and image data in which the watermark pattern is embedded to another server apparatus via the communication control apparatus 57.

FIG. 7 is a flowchart for explaining the procedure of the digital watermark detection process performed by the information processing apparatus 50. The following processing is performed by the CPU 51 according to the digital watermark detection program.
Of the following steps, step 50 and step 55 are processes in the image space, and steps 60 to 80 are processes in the frequency space.

First, a printed image with an embedded watermark pattern is prepared and photographed with a digital camera. Then, for example, image data captured using the input / output I / F 60 is transferred from the camera to the information processing apparatus 50. The transferred image data is stored in the data storage unit 62, for example.
Thus, the information processing apparatus 50 includes image data acquisition means for acquiring image data in which a pattern defined in the frequency space is embedded.

When performing a discrete Fourier transform on a watermark pattern, the size of the image used for the transformation needs to be the same sampling number as the watermark pattern created by the inverse transformation.
For this reason, when obtaining the image data in which the watermark pattern is embedded, the information processing apparatus 50 divides the input image into blocks having the same aspect ratio as the watermark pattern (step 50).

Next, the information processing apparatus 50 performs a discrete Fourier transform on the divided image data and converts it into a spectrum component in the frequency space (step 55).
In this spectral component, the information component (signal) of the watermark pattern is superimposed on the spectral component of the original image.

In general, since the spectral component of the original image has many direct current components, the spectral component of the original image appears strongly near the origin in the frequency space. On the other hand, since the watermark pattern has components set (arranged) in regions away from the imaginary axis and the real axis, the watermark pattern is restored by being distinguished from the spectrum of the original image.
As described above, the information processing apparatus 50 converts a pattern (signal pattern) in the frequency space by converting the pattern (watermark pattern) embedded in the image data acquired from a digital camera or the like from the image space to the frequency space. Pattern restoring means for restoring is provided.

  Next, the information processing apparatus 50 calculates the amplitude of the spectral component in the frequency space (step 60) and further integrates it (step 65). As a result, an image like the spectrum image 1a as shown in FIG. 9 is obtained.

Next, the information processing apparatus 50 detects an enlargement / reduction ratio for the obtained spectrum image (step 70).
This detection is performed at an arbitrary radius by detecting the same number of components of the enlargement / reduction ratio detection pattern (the number of signals on the circle 10), particularly those having a large integrated value of amplitude, and the circle on which the signal is arranged. This is done by detecting the radius of. The magnification / reduction ratio of the spectral image is determined by the detected size of the radius.

Next, the information processing apparatus 50 detects the rotation angle (step 75).
In this detection, the information processing apparatus 50 first calculates the radius of the circle 20 based on the previously detected enlargement / reduction ratio.
Then, on the circumference of the radius, the component distribution of the rotation angle detection pattern is formed, the spectrum having the largest total amplitude is set as a signal arranged on the circle 20, and the spectrum is detected by detecting the rotation angle of this signal. The rotation angle of the image is detected.

After detecting the enlargement / reduction ratio and rotation angle in this way, the information processing apparatus 50 corrects the enlargement / reduction ratio and rotation angle of the spectral image, and then detects the signal pattern (step 80).
In this detection, the information processing apparatus 50 first determines a signal position on the circumference of the circle 30 and the circle 40, and if there is a signal at the signal position, and if there is a signal, Whether it is a value signal or not is recognized by the feature value.

  The multi-value component has different characteristics depending on the shape of the signal. For example, when a first type signal composed of one pixel and a second type signal composed of a set of 13 pixels are distributed in a signal pattern. The maximum value of the amplitude in the local area is larger for one pixel and smaller for 13 pixels.

  This property is generally maintained even when a watermark pattern is superimposed on an original image. Therefore, in this case, the amplitude is used as a feature quantity, the signals are ranked in descending order of the maximum amplitude at the signal position for each circle, a signal having a relatively large maximum amplitude is a first type signal, and a signal having a relatively small amplitude. It is determined that there is no signal for the second type signal and other signal positions.

Note that the above is an example of the feature amount, and depending on the shape of the signal, the sum of the amplitudes in the local region can be used as the signal intensity.
As described above, the information processing apparatus 50 includes a detection unit that detects a position (component) of a signal and a feature amount of a signal arranged in a restored pattern (signal pattern), and a distribution of signals in the frequency space. Recognizing means for recognizing a pattern made up of signal features is provided.

  In the above, the signal pattern of the watermark pattern embedded in the original image is detected by the information processing apparatus 50. This detection function can also be realized by a mobile terminal such as a mobile phone or a user's personal computer, for example. It can also be used by a user to detect a signal pattern from an image.

Next, an example in which the digital watermark described above is applied to business will be described.
FIG. 8 is a diagram illustrating a configuration of the information providing system 100.
The information providing system 100 includes an information processing apparatus 50, a URL server 101, a printer 102, a receipt 103, a mobile phone 104, a service server 105, and the like.
In FIG. 8, only one printer 102 and one mobile phone 104 are shown. However, there can be a plurality of printers 102 provided in the store, and the mobile phone 104 exists for each mobile phone user. .
In the following, description will be given according to the numbers in parentheses in the figure.

(1) The information processing apparatus 50 generates a signal pattern and watermark image data in which a watermark pattern based on the signal pattern is embedded, and transmits it to the URL server 101 to provide it.
(2) When the URL server 101 receives these signal patterns and watermark image data, the URL server 101 assigns IDs thereto, and further associates URLs of link destinations with each other and stores them in the URL database.

(3) The printer 102 is installed in, for example, a cash register such as a supermarket, and prints a receipt.
At that time, the printer 102 requests watermark image data from the URL server 101, and prints an image with a digital watermark on the receipt 103 using the watermark image data transmitted from the URL server 101.

(4) Upon receipt of the receipt 103, the user takes a digital watermarked image printed on the receipt 103 with a digital camera built in the mobile phone 104.
The mobile phone 104 is preinstalled with an application for analyzing the digital watermark and connecting the mobile phone 104 to a website.
With this application, the mobile phone 104 restores a signal pattern by performing a discrete Fourier transform on the captured image.

(5) In addition, this application stores the correspondence between signal patterns and IDs similar to those stored in the URL database of the URL server 101, and the mobile phone 104 sets the ID corresponding to the restored signal pattern. It is transmitted to the URL server 101.
(6) When the URL server 101 receives the ID from the mobile phone 104, the URL server 101 searches the URL database for the URL corresponding to the ID and transmits it to the mobile phone 104.

(7) Upon receiving the URL from the URL server 101, the mobile phone 104 uses this to connect to the website of the service server 105.
In this way, the user can access the website of the service server 105 and receive various services by photographing the digital watermarked image printed on the receipt 103 with the mobile phone 104.

According to the embodiment described above, the following effects can be obtained.
(1) The amount of information embedded in the frequency space can be increased by multi-leveling the signals arranged in the frequency space.
(2) Since binary signals that can be detected robustly are assembled to generate a multilevel signal, the multilevel signal can be detected robustly even if a watermark pattern is embedded in the image.
(3) Depending on the shape and size of the signal in the frequency space, multiple values of the signal can be read with the characteristic amount specific to the signal.
(4) By increasing the number of watermark patterns, more various services can be provided.
(5) When a multilevel signal is used, the amount of information that can be embedded per unit area of an image is larger than that of a binary signal. Therefore, the required image area can be reduced even with the same amount of information. Also, the amount of information that can be embedded increases even with the same image area.

  In the embodiment described above, the discrete Fourier transform and the inverse discrete Fourier transform are used as an example of the orthogonal transform, but the signal embedded in the frequency space is multi-valued by using another orthogonal transform such as a wavelet transform. You can also

It is a figure for demonstrating an example of the signal pattern of the signal which concerns on this Embodiment. It is a figure for demonstrating distribution of a signal. It is a figure for demonstrating the example of the shape of a signal. FIG. 3 is a diagram for describing a hardware configuration of an information processing apparatus. It is the figure which showed an example of the signal pattern production | generation screen. It is a flowchart for demonstrating the procedure of a digital watermark embedding process. It is a flowchart for demonstrating the procedure of a digital watermark detection process. It is the figure which showed the structure of the information provision system. It is a figure for demonstrating a prior art example.

Explanation of symbols

1 frequency space 10 yen 20 yen 30 yen 40 yen 50 Information processing device 51 CPU
52 RAM
53 ROM
54 Input Device 55 Display Device 56 Printing Device 57 Communication Control Device 58 Storage Device 59 Storage Medium Drive Device 60 Input / Output I / F
61 Program storage unit 62 Data storage unit

Claims (12)

A position specifying means for specifying a position in the frequency space;
Signal selection means for selecting a plurality of types of signals having different feature quantities;
Signal placement means for placing the selected signal at the designated position;
Conversion means for converting a pattern formed by the arranged signals into an image space;
Equipped with,
The digital watermark generation apparatus , wherein the feature amount is an area occupied by the signal in a frequency space . A position specifying means for specifying a position in the frequency space;
Signal selection means for selecting a plurality of types of signals having different feature quantities;
Signal placement means for placing the selected signal at the designated position;
Conversion means for converting a pattern formed by the arranged signals into an image space;
Equipped with,
The digital watermark generation apparatus according to claim 1, wherein the feature amount is a shape formed by the signal in a frequency space . 3. The digital watermark generation apparatus according to claim 1, wherein the signal is composed of a set of binary signals. Pattern embedding means for embedding the pattern converted into the image space in an image;
Image output means for outputting an image in which the pattern is embedded;
Electronic watermark generating apparatus according to any one of claims of claims 1, characterized by comprising up to claim 3. In a computer comprising a position specifying means, a signal selecting means, a signal arranging means, and a converting means,
A position specifying step for specifying a position in a frequency space by the position specifying means;
A signal selection step of selecting a plurality of types of signals having different feature quantities by the signal selection means;
A signal placement step of placing the selected signal at the designated position by the signal placement means;
A conversion step of converting a pattern formed by the arranged signals into an image space by the conversion means;
Consisting of
The digital watermark generation method , wherein the feature amount is an area occupied by the signal in a frequency space . In a computer comprising a position specifying means, a signal selecting means, a signal arranging means, and a converting means,
A position specifying step for specifying a position in a frequency space by the position specifying means;
A signal selection step of selecting a plurality of types of signals having different feature quantities by the signal selection means;
A signal placement step of placing the selected signal at the designated position by the signal placement means;
A conversion step of converting a pattern formed by the arranged signals into an image space by the conversion means;
Consisting of
The digital watermark generation method , wherein the feature amount is a shape formed by the signal in a frequency space . A position specifying function for specifying a position in the frequency space;
A signal selection function for selecting multiple types of signals with different feature quantities;
A signal placement function for placing the selected signal at the designated position;
A conversion function for converting a pattern formed by the arranged signals into an image space;
The A watermark generation program for implementing by a computer,
The feature amount is an area occupied by the signal in a frequency space.
A digital watermark generation program characterized by the above . A position specifying function for specifying a position in the frequency space;
A signal selection function for selecting multiple types of signals with different feature quantities;
A signal placement function for placing the selected signal at the designated position;
A conversion function for converting a pattern formed by the arranged signals into an image space;
The A watermark generation program for implementing by a computer,
The feature amount is a shape formed by the signal in a frequency space.
A digital watermark generation program characterized by the above . Image data acquisition means for acquiring image data in which a pattern defined in a frequency space is embedded;
Pattern restoration means for restoring the pattern in the frequency space by converting the pattern embedded in the acquired image data from the image space to the frequency space;
Detecting means for detecting the position of the signal and the feature amount of the signal arranged in the restored pattern;
Equipped with,
The digital watermark detection apparatus , wherein the feature amount is an area occupied by the signal in a frequency space . Image data acquisition means for acquiring image data in which a pattern defined in a frequency space is embedded;
Pattern restoration means for restoring the pattern in the frequency space by converting the pattern embedded in the acquired image data from the image space to the frequency space;
Detecting means for detecting the position of the signal and the feature amount of the signal arranged in the restored pattern;
Equipped with,
The digital watermark detection apparatus , wherein the feature amount is a shape formed by the signal in a frequency space . An image data acquisition function for acquiring image data in which a pattern defined in a frequency space is embedded;
A pattern restoration function for restoring the pattern in the frequency space by converting the pattern embedded in the acquired image data from the image space to the frequency space;
A detection function for detecting the position of the signal and the feature amount of the signal arranged in the restored pattern;
The A digital watermark detection program for implementing by a computer,
The feature amount is an area occupied by the signal in a frequency space.
A digital watermark detection program. An image data acquisition function for acquiring image data in which a pattern defined in a frequency space is embedded;
A pattern restoration function for restoring the pattern in the frequency space by converting the pattern embedded in the acquired image data from the image space to the frequency space;
A detection function for detecting the position of the signal and the feature amount of the signal arranged in the restored pattern;
The A digital watermark detection program for implementing by a computer,
The feature amount is a shape formed by the signal in a frequency space.
A digital watermark detection program.

Images