JP2006050069A - Electronic watermark inserting method, electronic watermark inserting apparatus, and electronic watermark inserting program, and electronic watermark detecting method, electronic watermark detecting apparatus, and electronic watermark detection program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To increase data information amount of an electronic watermark to be embedded in an image without making the electronic watermark conspicuous. <P>SOLUTION: An electronic watermark for identifying a right image and an electronic watermark denoting first data D1 are inserted to the right image, and an electronic watermark for identifying a left image and an electronic watermark denoting second data D2 are inserted to the left image. A photographed image is divided into a photographed right image and a photographed left image, the first data are detected from a photographed half image (photographed right image or photographed left image) to which the electronic watermark for identifying the right image is inserted, and the second data are detected from a photographed half image (photographed left image or photographed right image) to which the electronic watermark for identifying the left image is inserted. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a digital watermark insertion method for inserting a digital watermark into image data, its apparatus and its program, a digital watermark detection method for detecting a digital watermark from image data into which the digital watermark has been inserted, its apparatus and its Regarding the program.

  In recent years, a technique for inserting a digital watermark into image data has been developed, and information for identifying a person who has a copyright, etc., information on usage methods and conditions for permitting the use of a copyrighted work, etc. in the form of a digital watermark. Insertion into image data has been put into practical use.

  Normally, a digital watermark is inserted and detected in the form of electronic data, but a digital watermark corresponding to printing has also appeared. This makes it possible to detect a digital watermark from the printed image data even when the image data into which the digital watermark is inserted is printed.

  However, conventional digital watermarking that also supports printing is premised on reading a printed matter using a scanner. Under such a premise, the scale of the read image is known in advance, and the image is never rotated. Further, since the scanner has a high resolution of about 6400 dpi, it can detect a digital watermark embedded in a high frequency component. Further, unlike the case of reading with a camera, there is no case of camera shake. Further, since the image is emitted from the scanner with a constant light source, the signal-to-noise ratio and the contrast are good.

  On the other hand, cellular phones with built-in cameras and electronic mail functions and WWW (World Wide Web) browser functions have recently become widespread. If an image with a digital watermark inserted is read by the camera of such a mobile phone, and the image is sent from the mobile phone as an attached file of an e-mail or form information on a homepage, the WWW server side detects the digital watermark, Based on this, information is retrieved from the database, and the mobile phone can receive various services such as provision of such information from a WWW server, a mail server, or the like.

  Therefore, the advent of a digital watermark insertion method, apparatus and program thereof, and digital watermark detection method, apparatus and program thereof that enable detection of a digital watermark from an image read by a built-in camera or the like of a mobile phone is desired. It was.

  However, the scale of an image read by a built-in camera of a mobile phone is indefinite due to factors such as the distance between the image and the camera. In addition, an image read by a built-in camera of a mobile phone is often rotated not a little because of a variation factor such as a relative rotation angle between the image and the camera. Furthermore, the resolution of the images read by the built-in cameras of most mobile phones is low. For example, the number of pixels when a 3 cm × 4 cm image is read by a 6400 dpi scanner is about 76 million, whereas the number of built-in cameras of the majority of mobile phones currently popular is about 300,000. It is clearer than there is. Further, an image read by a built-in camera of a mobile phone may cause camera shake. Furthermore, since an image read with a built-in camera of a mobile phone is read under ordinary external light, the signal-to-noise ratio and contrast are often insufficient.

Therefore, even if the image is unscaled, rotated, has low resolution, has camera shake, has a low signal-to-noise ratio, and has a poor contrast, the data embedded in the form of a digital watermark is stabilized. A robust digital watermark insertion method, apparatus and program thereof, and digital watermark detection method, apparatus and program thereof have been invented and disclosed in Patent Document 1. ing.
JP 2004-15396 A

  However, according to the invention disclosed in Patent Document 1, the amount of data that can be embedded in an image is not so large. In addition, if a large amount of data is to be embedded in an image, the amount of digital watermark representing the data must be increased.

  Therefore, the present invention provides an electronic watermark insertion method, an apparatus and program thereof, an electronic watermark detection method, an apparatus, and an electronic watermark insertion method that can increase the amount of data that can be embedded in an image without visually conspicuous electronic watermarks. The purpose is to provide the program.

  According to the present invention, in the frequency domain, the jth (j = 1, 2,..., N) data to be inserted among the first predetermined number of candidate positions arranged on the circle are combined with each other. A sub-step of generating a non-zero frequency component for a second predetermined number of positions less than the first predetermined number; detecting a rotation angle of the image in the frequency domain; and j-th region of the image A sub-step of generating a non-zero frequency component for identifying the non-zero frequency component, a sub-step of combining the generated non-zero frequency component from the frequency domain to the spatial domain, and all or one of the j-th region of the original image. There is provided a digital watermark insertion method comprising: a sub-step of adding or subtracting an image obtained by the conversion to a block of a part, and repeating for N j.

  According to the present invention, the rotation angle of the image is detected in the frequency domain, and a non-zero frequency component for identifying the j1 (j1 = 1, 2,..., N1) region of the image is generated. A first generation sub-step, a first conversion sub-step of converting the non-zero frequency component generated in the first generation sub-step from the frequency domain to the spatial domain, and all or one of the j1 areas of the original image A sub-step of adding or subtracting the image obtained in the first conversion sub-step to a block of a portion, a step of repeating N1 j1, and a first predetermined number of candidate positions arranged on a circle in the frequency domain Non-zero in the second predetermined number less than the first predetermined number, which are positions that are combined with each other by the j2 (j2 = 1, 2,..., N2) data to be inserted. frequency A second generation sub-step for generating a minute, a second conversion sub-step for converting the non-zero frequency component generated in the second generation sub-step from the frequency domain to a spatial domain, and the j2 region of the original image And a sub-step of adding or subtracting the image obtained in the second conversion sub-step to all or a part of the block, and a step of repeating for N2 j2s. Is done.

  Further, according to the present invention, in the frequency domain, the jth (j = 1, 2,..., N) data to be inserted among the first predetermined number of candidate positions arranged on the circle are mutually connected. A first generation sub-step for generating a non-zero frequency component in a second predetermined number less than the first predetermined number at the combined position; and the non-zero frequency generated in the first generation sub-step A first transformation sub-step for transforming a component from the frequency domain to a spatial domain; and a sub-step for adding or subtracting the image obtained in the first transformation sub-step to all or a part of blocks of the j-th region of the original image; Are repeated for N j, a second generation step for generating a non-zero frequency component for detecting a rotation angle of an image in the frequency domain, and the generation generated in the second generation step. A second conversion step of converting the zero frequency component from the frequency domain to the spatial domain, and a step of adding or subtracting the image obtained in the second conversion step to all or a part of the blocks of the original image. A featured watermark insertion method is provided.

  Further, according to the present invention, the sub-step of obtaining one or more image blocks from the k-th (k = 1, 2,..., N) region of the image, and the data of the one or more image blocks from the spatial region. A sub-step of converting to the frequency domain, a sub-step of obtaining an amplitude of each frequency component based on each frequency component of the one or more image blocks, and an amplitude of each frequency component in the frequency domain The rotation angle of the image is detected by detecting the rotation angle of the image and detecting the position of the non-zero frequency component for identifying the jth (j = 1, 2,... N) region of the image. Sub-step for identifying the j-th region of the image, and the first of the rotation angles detected at the first predetermined number of positions on the circumference in the frequency region. Less than a predetermined number of A sub-step for detecting, for each circle of different radii, whether there is a non-zero frequency component at any one of a predetermined number of positions, and detecting that the position is at any one of the second predetermined number of positions. A sub-step of detecting the j-th data inserted in the image based on the combination of the positions of the non-zero frequency components, the step of repeating for N k, the N-th j-th data, and There is provided a digital watermark detection method comprising: restoring data based on rank j.

  Further, according to the present invention, the sub-step of obtaining one or more image blocks from the k-th (k = 1, 2,..., N) region of the image, and the data of the one or more image blocks from the spatial region. A sub-step of converting to the frequency domain, a sub-step of obtaining an amplitude of each frequency component based on each frequency component of the one or more image blocks, and an amplitude of each frequency component in the frequency domain The rotation angle of the image is detected, the position of the non-zero frequency component for identifying the j-th (j = 1, 2,... N) region of the image is detected, and the detected non-zero frequency component is detected. A sub-step of detecting the data number j based on the position and the region number k; a sub-step of detecting the rotation angle of the image based on the detected position of the non-zero frequency component; First predetermined above It is different whether there is a non-zero frequency component at any of the second predetermined number smaller than the first predetermined number among the positions with the detected rotation angle as a rotation reference. Inserted into the image based on a combination of a sub-step for detecting each circle of radius and the position of the non-zero frequency component detected at any one of the second predetermined number of positions. A digital watermark detection method comprising: a sub-step of detecting j-th data; a step of repeating for N k; and a step of restoring data based on the N j-th data Provided.

  Further, according to the present invention, the sub-step of obtaining one or more image blocks from the k-th (k = 1, 2,..., M × N) region of the image, and the data of the one or more image blocks are spatialized. A sub-step of transforming from an area to a frequency domain, a sub-step of obtaining an amplitude of each frequency component based on each frequency component of the one or more image blocks, and an amplitude of each frequency component in the frequency domain A small rotation angle of the image by detecting a rotation angle of the image and detecting a position of a non-zero frequency component for identifying a jth (j = 1, 2,... N) region of the image. A sub-step of detecting the image, a step of repeating for M × N k, a rotation angle of the image is detected, and a j-th (j = 1, 2,..., N) region of the image is identified. M × N k of non-zero frequency components for A step of obtaining a correspondence relationship between j and k and a large rotation angle according to a distribution over a range, a step of obtaining a rotation angle of an image based on the small rotation angle and the large rotation angle, and M regions having the same value of j. The first predetermined number on the circumference in the frequency domain based on the step of adding the amplitudes of the frequency components and the amplitudes of the frequency components added together for each of the M regions having the same value of j. It is different whether there is a non-zero frequency component at any of the second predetermined number smaller than the first predetermined number among the positions with the detected rotation angle as a rotation reference. A combination of the step of detecting for each circle of radius and the position of the non-zero frequency component detected to be in any of the second predetermined number of positions for each of M regions having the same value of j Is inserted into the image based on Detecting data of the j that, digital watermark detection method characterized by comprising the steps of restoring the data based on the data and rank j of N j-th, is provided.

  Further, according to the present invention, the sub-step of obtaining one or more image blocks from the k-th (k = 1, 2,..., N) region of the image, and the data of the one or more image blocks from the spatial region. A sub-step of converting to the frequency domain, a sub-step of obtaining an amplitude of each frequency component based on each frequency component of the one or more image blocks, and an amplitude of each frequency component in the frequency domain A small rotation angle of the image by detecting a rotation angle of the image and detecting a position of a non-zero frequency component for identifying the sth (s = 1, 2,... N / M) region of the image. Detecting the rotation angle of the image, and identifying the sth (s = 1, 2,..., N) region of the image. Spans N k of non-zero frequency components Based on the cloth, the step of determining the correspondence between k and the data number j and the large rotation angle, the step of determining the rotation angle of the image based on the small rotation angle and the large rotation angle, and the amplitude of each frequency component, Any one of a second predetermined number smaller than the first predetermined number among the first predetermined number of positions on the circumference in the frequency domain and the rotation angle detected as a rotation reference. A sub-step for detecting whether or not there is a non-zero frequency component at each position of each circle of different radii, and the non-zero frequency component detected at any one of the second predetermined number of positions. A sub-step of detecting the j-th data inserted in the image based on a combination of positions; a step of repeating for N k; and data based on the N j-th data and the rank j Steps to restore Digital watermark detection method comprising: a is provided.

Furthermore, according to the present invention, a step of obtaining one or more image blocks from an image, a step of converting data of the one or more image blocks from a spatial domain to a frequency domain, and each frequency component of the one or more image blocks And obtaining the amplitude of each frequency component, and detecting the position of the non-zero frequency component for detecting the rotation angle of the image based on the amplitude of each frequency component in the frequency domain. Detecting the rotation angle of the image, and correcting the rotation angle of the image based on the rotation angle, and further, jth (j = 1, 2,...) Of the image whose rotation angle is corrected. N) a sub-step of obtaining one or more image blocks from the region;
A sub-step of transforming data of the one or more image blocks from a spatial domain to a frequency domain; a sub-step of obtaining an amplitude of each frequency component based on each frequency component of the one or more image blocks; Each of the circles with different radii indicates whether or not there is a non-zero frequency component at any one of the second predetermined number of positions less than the first predetermined number of the first predetermined number of positions on the circumference. J-th data inserted in the image based on a combination of a sub-step to be detected every time and the position of the non-zero frequency component detected to be at any one of the second predetermined number of positions A digital watermark detection method comprising: a sub-step of detecting N j, a step of repeating N for j, and a step of restoring data based on the N j-th data and the rank j Be done

  According to the present invention, since the data is divided into a plurality of pieces and each piece of divided data is inserted into each divided image in the form of a digital watermark, the data that can be embedded in the image while maintaining a small number of digital watermarks to be embedded in each divided image The amount of information can be increased.

  In addition, according to the present invention, the rotation angle of the image can be detected, and the digital watermark for identifying each area of the image can detect which divided image is at what angle and where. The original data can be restored based on the divided data.

  The best mode for carrying out the present invention will be described below in detail with reference to the drawings.

  FIG. 1 shows the amplitude in the frequency domain of a digital watermark used in an embodiment of the present invention.

  Referring to FIG. 1, there are two types of digital watermarks, the first digital watermark is inserted into the left half of the image (hereinafter referred to as “left image”), and the second digital watermark is referred to as the right half of the image (hereinafter referred to as “left image”). It is inserted into “right image”). Looking at the amplitude in the frequency domain of the digital watermark to be inserted into the left image, the first digital watermark has a non-zero frequency component distributed in the outer circle 601 and a non-zero frequency component distributed in the inner circle 602. Similarly, when the amplitude in the frequency domain of the digital watermark inserted into the right image is viewed, the second digital watermark includes a non-zero frequency component distributed in the outer circle 603 and a non-zero frequency component distributed in the inner circle 604.

The non-zero frequency components distributed in the outer circle are 20 points excluding two points overlapping the x-axis and two points overlapping the y-axis, out of 24 equal points obtained by dividing the outer circle into 24 equal parts. Arranged at 10 of the points. Since the distribution of the amplitudes of the non-zero frequency components is the object about the origin, the number when selecting 10 points from 20 points is equal to the number when selecting 5 points from 10 points. . Accordingly, ₁₀ C ₅ = 252 first data can be inserted into the left image. Similarly, ₁₀ C ₅ = 252 types of second data can be inserted into the right image. Therefore, a total of 252 ^ 2 = 63504 data can be inserted into the image.

  In the inner circle 602, a non-zero frequency component for detecting the rotation angle of the image is arranged. Similarly, a non-zero frequency component for detecting the rotation angle of the image is also arranged in the inner circle 604. In the example shown in FIG. 1, six non-zero frequency components are arranged in the inner circle 602, and six non-zero frequency components are arranged in the inner circle 604, and the non-zero frequency components are arranged in the inner circle 602. And the distribution of non-zero frequency components arranged in the inner circle 604 are different. The amplitude of the non-zero frequency component placed in each inner circle is point-symmetric with respect to the origin, but this is inevitable, and the point symmetry is lost no matter how the non-zero frequency component is placed. Therefore, it is impossible to detect that the image is viewed at 180 degrees only by non-zero frequency components distributed in one circle. Generally, the distribution of non-zero frequency components when rotated by an angle other than 180 degrees does not match the distribution of non-zero frequency components when not rotated, and the non-zero frequency components arranged in the inner circle 602 If one or more non-zero frequency components are arranged in each of the inner circles 602 and 604 so as to satisfy the two conditions that the distribution and the distribution of the non-zero frequency components arranged in the inner circle 604 do not match, A rotation angle in the range of 180 degrees can be detected, and whether the image is upside down (whether it is rotated 180 degrees) can be detected.

  The figure denoted by reference numeral 605 is a figure representing the non-zero frequency component for detecting the rotation angle by the center of gravity, and the figure denoted by reference numeral 606 is a figure representing data by symbols. The following description will be made using such a diagram.

  FIG. 2 shows the arrangement and orientation during printing of a digital watermark as shown in FIG. 1 and the arrangement and orientation during shooting with a camera.

  Referring to FIG. 2, at the time of printing, the center of gravity of the non-zero frequency component of the rotation angle detection digital watermark of the left image is at a position of 135 degrees and a position of 315 degrees. The center of gravity of the zero frequency component is at a position of 45 degrees and a position of 225 degrees. A digital watermark of the second data is inserted into the left image, and a digital watermark of the first data is inserted into the right image.

  When the camera is photographed at a rotation angle near 0 degrees, the center of gravity of the non-zero frequency component of the rotation angle detection digital watermark of the left image is at a position of 135 degrees and a position of 315 degrees, as in printing. The center of gravity of the non-zero frequency component of the digital image for detecting the rotation angle of the right image is at a position of 45 degrees and a position of 225 degrees, and the digital watermark of the second data is inserted in the left image, Is inserted with a digital watermark of the first data.

  On the other hand, when the camera is photographed at a rotation angle near 180 degrees, unlike the printing, the center of gravity of the non-zero frequency component of the left image rotation angle detection digital watermark is at a position of 45 degrees and a position of 225 degrees. The center of gravity of the non-zero frequency component of the digital image for detecting the rotation angle of the right image is at a position of 135 degrees and a position of 315 degrees, and the digital watermark of the first data is inserted in the left image. Is inserted with a digital watermark of the second data.

  Therefore, regardless of the rotation angle, the first data corresponds to the rotation angle detection digital watermark with the centroids at 45 degrees and 225 degrees, and the second data has the centroids at the positions 135 degrees and 225 degrees. This corresponds to a non-zero frequency component of a certain rotation angle detection digital watermark. Therefore, the image captured by the camera is divided into left and right, the left image and the right image are processed separately, the data and the rotation angle detection watermark are detected for each image, and the value of the first data or the second data By identifying whether the data is the first data or the second data, the final entire data can be obtained. Specifically, for example, if the data of each divided image is inserted by arranging non-zero frequency components at 10 positions out of 20 positions, the first data is multiplied by 252. A value obtained by adding the second data to the value is the final overall data.

  2 shows the state of digital watermark insertion and digital watermark detection when the image is divided into left and right parts, while FIG. 3 shows digital watermark insertion and digital watermark when the image is divided into four parts vertically and horizontally. This shows the state of detection.

  Referring to FIG. 3, at the time of printing, the center of gravity of the non-zero frequency component of the rotation angle detection digital watermark in the upper right image is at 45 degrees and 225 degrees. The centroid of the zero frequency component is at 90 degrees and 270 degrees, the centroid of the non-zero frequency component of the rotation angle detection digital watermark in the lower left image is at 135 degrees and 315 degrees, The center of gravity of the non-zero frequency component of the watermark for detecting the rotation angle of the lower image is at a position of 0 degrees and a position of 180 degrees, the digital watermark of the first data is inserted into the upper right image, and the first watermark is inserted into the upper left image. An electronic watermark of two data is inserted, an electronic watermark of the third data is inserted into the lower left image, and an electronic watermark of the fourth data is inserted into the lower right image.

  In the example shown in FIG. 3, for example, if data is inserted into each divided image by arranging non-zero frequency components at 10 positions out of 20 positions, 252 ^ 4 = 4 032,758,016 types of data can be represented.

  When the camera is photographed at a rotation angle near 0 degrees, the center of gravity of the non-zero frequency component of the rotation angle detection digital watermark in the upper right image is at a position of 45 degrees and a position of 225 degrees, as in printing. The centroid of the non-zero frequency component of the rotation angle detection digital watermark in the upper left image is at a position of 90 degrees and 270 degrees, and the centroid of the non-zero frequency component of the rotation angle detection digital watermark in the lower left image is 135 degrees. The center of gravity of the non-zero frequency component of the digital image for detecting the rotation angle of the lower right image is at the position of 0 degree and the position of 180 degrees, and the upper right image has the first data The digital watermark is inserted, the digital watermark of the second data is inserted in the upper left image, the digital watermark of the third data is inserted in the lower left image, and the fourth data of the fourth data is inserted in the lower right image. Insert watermark It has been.

  On the other hand, when the camera is photographed at a rotation angle near 90 degrees, unlike the case of printing, the center of gravity of the non-zero frequency component of the rotation angle detection digital watermark in the upper right image is at the positions of 90 degrees and 270 degrees. The center of gravity of the non-zero frequency component of the rotation angle detection digital watermark of the upper left image is at the positions of 135 degrees and 315 degrees, and the center of gravity of the non-zero frequency component of the rotation angle detection digital watermark of the lower left image is 0 The center of gravity of the non-zero frequency component of the digital watermark for detecting the rotation angle of the lower right image is at the position of 45 degree and the position of 225 degree. Of the first data is inserted in the upper left image, the second data digital watermark is inserted in the lower left image, and the third data is inserted in the lower right image. Electronic transparency There has been inserted.

  Also, when the image is taken at a rotation angle of about 180 degrees by the camera, the center of gravity of the non-zero frequency component of the digital image for detecting the rotation angle of the upper right image is at a position of 135 degrees and a position of 315 degrees, unlike at the time of printing. The center of gravity of the non-zero frequency component of the digital watermark for detecting the rotation angle of the upper left image is at a position of 0 degree and the position of 180 degree, and the center of gravity of the non-zero frequency component of the digital watermark for detecting the rotation angle of the lower left image is 45 The center of gravity of the non-zero frequency component of the digital watermark for detecting the rotation angle of the lower right image is at the position of 90 degrees and the position of 270 degrees. The digital watermark of the fourth data is inserted in the upper left image, the digital watermark of the first data is inserted in the lower left image, and the second data is inserted in the lower right image. Electronic transparency Teeth are inserted.

  Furthermore, when the camera is photographed at a rotation angle of about 270 degrees, unlike the printing, the center of gravity of the non-zero frequency component of the rotation angle detection digital watermark in the upper right image is at a position of 0 degrees and a position of 180 degrees. The center of gravity of the non-zero frequency component of the rotation angle detection digital watermark of the upper left image is at 45 degrees and 225 degrees, and the center of gravity of the non-zero frequency component of the rotation angle detection digital watermark of the lower left image is 90 The center of gravity of the non-zero frequency component of the digital image for detecting the rotation angle of the lower right image is at the position of 135 degrees and the position of 315 degrees, and the second data Is inserted in the upper left image, the fourth data watermark is inserted in the lower left image, and the first data is inserted in the lower right image. Electronic transparency Teeth are inserted.

  In the case of the left and right two divisions shown in FIG. 2, the center of gravity of the non-zero frequency component for detecting the rotation angle of the other divided images is not considered, and only the center of gravity of the non-zero frequency component for detecting the rotation angle of the divided image is used. It was possible to determine whether the data of the divided image is the first data or the second data. On the other hand, such a determination cannot be made in the case of the four divisions shown in FIG. That is, for example, when the rotation angle is 0 degrees, the data corresponding to the centroids of the rotation angle detection non-zero frequency components of 45 degrees and 225 degrees is the first data, whereas the rotation angle is 90 degrees. If the rotation angle is 180 degrees, the data corresponding to the centroids of the non-zero frequency components for rotation angle detection of 45 degrees and 225 degrees is the third data, and when the rotation angle is 180 degrees, 45 degrees and 225 degrees. The data corresponding to the center of gravity of the non-zero frequency component for rotation angle detection is the first data. When the rotation angle is 270 degrees, it corresponds to the center of gravity of the non-zero frequency component for rotation angle detection of 45 degrees and 225 degrees. The data to be performed is third data. However, by detecting which divided image the divided image is and which rotation angle detection digital watermark having a center of gravity is detected from the divided image, what number is added to the divided image. Can be detected. Specifically, taking the upper right divided image as an example, the divided image is the upper right divided image, and a viewpoint angle detection digital watermark having centroids of 45 degrees and 225 degrees is detected in the divided image. Thus, it can be detected that the first data is inserted into the divided image, the divided image is the upper right divided image, and the divided image has a center of gravity of 90 degrees and 270 degrees. Since the viewpoint angle detection digital watermark is detected, it can be detected that the fourth data is inserted into the divided image, the divided image is the upper right divided image, and the divided image. Since the viewpoint angle detection digital watermark having centroids of 135 degrees and 315 degrees is detected, it can be detected that the third data is inserted into the divided image. Since the divided image is the upper right divided image and the viewpoint angle detection digital watermark having centroids of 0 degrees and 180 degrees is detected in the divided image, the second data is inserted into the divided image. Can be detected.

  In the method shown in FIG. 3, it is possible to detect whether the rotation angle of the image is around 0 degree, around 90 degrees, around 180 degrees, or around 270 degrees. In the method shown in FIG. 2, it can be detected whether the rotation angle of the image is around 0 degree or around 180 degrees. However, the rotation angle of the image is around 90 degrees and the image It cannot be detected that the rotation angle is around 270 degrees. However, not only when taking an image with the camera lying sideways, but also when taking an image with the camera held vertically, the rotation angle of the image is around 90 degrees and the rotation angle of the image. It is preferable that it is possible to detect that is around 270 degrees. FIG. 4 is a diagram for explaining a method that enables this.

  In this method, as in the case shown in FIG. 2, the image is divided into left and right parts and a digital watermark is inserted, but the image is divided into four parts vertically and horizontally to detect the digital watermark.

  When the camera is photographed at a rotation angle close to 0 degrees, the center of gravity of the non-zero frequency component of the rotation angle detection digital watermark in the upper right image is at a position of 45 degrees and a position of 225 degrees, and the rotation angle of the upper left image is detected. The centroid of the non-zero frequency component of the digital watermark is at the positions of 135 degrees and 315 degrees, and the centroid of the non-zero frequency component of the rotation angle detection watermark of the lower left image is at the position of 135 degrees and 315 degrees. The center of gravity of the non-zero frequency component of the digital watermark for detecting the rotation angle of the lower right image is at the position of 45 degrees and the position of 225 degrees, and the digital watermark of the first data is inserted in the upper right image. In addition, a digital watermark of the second data is inserted in the upper left image, a digital watermark of the second data is inserted in the lower left image, and a digital watermark of the first data is inserted in the lower right image. Yes.

  On the other hand, the center of gravity of the non-zero frequency component of the rotation angle detection digital watermark in the upper right image is at a position of 135 degrees and a position of 315 degrees when the image is photographed at a rotation angle of about 90 degrees by the camera. The centroid of the non-zero frequency component of the angle detection digital watermark is at a position of 135 degrees and 315 degrees, and the centroid of the non-zero frequency component of the rotation angle detection digital watermark of the lower left image is a position of 45 degrees and 225. The center of gravity of the non-zero frequency component of the digital watermark for detecting the rotation angle of the lower right image is at the position of 45 degrees and the position of 225 degrees, and the digital watermark of the first data is inserted in the upper right image The digital watermark of the first data is inserted in the upper left image, the digital watermark of the second data is inserted in the lower left image, and the digital watermark of the second data is inserted in the lower right image. Is There.

  When the camera is photographed at a rotation angle of about 180 degrees, the center of gravity of the non-zero frequency component of the rotation angle detection digital watermark of the upper right image is at a position of 135 degrees and a position of 315 degrees. The center of gravity of the non-zero frequency component of the digital watermark for angle detection is at the position of 45 degrees and 225 degrees, and the center of gravity of the non-zero frequency component of the watermark for detection of the rotation angle of the lower left image is at the position of 45 degrees and 225. The center of gravity of the non-zero frequency component of the watermark for detecting the rotation angle of the lower right image is at the position of 135 degrees and 315 degrees, and the watermark of the second data is inserted in the upper right image. The digital watermark of the first data is inserted in the upper left image, the digital watermark of the first data is inserted in the lower left image, and the digital watermark of the second data is inserted in the lower right image. The To have.

  Furthermore, when the camera is photographed at a rotation angle of about 270 degrees, the center of gravity of the non-zero frequency component of the rotation angle detection digital watermark in the upper right image is at a position of 45 degrees and a position of 225 degrees. The centroids of the non-zero frequency components of the angle detection watermark are located at 45 degrees and 225 degrees, and the centroids of the non-zero frequency components of the rotation angle detection watermark of the lower left image are 135 degrees and 315 The center of gravity of the non-zero frequency component of the watermark for detecting the rotation angle of the lower right image is at the position of 135 degrees and 315 degrees, and the watermark of the second data is inserted in the upper right image. A digital watermark of the second data is inserted in the upper left image, a digital watermark of the first data is inserted in the lower left image, and a digital watermark of the first data is inserted in the lower right image The To have.

  In the case of the method shown in FIG. 4, the center of gravity of the non-zero frequency component for detecting the rotation angle of the other divided image is not considered, and only the center of gravity of the non-zero frequency component for detecting the rotation angle of the divided image is used. Whether the data of the divided image is the first data or the second data cannot be determined. That is, for example, when the rotation angle is 0 degree, the first data corresponds to the centroids of the non-zero frequency components for rotation angle detection of 45 degrees and 225 degrees in the first quadrant, and 45 degrees and 45 degrees in the fourth quadrant. When the first data corresponds to the center of gravity of the non-zero frequency component for detecting the rotation angle of 225 degrees, while the rotation angle is 90 degrees, the non-detection angle for detecting the rotation angles of 135 degrees and 315 degrees in the first quadrant. The first data corresponds to the center of gravity of the zero frequency component, and the first data corresponds to the center of gravity of the non-zero frequency component for rotation angle detection of 135 degrees and 315 degrees in the second quadrant, and the rotation angle is 180 degrees. Corresponds to the center of gravity of the non-zero frequency component for rotation angle detection of 45 degrees and 225 degrees in the third quadrant, and the non-zero frequency component for rotation angle detection of 45 degrees and 225 degrees in the fourth quadrant. When the first data corresponds to the heart and the rotation angle is 270 degrees, the first data corresponds to the centroids of the non-zero frequency components for rotation angle detection of 135 degrees and 315 degrees in the third quadrant. The first data corresponds to the centroids of the non-zero frequency components for rotation angle detection of 135 degrees and 315 degrees in the quadrant.

  Therefore, in the case of the left / right two-division detection at the time of insertion shown in FIG. It is determined whether it is around 0 degree, around 90 degrees, around 180 degrees, or around 270 degrees.

  In this way, the rotation angle of the image is detected by comprehensively looking at the center of gravity of the non-zero frequency component for rotation angle detection in each divided image, and the data detected from each divided image is based on the rotation angle. It is possible to determine what number the data is, and to obtain final overall data based on the data detected from each divided image and its number. Specifically, for example, if the data of each divided image is inserted by arranging non-zero frequency components at 10 positions out of 20 positions, 252 ^ 2 is set as the first data. A value obtained by adding the second data to the multiplied value is the final overall data.

  In any of the above methods, after the image taken by the camera is divided, the digital watermark for detecting the rotation angle is detected in each divided image. In contrast, the method described with reference to FIG. 5 detects the rotation angle detection digital watermark inserted into the entire image before dividing the image taken by the camera, and based on this, the rotation angle of the image is detected. After the image is detected and the rotation of the image is corrected based on the rotation angle, the image is divided, and the digital watermark of the data is detected in each divided image.

  Referring to FIG. 5, as indicated by reference numeral 607, a digital watermark for detecting the rotation angle of the image is inserted into the entire image. Further, for example, when the data is divided into two and 252 ^ 2 = 63504 data is represented, as shown by reference numeral 608, the image is divided into two, and the digital watermark representing the i-th data in the i-th divided image Is inserted (i = 1, 2). Similarly, for example, when the data is divided into four and 252 ^ 4 = 40327558016 data is represented, as shown by reference numeral 609, the image is divided into four and the i-th divided image represents the i-th data. A watermark is inserted (i = 1 to 4). For example, when the data is divided into six to represent 252 ^ 6 = 2.561 × 10 ^ 14 kinds of data, as indicated by reference numeral 609, the image is divided into six and the i-th divided image is divided into the i-th data. Is inserted (i = 1 to 6). For example, when the data is divided into nine and 252 ^ 9 = 4.098 × 10 ^ 21 kinds of data are represented, as shown by reference numeral 609, the image is divided into nine and the i-th divided image is divided into the i-th data. Is inserted (i = 1 to 9).

  Next, a digital watermark insertion apparatus and a digital watermark detection apparatus that perform the above-described digital watermark insertion method and digital watermark detection method will be described.

  FIG. 6 is a block diagram showing the configuration of the digital watermark insertion apparatus according to the embodiment of the present invention.

  Referring to FIG. 6, a digital watermark insertion apparatus according to an embodiment of the present invention includes a rotation angle detection pattern generation unit 101, a data holding unit 102, a data pattern generation unit 103, a synthesis unit 104, an inverse discrete Fourier transform unit 105, two A valuation unit 106 and an addition unit (subtraction unit) 107 are provided. These parts may be realized by hardware, but may be realized by a computer reading and executing a program for causing a computer to function as these parts.

  Next, with reference to FIGS. 6 and 7, a digital watermark insertion method performed by the digital watermark insertion apparatus in the case of inserting a digital watermark for detecting a rotation angle and partial data different for each divided image for each divided image will be described. explain.

  In this digital watermark insertion method, steps S152 to S156 are repeated for the first divided image to the Nth divided image (step S151). In each iteration, do the following:

  First, the rotation angle detection pattern generation unit 101 generates a jth rotation angle detection pattern (step S151). This pattern is composed of non-zero frequency components.

Next, the data pattern generation unit 103 reads the entire data held in the data holding unit 102, obtains the j-th data according to the value of j currently handled, and the j-th data according to the j-th data. The data pattern is generated (step S152). This pattern is also composed of non-zero frequency components. In the case where the jth data pattern is as shown in FIG.
Dj = mod (D, 252 ^ (N + 1−j)) / 252 ^ (N−j)
It is represented by

  Next, the combining unit 104 generates a combined pattern by adding the jth rotation angle detection pattern generated in step S151 and the jth data pattern generated in step S152 (step S153).

  Next, the inverse discrete Fourier transform unit 105 obtains a spatial domain pattern by applying an inverse discrete Fourier transform to the pattern synthesized in step S153 (step S154).

  Next, the binarization unit 106 binarizes the space area pattern obtained in step S154 (step S155). That is, for each pixel, if the pixel value of the spatial pattern obtained in step S154 is 0 or more, a predetermined value (for example, 10) is set. If the pixel value of the spatial pattern obtained in step S154 is less than 0, , And a predetermined value (for example, −10).

  Next, the adding unit (subtracting unit) 107 adds or subtracts the spatial region pattern binarized in step S106 to all or a part of blocks of the j-th region of the yellow component of the original image (step S156). Note that the binarization unit 106 performs binarization in order to facilitate detection of the digital watermark by the digital watermark detection apparatus, and may be omitted. In this case, the spatial pattern output from the inverse discrete Fourier transform unit 105 is directly added to or subtracted from the j-th region of the yellow component of the original image without being binarized.

  Thereafter, the image into which the digital watermark is inserted by the above method is printed on a printed matter.

  Next, refer to FIGS. 6 and 8 for a digital watermark insertion method performed by the digital watermark insertion apparatus in the case of inserting a rotation angle detection digital watermark that is common among divided images (common to the entire image) into the entire image. To explain.

  First, the rotation angle detection pattern generation unit 101 generates a rotation angle detection pattern common to the divided images (step S151). This pattern is composed of non-zero frequency components.

  Next, steps S553 to S557 are repeated for the first divided image to the Nth divided image (step S552). In each iteration, do the following:

First, the data pattern generation unit 103 reads the entire data held in the data holding unit 102, obtains the j-th data according to the value of j currently handled, and the j-th data according to the j-th data. A data pattern is generated (step S553). This pattern is also composed of non-zero frequency components. In the case where the jth data pattern is as shown in FIG.
Dj = mod (D, 252 ^ (N + 1−j)) / 252 ^ (N−j)
It is represented by

  Next, the synthesis unit 104 generates a synthesis pattern by adding the common rotation angle detection pattern generated in step S551 and the jth data pattern generated in step S553 (step S554).

  Next, the inverse discrete Fourier transform unit 105 obtains a spatial domain pattern by performing inverse discrete Fourier transform on the pattern synthesized in step S554 (step S555).

  Next, the binarization unit 106 binarizes the space area pattern obtained in step S555 (step S556). That is, for each pixel, if the pixel value of the spatial pattern obtained in step S555 is 0 or more, a predetermined value (for example, 10) is set. If the pixel value of the spatial pattern obtained in step S555 is less than 0, , And a predetermined value (for example, −10).

  Next, the adding unit (subtracting unit) 107 adds or subtracts the spatial region pattern binarized in step S556 to all or a part of blocks of the j-th region of the yellow component of the original image (step S557). Note that the binarization unit 106 performs binarization in order to facilitate detection of the digital watermark by the digital watermark detection apparatus, and may be omitted. In this case, the spatial pattern output from the inverse discrete Fourier transform unit 105 is directly added to or subtracted from the j-th region of the yellow component of the original image without being binarized.

  Thereafter, the image into which the digital watermark is inserted by the above method is printed on a printed matter.

  FIG. 9 is a block diagram illustrating a configuration of a digital watermark detection apparatus for performing the detection method (data two-division, rotation angle detection digital watermark two-division, and detection two-division) described with reference to FIG.

  Referring to FIG. 9, this digital watermark detection apparatus includes a spatial filter 201, a blocking unit 202, a discrete Fourier transform unit 203, an amplitude calculation unit 204, an amplitude integration unit 205, a rotation angle detection / region number identification unit 206, and partial data. A detection unit 207 and a data detection unit 208 are provided. These parts may be realized by hardware, but may be realized by a computer reading and executing a program for causing the computer to function as these parts.

  Next, a digital watermark detection method performed by the digital watermark detection apparatus shown in FIG. 9 will be described with reference to FIGS.

  In this digital watermark detection method, steps S252 to S259 are repeated for the first divided image to the second divided image (step S251). In each iteration, do the following:

  First, the spatial filter 201 is applied to the current divided image (kth divided image) of the blue component of the image captured by the camera (step S252). This spatial filter has tap coefficients as shown in FIG. 11, for example. As is apparent from the tap coefficient, this spatial filter has a band-pass characteristic. Therefore, by applying this spatial filter 201 to the image, it is possible to suppress low frequency components and high frequency components resulting from the original image.

  Next, the blocking unit 202 divides the divided image to which the spatial filter is applied into one or more blocks having a size for performing discrete Fourier transform (step S253).

  Next, the discrete Fourier transform unit 203 performs a discrete Fourier transform on the image of each block to obtain a frequency component (step S254).

  Next, the amplitude calculation unit 204 obtains the amplitude of each frequency component of each block (step S255).

  Next, the amplitude integrating unit 205 obtains an integrated amplitude of each frequency component for the divided image by integrating frequency components of the same frequency among the blocks included in the divided image (step S256).

  Next, the rotation angle detection / region number identification unit 206 inputs the integrated amplitude of each frequency component and observes the distribution of non-zero frequency components distributed in the inner circle 602 in FIG. Is a first divided image or a second divided image (step S257). That is, referring to FIG. 2, when the center of gravity is at a position of 45 degrees and a position of 225 degrees, it is determined that the current divided image is the first divided image, and the position of the center of gravity is 135 degrees and 315 degrees. If it is in the position, it is determined that the current divided image is the second divided image.

  Note that obtaining the two centroids of the non-zero frequency component for detecting the rotation angle is, for example, selecting an arbitrary one non-zero frequency component and having a relatively short distance from the non-zero frequency component ( It is assumed that half of the non-zero frequency components including the number of non-zero frequency components are extracted, and that the half of the non-zero frequency components belong to the first group and the remaining half belong to the second group. This is possible by obtaining the center of gravity of each group.

  Next, the rotation angle detection / region number identification unit 206 inputs the integrated amplitude of each frequency component, sees the distribution of non-zero frequency components distributed in the inner circle 602 of FIG. 1, and further makes a determination in step S257. By looking at the number of the current divided image, the rotation angle of the image is detected (step S258). That is, if it is determined that the current divided image is the left image and the second divided image, the non-zero frequency component with the center of gravity at the position of 135 degrees when the image is not rotated is used as a reference. Based on the position of the actual non-zero frequency component whose center of gravity is near 135 degrees, a rotation angle near 0 degrees is detected, and it is determined that the current divided image is the right image and the first divided image. If the image is not rotated, the center of gravity when the image is not rotated is based on the non-zero frequency component at a position of 45 degrees. If a rotation angle near 0 degrees is detected and it is determined that the current divided image is the left image and the first divided image, the center of gravity when the image is not rotated is at a position of 45 degrees. Center of gravity based on non-zero frequency components Since the rotation angle near 180 degrees is detected based on the position of the actual non-zero frequency component at the position near 5 degrees, it is determined that the current divided image is the right image and the second divided image. If there is a non-zero frequency component with the center of gravity at a position of 135 degrees when the image is not rotated, the center of gravity is around 180 degrees based on the actual position of the non-zero frequency component at a position near 135 degrees. The rotation angle of is detected.

  Next, the partial data detection unit 206 receives the integrated amplitude of each frequency component, detects data by looking at the distribution of non-zero frequency components distributed in the outer circle 601 in FIG. j data (step S259). Here, the number j is the number of the divided image detected in step S257.

  When the repetition of step S251 ends, the entire data is detected based on the Nth data from the first data obtained by performing step S259 N times (step S260).

  In steps S257 and S258, it is necessary to detect the polar angle in the frequency domain of each non-zero frequency component for detecting the rotation angle data distributed in the inner circle 602. For this purpose, the following method is used. .

  That is, referring to FIG. 12, first, the radius as a parameter is initialized to zero (step S901), and there are N1 (6 in the example of FIG. 1) non-zero frequency components on the circumference of the current radius. Whether or not (step S902).

  In step S902, the following is performed. That is, in the example of FIG. 1, a window is provided on a circle of the current radius for each angle that equally divides the circumference into 16 parts, and the non-zero amplitude exceeding the threshold value after passing all frequency components through the window. Whether or not N1 frequency components can be detected is examined while shifting the window angle. If it can be detected, the N1 non-zero frequency components are non-zero frequency components to be detected.

  If not (NO in step S902), the radius is increased by the unit length ΔR in the frequency domain (step S903), and the process returns to step S902.

  If there is (YES in step S902), the angle of each non-zero frequency component for detecting the rotation angle distributed on the circle of the current radius is detected in polar coordinates in the frequency domain (step S904).

  Although not shown in FIG. 12, if N1 non-zero frequency components cannot be detected even when the radius reaches the maximum radius, the width of the window in the radial direction is increased and Steps S902 and S903 are repeated. If N1 non-zero frequency components cannot be detected at any radius, the threshold value is lowered after returning the radial width of the window, and steps S902 and S903 are repeated. If N1 non-zero frequency components cannot be detected at any radius, the width of the window in the radial direction is expanded again and steps S902 and S903 are repeated. The same is repeated thereafter.

  In step S259, it is necessary to detect the angle of each non-zero frequency component representing the data distributed in the outer circle 601 in the frequency domain in the frequency domain. For this purpose, the following method is used.

  That is, referring to FIG. 13, first, the amplitude positions of all frequency components are corrected based on the rotation angle detected in step 258 (step S921).

  Next, the parameter radius is initialized to the radius of the inner circle detected in step 257 (step S922), and N2 (10 in the example of FIG. 1) non-zero frequency components are present on the circumference of the current radius. It is determined whether or not there is (step S923).

  In step S923, the following is performed. That is, in the example shown in FIG. 1, a window having a predetermined width is provided at 20 points where there may be non-zero frequency components representing data on a circle of the current radius, and all frequency components are It is checked whether 10 non-zero frequency components having an amplitude exceeding the threshold value can be detected after passing through the window. If it can be detected, the 10 non-zero frequency components are non-zero frequency components to be detected.

  If not (NO in step S923), the radius is increased by the unit length ΔR in the frequency domain (step S924), and the process returns to step S923.

  If there is (YES in step S923), the j-th data is detected by the following method (step S925). That is, the angle of each non-zero frequency component representing the data distributed on the circle with the current radius in the frequency domain is detected. Then, the value of the data represented by the non-zero frequency component is obtained from the distribution of the non-zero frequency component. Then, according to the number j detected in step S257, the data is set as jth data.

  Although not shown in FIG. 13, when ten non-zero frequency components cannot be detected even when the radius reaches the maximum radius, the width of the window in the radial direction is widened, and Steps S923 and S924 are repeated. If 10 non-zero frequency components cannot be detected at any radius, the threshold value is lowered after returning the radial width of the window, and steps S923 and S924 are repeated. If 10 non-zero frequency components cannot be detected at any radius, the width of the window in the radial direction is increased again and steps S923 and S924 are repeated. The same is repeated thereafter.

  FIG. 14 is a block diagram showing a configuration of a digital watermark detection apparatus for performing the detection method (data 4-division / rotation angle detection digital watermark 4-division / detection 4-division) described with reference to FIG.

  Referring to FIG. 14, this digital watermark detection apparatus includes a spatial filter 201, a blocking unit 202, a discrete Fourier transform unit 203, an amplitude calculation unit 204, an amplitude integration unit 205, a rotation angle detection / pattern number identification unit 301, and partial data. A number detection unit 302, a partial data detection unit 303, and a data detection unit 304 are provided. These parts may be realized by hardware, but may be realized by a computer reading and executing a program for causing the computer to function as these parts.

  Next, a digital watermark detection method performed by the digital watermark detection apparatus shown in FIG. 14 will be described with reference to FIGS.

  In this digital watermark detection method, first, steps S352 to S360 are repeated for the first to fourth divided images (step S351). In each iteration, do the following:

  First, the spatial filter 201 is applied to the current divided image (kth divided image) of the blue component of the image captured by the camera (step S352). This spatial filter has tap coefficients as shown in FIG. 11, for example. As is apparent from the tap coefficient, this spatial filter has a band-pass characteristic. Therefore, by applying this spatial filter 201 to the image, it is possible to suppress low frequency components and high frequency components resulting from the original image.

  Next, the blocking unit 202 divides the divided image to which the spatial filter is applied into one or more blocks having a size for performing discrete Fourier transform (step S353).

  Next, the discrete Fourier transform unit 203 performs a discrete Fourier transform on the image of each block to obtain a frequency component (step S354).

  Next, the amplitude calculation unit 204 obtains the amplitude of each frequency component of each block (step S355).

  Next, the amplitude integrating unit 205 obtains an integrated amplitude of each frequency component for the divided image by integrating frequency components having the same frequency among the blocks included in the divided image (step S356).

  Next, the rotation angle detection / pattern number identification unit 301 inputs the integrated amplitude of each frequency component and determines the pattern number p by looking at the distribution of non-zero frequency components distributed in the inner circle 602 of FIG. Identify (step S357). That is, referring to FIG. 3, when the center of gravity is in the vicinity of the position of 0 degrees and the position of 180 degrees, it is determined that the pattern number included in the current divided image is 1, and the center of gravity is If it is in the vicinity of the 45 degree position and the 225 degree position, it is determined that the pattern number included in the current divided image is 2, and the center of gravity is in the vicinity of the 90 degree position and the 270 degree position. If it is in the vicinity of the position, it is determined that the pattern number included in the current divided image is 3. If the center of gravity is in the vicinity of the position of 135 degrees and the vicinity of the position of 315 degrees, It is determined that the pattern number included in the divided image is 4.

  Next, the partial data number detection unit 302 is based on the entire correspondence between the number k of the divided image that matches the current repetition number k and the pattern number p detected when the k-th divided image is processed. It is detected what number of partial data corresponds to the divided image being processed (that is, the correspondence between the captured divided image number k and the partial data number j) (step S358). That is, referring to FIG. 3, if the number k of the captured divided image is a quadrant number, when k = 1 is repeated and the correspondence between k and p is (1, 2), If the correspondence between k and j is (1, 1), and the correspondence between k and p is (1, 3), the correspondence between k and j is ( 1 and 4), and when the correspondence between k and p is (1, 4), it is detected that the correspondence between k and j is (1, 3). If the correspondence between k and p is (1, 1), it is detected that the correspondence between k and j is (1, 2). In the repetition of k = 2, when the correspondence between k and p is (2, 3), it is detected that the correspondence between k and j is (2, 2). When the correspondence between p and (2, 4) is detected, the correspondence between k and j is detected as (2, 1), and the correspondence between k and p is (2, 1), it is detected that the correspondence between k and j is (2, 4), and when the correspondence between k and p is (2, 2), k and It is detected that the correspondence relationship with j is (2, 3). In the repetition of k = 3, when the correspondence between k and p is (3, 4), it is detected that the correspondence between k and j is (3, 3). When the correspondence relationship with p is (3, 1), it is detected that the correspondence relationship between k and j is (3, 2), and the correspondence relationship between k and p is (3, 2), it is detected that the correspondence between k and j is (3, 1), and when the correspondence between k and p is (3, 3), k and It is detected that the correspondence with j is (3, 4). In the repetition of k = 4, when the correspondence between k and p is (4, 1), it is detected that the correspondence between k and j is (4, 4). When the correspondence relationship with p is (4, 2), it is detected that the correspondence relationship between k and j is (4, 3), and the correspondence relationship between k and p is (4, 3), it is detected that the correspondence between k and j is (4, 2), and when the correspondence between k and p is (4, 4), k and It is detected that the correspondence relationship with j is (4, 1).

  In the method described with reference to FIG. 10 corresponding to the data division into two, the number of the partial data can be detected immediately only from the center of gravity of the non-zero frequency component for detecting the rotation angle. On the other hand, in the method corresponding to the data division into 4, the number of the partial data cannot be detected only from the center of gravity of the non-zero frequency component for detecting the rotation angle. The number of partial data can be detected from the correspondence between the center of gravity and the quadrant of the currently processed divided image.

  After step S358, the rotation angle of the image is detected based on the digital watermark for detecting the rotation angle (step S359). That is, referring to FIG. 3, since it is detected which of the rotation angle detection digital watermark patterns is one of patterns 1 to 4 in step S357 of the current iteration, the number when the image is not rotated The rotation angle within the section is detected based on the angle of the pattern. That is, the difference between the angle of the numbered pattern and the detected pattern angle when the image is not rotated is detected as a small rotation angle. Further, depending on the correspondence between the number k of the current divided image and the pattern number p, whether the rotation angle of the image is near 0 degrees, near 90 degrees, around 180 degrees, or around 270 degrees And 0 degrees, 90 degrees, 180 degrees, or 270 degrees is detected as a large rotation angle accordingly. A combination of the small rotation angle and the large rotation angle is defined as the rotation angle.

  Next, the partial data detection unit 303 detects partial data from the divided image (step S360). The detection method is the same as the method described with reference to FIG. At this time, the rotation angle detected in step S359 is used as a correction angle in step S921. Further, the number j obtained in step S358 is used as the number of partial data.

  When the repetition of step S351 ends, the data detection unit 304 detects data based on the first to fourth partial data obtained in step S360 in all repetitions (step S362).

  Note that spatial filter application (step S352), blocking (step S353), discrete Fourier transform (step S354), and amplitude calculation (step S355) are not performed within the repetition of step S351, but before the repetition of step S351. Alternatively, it may be performed on the entire image.

  Further, the digital watermark detection apparatus shown in FIG. 14 and the digital watermark detection method described with reference to FIG. 15 can be applied to the pattern of FIG.

  FIG. 16 is a block diagram illustrating a configuration of a digital watermark detection apparatus for performing the detection method (data two-division / rotation angle detection digital watermark two-division / detection four-division) described with reference to FIG.

  Referring to FIG. 16, this digital watermark detection apparatus includes a spatial filter 201, a blocking unit 202, a discrete Fourier transform unit 203, an amplitude calculation unit 204, an amplitude integration unit 205, a rotation angle detection / pattern number identification unit 401, partial data. A number detection unit 402, an amplitude summation unit 403, a partial data detection unit 404, and a data detection unit 405 are provided. These parts may be realized by hardware, but may be realized by a computer reading and executing a program for causing the computer to function as these parts.

  Next, a digital watermark detection method performed by the digital watermark detection apparatus shown in FIG. 16 will be described with reference to FIGS.

  In this digital watermark detection method, first, steps S452 to S458 are repeated for the first to fourth divided images (step S451). In each iteration, do the following:

  First, the spatial filter 201 is applied to the current divided image (kth divided image) of the blue component of the image captured by the camera (step S452). This spatial filter has tap coefficients as shown in FIG. 11, for example. As is apparent from the tap coefficient, this spatial filter has a band-pass characteristic. Therefore, by applying this spatial filter 201 to the image, it is possible to suppress low frequency components and high frequency components resulting from the original image.

  Next, the blocking unit 202 divides the divided image to which the spatial filter is applied into one or more blocks having a size for performing discrete Fourier transform (step S453).

  Next, the discrete Fourier transform unit 203 performs a discrete Fourier transform on the image of each block to obtain a frequency component (step S454).

  Next, the amplitude calculation unit 204 obtains the amplitude of each frequency component of each block (step S455).

  Next, the amplitude integrating unit 205 obtains an integrated amplitude of each frequency component for the divided image by integrating frequency components of the same frequency among the blocks included in the divided image (step S456).

  Next, the rotation angle detection / pattern number identification unit 401 inputs the integrated amplitude of each frequency component and determines the pattern number p by looking at the distribution of non-zero frequency components distributed in the inner circle 602 of FIG. Identify (step S457). That is, referring to FIG. 4, when the center of gravity is in the vicinity of the position of 45 degrees and the position of 225 degrees, it is determined that the pattern number included in the current divided image is 1, and the center of gravity is If it is in the vicinity of the position of 135 degrees and the vicinity of the position of 315 degrees, it is determined that the pattern number included in the current divided image is 2.

  Next, the small rotation angle of the image is detected based on the digital watermark for detecting the rotation angle (step S458). That is, referring to FIG. 4, since it is detected whether the rotation angle detection digital watermark pattern is pattern 1 or 2 in step S 457 of the current repetition, the number when the image is not rotated. The small rotation angle is detected based on the angle of the pattern. That is, the difference between the angle of the numbered pattern and the detected pattern angle when the image is not rotated is detected as a small rotation angle.

  When the repetition of step S451 is completed, the partial data number detection unit 404 counts what number based on the entire correspondence between the number k of the divided image and the pattern number p detected when the k-th divided image is processed. It is detected what number of partial data corresponds to the divided image (that is, the correspondence relationship between the captured divided image number k and the partial data number j) (step S459). That is, referring to FIG. 4, when the pattern number detected in the first quadrant to the pattern number detected in the fourth quadrant is (1, 2, 2, 1), the first divided image and the first The first partial data corresponds to the four divided images, the second partial data corresponds to the second divided image and the third divided image, and is detected in the pattern number to the fourth quadrant detected in the first quadrant. When the obtained pattern number is (2, 2, 1, 1), the first partial data corresponds to the first divided image and the second divided image, and the third divided image and the fourth divided image When the second partial data corresponds to the divided image and the pattern number detected in the first quadrant to the pattern number detected in the fourth quadrant is (2, 1, 1, 1, 2), The first partial data corresponds to the divided image and the third divided image, and the first divided image and the fourth divided image. When the second partial data corresponds to the image and the pattern number detected in the first quadrant to the pattern number detected in the fourth quadrant is (1, 1, 2, 2), the third division The first partial data corresponds to the image and the fourth divided image, and the second partial data corresponds to the first divided image and the second divided image.

  Incidentally, in the method described with reference to FIG. 10 corresponding to the data division into two and corresponding to the variation of the camera angle in units of 180 degrees, the partial data number is immediately obtained from only the centroid of the non-zero frequency component for detecting the rotation angle. In the method described with reference to FIG. 15 that can be detected and corresponds to data division into 4, the correspondence between the center of gravity of the non-zero frequency component for detecting the rotation angle and the quadrant of the currently processed divided image The number of the partial data can be detected from the above, but in the method corresponding to the data division into two and corresponding to the variation of the camera angle in units of 90 degrees, the center of gravity of the non-zero frequency component for detecting the rotation angle and the current processing The number of partial data cannot be detected from the correspondence relationship between the quadrants of the divided images and the non-zero frequency for rotation angle detection for all the divided images. Which portion the data to which the divided image from the whole of the center of gravity of the component can detect whether they correspond.

  Next, based on the entire correspondence between the number k of the divided image and the pattern number p detected when the k-th divided image is processed, the rotation angle of the image is about 0 degrees or about 90 degrees. , Around 180 degrees, or around 270 degrees, and accordingly, 0 degree, 90 degrees, 180 degrees, or 270 degrees is detected as a large rotation angle (step S460).

  Next, a combination of the small rotation angle and the large rotation angle is set as the rotation angle (step S461).

  Following step S461, steps S463 and S464 are repeated for the first partial data to the second partial data (step S462).

  First, the integrated amplitude value for each frequency for all the divided images corresponding to the partial data of the current number is added (step S463). That is, if the current number is 1 and the rotation angle is near 0 degrees, the integrated amplitude values for each frequency for the first quadrant and the fourth quadrant corresponding to the first partial data are added together, And the rotation angle is in the vicinity of 90 degrees, the sum of the integrated amplitude values for each frequency for the first quadrant and the second quadrant corresponding to the first partial data, and the current number is 1 and if the rotation angle is around 180 degrees, the sum of the amplitude integrated values for each frequency for the second quadrant and the third quadrant corresponding to the first partial data, and the current number is 1. If the rotation angle is in the vicinity of 270 degrees, the integrated amplitude values for each frequency for the third and fourth quadrants corresponding to the first partial data are added together. Further, if the current number is 2 and the rotation angle is near 0 degrees, the sum of the amplitude integrated values for each frequency for the second quadrant and the third quadrant corresponding to the second partial data is added. 2 and the rotation angle is near 90 degrees, the sum of the amplitude integrated values for each frequency for the third quadrant and the fourth quadrant corresponding to the second partial data is added, and the current number is 2 and if the rotation angle is around 180 degrees, the sum of the integrated amplitude values for each frequency for the first quadrant and the fourth quadrant corresponding to the second partial data, and the current number is 2. If the rotation angle is in the vicinity of 270 degrees, the integrated amplitude values for each frequency for the first quadrant and the second quadrant corresponding to the second partial data are added together.

  Next, the partial data detection unit 403 detects partial data from the amplitude of the combined frequency (step S464). The detection method is the same as the method described with reference to FIG. At this time, the rotation angle calculated in step S461 is used as a correction angle in step S921. The current repetition number j is used as the partial data number.

  When the repetition of step S462 is completed, the data detection unit 405 detects data based on the first partial data and the second partial data obtained in step S464 in all repetitions (step S465).

  Note that spatial filter application (step S452), blocking (step S453), discrete Fourier transform (step S454), and amplitude calculation (step S455) are performed before the first iteration instead of being performed within the first iteration. Alternatively, it may be performed on the entire image.

  As shown in FIG. 18, the first digital watermark for detecting the rotation angle is inserted into the left image, the second digital watermark for detecting the rotation angle is inserted into the right image, and the first partial data is The digital watermark detection shown in FIG. 16 is also inserted when the second partial data is inserted into the upper left image, the third partial data is inserted into the lower left image, and the fourth partial data is inserted into the lower right screen. The digital watermark detection method described with reference to the apparatus and FIG. 17 can be applied. However, in this case, the number of partial screens corresponding to one partial data is only one in step S461.

  A method of inserting a digital watermark as shown in FIG. 18 will be described with reference to FIG.

  In this digital watermark insertion method, steps S652 to S655 are repeated for the first divided image to the N1th divided image (step S651). In each iteration, do the following:

  First, the rotation angle detection pattern generation unit 101 generates a j1th rotation angle detection pattern (step S652). This pattern is composed of non-zero frequency components.

  Next, the inverse discrete Fourier transform unit 105 obtains a spatial domain pattern by performing inverse discrete Fourier transform on the pattern generated in step S652 (step S653).

  Next, the binarization unit 106 binarizes the space area pattern obtained in step S653 (step S654). That is, for each pixel, if the pixel value of the spatial pattern obtained in step S653 is 0 or more, a predetermined value (for example, 10) is set. If the pixel value of the spatial pattern obtained in step S653 is less than 0, , And a predetermined value (for example, −10).

  Next, the adding unit (subtracting unit) 107 adds or subtracts the spatial region pattern binarized in step S654 to all or a part of blocks of the j-th region of the yellow component of the original image (step S655). Note that the binarization unit 106 performs binarization in order to facilitate detection of the digital watermark by the digital watermark detection apparatus, and may be omitted. In this case, the spatial pattern output from the inverse discrete Fourier transform unit 105 is directly added to or subtracted from the j-th region of the yellow component of the original image without being binarized.

  Next, Steps S657 to S670 are repeated for the first to N-th divided images (Step S151). Here, each divided region when the image is divided into N1 and each divided region when the image is divided into N2 are generally different. In each iteration, do the following:

First, the data pattern generation unit 103 reads the entire data held in the data holding unit 102, obtains the j2th data according to the value of j2 currently handled, and the j2th data according to the j2th data. A data pattern is generated (step S657). This pattern is composed of non-zero frequency components. In the case where the j2nd data pattern is as shown in FIG. 18, the jth data Dj is represented by D as the whole data.
Dj = mod (D, 252 ^ (N + 1−j)) / 252 ^ (N−j)
It is represented by

  Next, the inverse discrete Fourier transform unit 105 obtains a spatial domain pattern by performing inverse discrete Fourier transform on the pattern generated in step S657 (step S658).

  Next, the binarization unit 106 binarizes the space area pattern obtained in step S658 (step S659). That is, for each pixel, if the pixel value of the spatial region pattern obtained in step S658 is 0 or more, a predetermined value (for example, 10) is set. If the pixel value of the spatial pattern obtained in step S154 is less than 0, , And a predetermined value (for example, −10).

  Next, the adding unit (subtracting unit) 107 adds or subtracts the spatial region pattern binarized in step S659 to all or a part of blocks of the j-th region of the yellow component of the original image (step S670). Note that the binarization unit 106 performs binarization in order to facilitate detection of the digital watermark by the digital watermark detection apparatus, and may be omitted. In this case, the spatial pattern output from the inverse discrete Fourier transform unit 105 is directly added to or subtracted from the j-th region of the yellow component of the original image without being binarized.

  Thereafter, the image into which the digital watermark is inserted by the above method is printed on a printed matter.

  Steps S654, S655, S659, and S670 are deleted, and after repeating step S656, the rotation angle detection digital watermark and the data digital watermark are added to the entire image, and the result of the addition is added to the entire image. Or you may subtract. The addition result may be binarized as necessary before addition or subtraction.

  The digital watermark shown in FIG. 18 can also be detected by the method described below.

  Referring to FIG. 20, in this digital watermark detection method, first, steps S752 to S758 are repeated for the first to fourth divided images (step S751). In each iteration, do the following:

  First, the spatial filter 201 is applied to the current divided image (kth divided image) of the blue component of the image captured by the camera (step S752). This spatial filter has tap coefficients as shown in FIG. 11, for example. As is apparent from the tap coefficient, this spatial filter has a band-pass characteristic. Therefore, by applying this spatial filter 201 to the image, it is possible to suppress low frequency components and high frequency components resulting from the original image.

  Next, the blocking unit 202 divides the divided image to which the spatial filter is applied into one or more blocks having a size for performing discrete Fourier transform (step S753).

  Next, the discrete Fourier transform unit 203 performs a discrete Fourier transform on the image of each block to obtain a frequency component (step S754).

  Next, the amplitude calculation unit 204 obtains the amplitude of each frequency component of each block (step S755).

  Next, the amplitude integrating unit 205 obtains an integrated amplitude of each frequency component for the divided image by integrating the frequency components of the same frequency among the blocks included in the divided image (step S756).

  Next, the rotation angle detection / pattern number identification unit 401 inputs the integrated amplitude of each frequency component and determines the pattern number p by looking at the distribution of non-zero frequency components distributed in the inner circle 602 of FIG. Identify (step S457). That is, referring to FIG. 18, when the center of gravity is in the vicinity of the position of 45 degrees and the position of 225 degrees, it is determined that the pattern number included in the current divided image is 1, and the center of gravity is If it is in the vicinity of the position of 135 degrees and the vicinity of the position of 315 degrees, it is determined that the pattern number included in the current divided image is 2.

  Next, the small rotation angle of the image is detected based on the digital watermark for detecting the rotation angle (step S758). That is, referring to FIG. 18, since it is detected whether the rotation angle detection digital watermark pattern is pattern 1 or 2 in the current iteration step S757, the number when the image is not rotated. The small rotation angle is detected based on the angle of the pattern. That is, the difference between the angle of the numbered pattern and the detected pattern angle when the image is not rotated is detected as a small rotation angle.

  When the repetition of step S751 is completed, the partial data number detection unit 404 determines what number based on the entire correspondence between the number k of the divided image and the pattern number p detected when the kth divided image is processed. It is detected what number of partial data corresponds to the divided image (that is, the correspondence relationship between the captured divided image number k and the partial data number j) (step S759). That is, referring to FIG. 4, when the pattern number detected in the first quadrant to the pattern number detected in the fourth quadrant is (1, 2, 2, 1), the first divided image includes 1 partial data corresponds, the second partial data corresponds to the second divided image, the third partial data corresponds to the third divided image, and the fourth partial data corresponds to the fourth divided image. Correspondingly, when the pattern number detected in the first quadrant to the pattern number detected in the fourth quadrant is (2, 2, 1, 1), the fourth partial data is included in the first partial image. The first partial data corresponds to the second partial image, the second partial data corresponds to the third partial image, the third partial data corresponds to the fourth partial image, The pattern number detected in the quadrant to the pattern number detected in the fourth quadrant are (2, 1, 1, 1, 2). The third partial data corresponds to the first divided image, the fourth partial data corresponds to the second divided image, the first partial data corresponds to the third divided image, and If the second partial data corresponds to the four divided images, and the pattern number detected in the first quadrant to the pattern number detected in the fourth quadrant is (1, 1, 2, 2), The fourth partial data corresponds to one divided image, the third partial data corresponds to the second divided image, the second partial data corresponds to the third divided image, and the fourth divided image corresponds to the fourth divided image. The first partial data corresponds.

  Next, based on the entire correspondence between the number k of the divided image and the pattern number p detected when the k-th divided image is processed, the rotation angle of the image is about 0 degrees or about 90 degrees. , Around 180 degrees, or around 270 degrees, and accordingly, 0 degree, 90 degrees, 180 degrees, or 270 degrees is detected as a large rotation angle (step S760).

  Next, a combination of the small rotation angle and the large rotation angle is set as the rotation angle (step S761).

  After step S761, step S763 is repeated for the first to fourth partial data (step S462).

  In step S462, the partial data detection unit 403 detects partial data from the amplitude obtained in step S756. The detection method is the same as the method described with reference to FIG. At this time, the rotation angle calculated in step S761 is used as a correction angle in step S921. The current repetition number j is used as the partial data number.

  If the repetition of step S762 ends, the data detection unit 405 detects data based on the first to fourth partial data obtained in step S763 in all repetitions (step S764).

  Note that the spatial filter application (step S752), blocking (step S753), discrete Fourier transform (step S754), and amplitude calculation (step S755) are performed before entering the first iteration instead of being performed within the first iteration. Alternatively, it may be performed on the entire image.

  FIG. 21 is a block diagram illustrating a configuration of a digital watermark detection apparatus for performing the detection method (the entire image pre-data detection rotation correction) described with reference to FIG.

  Referring to FIG. 21, this digital watermark detection apparatus includes a spatial filter 201, a blocking unit 202, a discrete Fourier transform unit 203, an amplitude calculation unit 204, an amplitude integration unit 205, a rotation angle detection unit 501, a rotation correction unit 503, a space A filter 504, a blocking unit 505, a discrete Fourier transform unit 506, an amplitude calculation unit 507, an amplitude integration unit 508, a partial data detection unit 509, and a data detection unit 510 are provided. These parts may be realized by hardware, but may be realized by a computer reading and executing a program for causing the computer to function as these parts.

  Next, a digital watermark detection method performed by the digital watermark detection apparatus shown in FIG. 21 will be described with reference to FIGS.

  First, the spatial filter 201 is applied to the entire blue component of the image captured by the camera (step S551). This spatial filter has tap coefficients as shown in FIG. 11, for example. As is apparent from the tap coefficient, this spatial filter has a band-pass characteristic. Therefore, by applying this spatial filter 201 to the image, it is possible to suppress low frequency components and high frequency components resulting from the original image.

  Next, the blocking unit 202 divides the entire image to which the spatial filter is applied into one or more blocks having a size for performing discrete Fourier transform (step S552).

  Next, the discrete Fourier transform unit 203 performs a discrete Fourier transform on the image of each block to obtain a frequency component (step S553).

  Next, the amplitude calculation unit 204 obtains the amplitude of each frequency component of each block (step S554).

  Next, the amplitude integrating unit 205 obtains the integrated amplitude of each frequency component for the entire image by integrating the frequency components of the same frequency among the blocks included in the entire image (step S555).

  Next, the rotation angle detection unit 501 receives the integrated amplitude of each frequency component and detects the rotation angle of the entire image by observing the distribution of the non-zero frequency components distributed in the inner circle 602 of FIG. (Step S556). In order to detect the rotation angle, the method described with reference to FIG. 12 can be used.

  Next, the rotation correction unit 503 corrects the rotation angle of the entire image using the rotation angle detected in step S556 (step S557).

  Next, steps S559 to S564 are repeated for the first partial data to the Nth partial data (step S558). In each iteration, do the following:

  First, the spatial filter 504 is applied to the current divided image (jth divided image) of the blue component of the rotation-corrected image (step S559). This spatial filter has tap coefficients as shown in FIG. 11, for example. As is apparent from the tap coefficient, this spatial filter has a band-pass characteristic. Therefore, by applying this spatial filter 201 to the image, it is possible to suppress low frequency components and high frequency components resulting from the original image.

  Next, the blocking unit 505 divides the divided image to which the spatial filter is applied into blocks having a size for performing discrete Fourier transform (step S560).

  Next, the discrete Fourier transform unit 506 performs a discrete Fourier transform on the image of each block to obtain a frequency component (step S561).

  Next, the amplitude calculation unit 507 calculates the amplitude of each frequency component of each block (step S562).

  Next, the amplitude integrating unit 508 obtains the integrated amplitude of each frequency component for the divided image by integrating the frequency components of the same frequency among the blocks included in the divided image (step S563).

  Next, the j-th partial data is detected from the integrated amplitude of each frequency component (step S564). In order to detect the j-th partial data, the detection method described with reference to FIG. 13 can be used. However, since the rotation correction has already been performed in step S557, step S921 is omitted.

  When the repetition of step S558 is completed, the entire data is detected based on the first to Nth partial data detected in step S564 of each repetition (step S565).

  In order to obtain data corresponding to a digital watermark by inserting a digital watermark into an image printed on a printed material, photographing the image with a camera or the like, and detecting the digital watermark from the captured image Can be used.

It is a figure which shows the digital watermark in the case of the data 2 division | segmentation utilized by embodiment of this invention. It is a figure which shows the digital watermark in the printed image and picked-up image in the case of the data 2 division | segmentation and the electronic watermark 2 division | segmentation at the time of a detection, and 2 division | segmentation at the time of detection utilized in embodiment of this invention. FIG. 4 is a diagram illustrating a digital image in a print image and a captured image in the case of data division into four and digital watermark for rotation angle detection and division into four at detection used in the embodiment of the present invention. FIG. 5 is a diagram illustrating a digital image in a print image and a captured image in the case of data division into two and digital watermark for rotation angle detection divided into two and four divisions at the time of detection used in the embodiment of the present invention. It is a figure which shows the digital watermark in the printing image in the case of the whole image pre-data detection rotation correction utilized in embodiment of this invention. It is a block diagram which shows the structure of the digital watermark insertion apparatus by embodiment of this invention. 5 is a flowchart illustrating a digital watermark insertion method according to an embodiment of the present invention. 6 is a flowchart illustrating a digital watermark insertion method according to another embodiment of the present invention. FIG. 2 is a block diagram showing a configuration of a digital watermark detection apparatus according to an embodiment in the case of data division into two and digital watermark division into two for rotation angle detection and division into two at detection according to the present invention; 6 is a flowchart illustrating a digital watermark detection method according to an embodiment of the present invention when data is divided into two parts / digital watermark for rotation angle detection is divided into two parts and divided into two at the time of detection. It is a figure which shows the tap coefficient of the spatial filter utilized in embodiment of this invention. 3 is a flowchart illustrating a method for detecting a rotation angle detection digital watermark according to an embodiment of the present invention; 4 is a flowchart illustrating a method for detecting a digital watermark representing data according to an embodiment of the present invention. FIG. 2 is a block diagram showing a configuration of a digital watermark detection apparatus according to an embodiment in the case of data division into four and digital watermark division into four for rotation angle detection and division into four at detection according to the present invention; 4 is a flowchart illustrating a digital watermark detection method according to an embodiment of the present invention in the case of data division into four / digital watermark for rotation angle detection and division into four at detection. FIG. 3 is a block diagram showing a configuration of a digital watermark detection apparatus according to an embodiment in the case of data division into two and digital watermark division into two for detection of rotation angle and division into four at detection according to the present invention; It is a flowchart which shows the digital watermark detection method by embodiment in the case of the data 2 division | segmentation and the digital watermark 2 division | segmentation at the time of detection of this invention, and 4 division | segmentation at the time of a detection. FIG. 4 is a diagram illustrating a digital image in a print image and a photographed image in the case of data division into four / digital watermark for rotation angle detection according to the present invention and division into two when detection is performed. 10 is a flowchart illustrating a digital watermark insertion method according to still another embodiment of the present invention. 4 is a flowchart illustrating a digital watermark detection method according to an embodiment of the present invention when data is divided into four and a digital watermark for rotation angle detection is divided into two and divided into four at the time of detection. It is a block diagram which shows the structure of the digital watermark detection apparatus by embodiment in the case of the whole image rotation correction before data detection of this invention. It is a flowchart which shows the digital watermark detection method by embodiment in the case of the whole image rotation correction before data detection of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Rotation angle detection pattern generation part 102 Data holding part 103 Data pattern generation part 104 Synthesis | combination part 105 Inverse discrete Fourier transform part 106 Binarization part 107 Addition part (subtraction part)
201 Spatial Filter 202 Blocking Unit 203 Discrete Fourier Transform Unit 204 Amplitude Calculation Unit 205 Amplitude Integration Unit 206 Rotation Angle Detection / Region Number Identification Unit 207 Partial Data Detection Unit 208 Data Detection Unit 301 Rotation Angle Detection / Pattern Number Identification Unit 302 Partial Data Number detection unit 303 Partial data detection unit 304 Data detection unit 401 Rotation angle detection / pattern number identification unit 402 Partial data number detection unit 403 Amplitude summation unit 404 Partial data detection unit 405 Data detection unit 501 Rotation angle detection unit 503 Rotation correction unit 504 Spatial filter 505 Blocking unit 506 Discrete Fourier transform unit 507 Amplitude calculation unit 508 Amplitude integration unit 509 Partial data detection unit 510 Data detection unit

Claims (19)

Of the first predetermined number of candidate positions arranged on a circle in the frequency domain, the positions are combined with each other by the j-th data (j = 1, 2,..., N) to be inserted, and Generating a non-zero frequency component in a second predetermined number less than a predetermined number of 1;
Detecting a rotation angle of the image in the frequency domain and generating a non-zero frequency component for identifying the jth region of the image;
A sub-step of combining the generated non-zero frequency components into a spatial domain from the frequency domain;
A sub-step of adding or subtracting the image obtained by the conversion to all or a part of blocks of the j-th region of the original image;
A method for inserting a digital watermark, comprising: repeating N for N j. A first generation sub-step for detecting a rotation angle of the image in the frequency domain and generating a non-zero frequency component for identifying the j1 (j1 = 1, 2,..., N1) region of the image;
A first transformation substep for transforming the non-zero frequency component generated in the first generation substep from the frequency domain to a spatial domain;
A substep of adding or subtracting the image obtained in the first conversion substep to all or a part of blocks of the j1 region of the original image;
Repeating for N1 j1s;
In the frequency domain, among the first predetermined number of candidate positions arranged on a circle, the positions are combined with each other by j2th data (j2 = 1, 2,..., N2) to be inserted, A second generation sub-step for generating a non-zero frequency component in a second predetermined number less than the first predetermined number;
A second transformation substep for transforming the non-zero frequency component generated in the second generation substep from the frequency domain to a spatial domain;
A sub-step of adding or subtracting the image obtained in the second conversion sub-step to all or a part of the blocks of the j2 region of the original image;
Repeating for N2 j2s;
An electronic watermark insertion method comprising: Of the first predetermined number of candidate positions arranged on a circle in the frequency domain, the positions are combined with each other by the j-th data (j = 1, 2,..., N) to be inserted, and A first generation sub-step for generating non-zero frequency components in a second predetermined number less than a predetermined number of 1;
A first transformation substep for transforming the non-zero frequency component generated in the first generation substep from the frequency domain to a spatial domain;
A sub-step of adding or subtracting the image obtained in the first conversion sub-step to all or a part of blocks of the j-th region of the original image;
Repeating for N j;
A second generation step of generating a non-zero frequency component for detecting an image rotation angle in the frequency domain;
A second transforming step for transforming the non-zero frequency component generated in the second generating step from the frequency domain to a spatial domain;
Adding or subtracting the image obtained in the second conversion step to all or a part of the blocks of the original image;
An electronic watermark insertion method comprising:   A printed matter on which an image generated by the digital watermark insertion method according to claim 1 is printed. A sub-step of obtaining one or more image blocks from the k-th (k = 1, 2,..., N) region of the image;
Sub-step of transforming the data of the one or more image blocks from a spatial domain to a frequency domain;
A sub-step of obtaining an amplitude of each frequency component based on each frequency component of the one or more image blocks;
A non-zero frequency for detecting the rotation angle of the image based on the amplitude of each frequency component in the frequency domain and identifying the jth (j = 1, 2,... N) region of the image. Detecting a rotation angle of the image by detecting a position of a component and identifying a j-th region of the image;
Any one of a second predetermined number smaller than the first predetermined number among the first predetermined number of positions on the circumference in the frequency domain and the rotation angle detected as a rotation reference. A sub-step for detecting for each circle of different radii whether there is a non-zero frequency component at the position of
Detecting a j-th data inserted in the image based on a combination of positions of the non-zero frequency components detected to be at any one of the second predetermined number of positions;
Repeating for N k;
Restoring the data based on the N jth data and the rank j;
An electronic watermark detection method comprising: A sub-step of obtaining one or more image blocks from the k-th (k = 1, 2,..., N) region of the image;
Sub-step of transforming the data of the one or more image blocks from a spatial domain to a frequency domain;
A sub-step of obtaining an amplitude of each frequency component based on each frequency component of the one or more image blocks;
A non-zero frequency for detecting the rotation angle of the image based on the amplitude of each frequency component in the frequency domain and identifying the jth (j = 1, 2,... N) region of the image. A sub-step of detecting the position of the component and detecting the data number j based on the position of the detected non-zero frequency component and the region number k;
A sub-step of detecting a rotation angle of the image based on the detected position of the non-zero frequency component;
Any one of a second predetermined number smaller than the first predetermined number among the first predetermined number of positions on the circumference in the frequency domain and the rotation angle detected as a rotation reference. A sub-step for detecting for each circle of different radii whether there is a non-zero frequency component at the position of
Detecting a j-th data inserted in the image based on a combination of positions of the non-zero frequency components detected to be at any one of the second predetermined number of positions;
Repeating for N k;
Restoring the data based on the N jth data;
An electronic watermark detection method comprising: A sub-step of obtaining one or more image blocks from a k-th (k = 1, 2,..., M × N) region of the image;
Sub-step of transforming the data of the one or more image blocks from a spatial domain to a frequency domain;
A sub-step of obtaining an amplitude of each frequency component based on each frequency component of the one or more image blocks;
A non-zero frequency for detecting the rotation angle of the image based on the amplitude of each frequency component in the frequency domain and identifying the jth (j = 1, 2,... N) region of the image. A sub-step of detecting a small rotation angle of the image by detecting a position of the component;
Repeating for M × N k;
According to the distribution over M × N k of non-zero frequency components for detecting the rotation angle of the image and identifying the j th (j = 1, 2,..., N) region of the image. Obtaining the correspondence between k and k and a large rotation angle;
Obtaining a rotation angle of the image based on the small rotation angle and the large rotation angle;
adding the amplitude of each frequency component between M regions having the same value of j;
For each of the M regions having the same value of j, the rotation angle detected at the first predetermined number of positions on the circumference in the frequency region based on the amplitudes of the frequency components added together. Detecting for each circle of different radii whether there is a non-zero frequency component at any of a second predetermined number less than the first predetermined number of rotation references;
Inserted into the image based on a combination of the positions of the non-zero frequency components detected to be in any one of the second predetermined number of positions for every M areas having the same value of j Detecting the jth data;
Restoring the data based on the N jth data and the rank j;
An electronic watermark detection method comprising: A sub-step of obtaining one or more image blocks from the k-th (k = 1, 2,..., N) region of the image;
Sub-step of transforming the data of the one or more image blocks from a spatial domain to a frequency domain;
A sub-step of obtaining an amplitude of each frequency component based on each frequency component of the one or more image blocks;
In the frequency domain, a rotation angle of the image is detected on the basis of the amplitude of each frequency component, and a non-sigma for identifying the sth (s = 1, 2,... N / M) region of the image. A sub-step of detecting a small rotation angle of the image by detecting a position of a zero frequency component;
Repeating for N k;
K according to the distribution over N k of non-zero frequency components for detecting the rotation angle of the image and identifying the s th (s = 1, 2,..., N) region of the image. Obtaining a correspondence with the data number j and a large rotation angle;
Obtaining a rotation angle of the image based on the small rotation angle and the large rotation angle;
Based on the amplitude of each frequency component, in the frequency domain, a first predetermined number of positions on the circumference, and the rotation angle detected above the first predetermined number out of the rotation angles detected as rotation references. A sub-step for detecting, for each circle of different radii, whether there is a non-zero frequency component at any of a small second predetermined number of positions;
Detecting a j-th data inserted in the image based on a combination of positions of the non-zero frequency components detected to be at any one of the second predetermined number of positions;
Repeating for N k;
Restoring the data based on the N jth data and the rank j;
An electronic watermark detection method comprising: Obtaining one or more image blocks from the image;
Converting the data of the one or more image blocks from a spatial domain to a frequency domain;
Obtaining the amplitude of each frequency component based on each frequency component of the one or more image blocks;
Detecting the rotation angle of the image by detecting the position of a non-zero frequency component for detecting the rotation angle of the image based on the amplitude of each frequency component in the frequency domain; and
Correcting the rotation angle of the image by the rotation angle;
In addition,
A sub-step of obtaining one or more image blocks from the j-th (j = 1, 2,..., N) region of the image whose rotation angle is corrected;
Sub-step of transforming the data of the one or more image blocks from a spatial domain to a frequency domain;
A sub-step of obtaining an amplitude of each frequency component based on each frequency component of the one or more image blocks;
Whether or not there is a non-zero frequency component at any one of the second predetermined number smaller than the first predetermined number of the first predetermined number of positions on the circumference in the frequency domain. A substep to detect for each circle of radius;
Detecting a j-th data inserted in the image based on a combination of positions of the non-zero frequency components detected to be at any one of the second predetermined number of positions;
Repeating for N j;
Restoring the data based on the N jth data and the rank j;
An electronic watermark detection method comprising: Of the first predetermined number of candidate positions arranged on a circle in the frequency domain, the positions are combined with each other by the j-th data (j = 1, 2,..., N) to be inserted, and Generating non-zero frequency components in a second predetermined number less than a predetermined number of ones;
Detecting a rotation angle of the image in the frequency domain and generating a non-zero frequency component for identifying the jth region of the image;
Combining the generated non-zero frequency components from the frequency domain to the spatial domain;
Adding or subtracting the image obtained by the conversion to all or a part of the blocks of the jth region of the original image;
A digital watermark insertion apparatus comprising means for repeating N for j. A first generation step of detecting a rotation angle of the image in the frequency domain and generating a non-zero frequency component for identifying the j1 (j1 = 1, 2,..., N1) region of the image;
A first conversion step of converting the non-zero frequency component generated in the first generation step from the frequency domain to a spatial domain;
Adding or subtracting the image obtained in the first conversion step to all or a part of blocks of the j1 region of the original image;
Means for repeating N1 j1s;
In the frequency domain, among the first predetermined number of candidate positions arranged on a circle, the positions are combined with each other by j2th data (j2 = 1, 2,..., N2) to be inserted, A second generation step of generating non-zero frequency components in a second predetermined number less than the first predetermined number;
A second transformation step for transforming the non-zero frequency component generated in the second generation step from the frequency domain to a spatial domain;
Adding or subtracting the image obtained in the second conversion step to all or a part of the blocks of the j2 region of the original image;
Means for repeating N2 j2s;
An electronic watermark insertion apparatus comprising: Of the first predetermined number of candidate positions arranged on a circle in the frequency domain, the positions are combined with each other by the j-th data (j = 1, 2,..., N) to be inserted, and Generating a non-zero frequency component for a second predetermined number less than a predetermined number of ones;
A transforming step of transforming the non-zero frequency component generated in the generating step from the frequency domain to a spatial domain;
Adding or subtracting the image obtained in the conversion step to all or a part of blocks of the j-th region of the original image;
Means for repeating for N j;
Generating means for generating a non-zero frequency component for detecting the rotation angle of the image in the frequency domain;
Transforming means for transforming the non-zero frequency component generated by the generating means from the frequency domain to a spatial domain;
Means for adding or subtracting the image obtained by the conversion means to all or a part of the blocks of the original image;
An electronic watermark insertion apparatus comprising: Obtaining one or more image blocks from the kth (k = 1, 2,..., N) region of the image;
Converting the data of the one or more image blocks from a spatial domain to a frequency domain;
Obtaining the amplitude of each frequency component based on each frequency component of the one or more image blocks;
A non-zero frequency for detecting the rotation angle of the image based on the amplitude of each frequency component in the frequency domain and identifying the jth (j = 1, 2,... N) region of the image. Detecting a rotation angle of the image by detecting a position of a component and identifying a j-th region of the image;
Any one of a second predetermined number smaller than the first predetermined number among the first predetermined number of positions on the circumference in the frequency domain and the rotation angle detected as a rotation reference. Detecting for each circle of different radii whether there is a non-zero frequency component at the position of
Detecting the jth data inserted in the image based on a combination of the positions of the non-zero frequency components detected to be in any of the second predetermined number of positions;
Means for repeating N for k,
Means for restoring data based on the N jth data and the rank j;
An electronic watermark detection apparatus comprising: Obtaining one or more image blocks from the kth (k = 1, 2,..., N) region of the image;
Converting the data of the one or more image blocks from a spatial domain to a frequency domain;
Obtaining the amplitude of each frequency component based on each frequency component of the one or more image blocks;
A non-zero frequency for detecting the rotation angle of the image based on the amplitude of each frequency component in the frequency domain and identifying the jth (j = 1, 2,... N) region of the image. Detecting the position of the component and detecting the data number j based on the detected position of the non-zero frequency component and the region number k;
Detecting a rotation angle of the image based on the detected position of the non-zero frequency component;
Any one of a second predetermined number smaller than the first predetermined number among the first predetermined number of positions on the circumference in the frequency domain and the rotation angle detected as a rotation reference. Detecting for each circle of different radii whether there is a non-zero frequency component at the position of
Detecting the jth data inserted in the image based on a combination of the positions of the non-zero frequency components detected to be in any of the second predetermined number of positions;
Means for repeating N for k,
Means for restoring data based on N j-th data;
An electronic watermark detection apparatus comprising: Obtaining one or more image blocks from the k-th (k = 1, 2,..., M × N) region of the image;
Converting the data of the one or more image blocks from a spatial domain to a frequency domain;
Obtaining the amplitude of each frequency component based on each frequency component of the one or more image blocks;
A non-zero frequency for detecting the rotation angle of the image based on the amplitude of each frequency component in the frequency domain and identifying the jth (j = 1, 2,... N) region of the image. Detecting a small rotation angle of the image by detecting the position of the component;
Means for repeating M × N k;
According to the distribution over M × N k of non-zero frequency components for detecting the rotation angle of the image and identifying the j th (j = 1, 2,..., N) region of the image. Means for determining the correspondence between k and k and a large rotation angle;
Means for determining a rotation angle of the image based on the small rotation angle and the large rotation angle;
means for adding the amplitude of each frequency component between M regions having the same value of j;
For each of the M regions having the same value of j, the rotation angle detected at the first predetermined number of positions on the circumference in the frequency region based on the amplitudes of the frequency components added together. Means for detecting, for each circle of different radii, whether or not there is a non-zero frequency component at any of a second predetermined number less than the first predetermined number of rotation references;
Inserted into the image based on a combination of the positions of the non-zero frequency components detected to be in any one of the second predetermined number of positions for every M areas having the same value of j Means for detecting the jth data;
Means for restoring data based on the N jth data and the rank j;
An electronic watermark detection apparatus comprising: Obtaining one or more image blocks from the kth (k = 1, 2,..., N) region of the image;
Converting the data of the one or more image blocks from a spatial domain to a frequency domain;
Obtaining the amplitude of each frequency component based on each frequency component of the one or more image blocks;
In the frequency domain, a rotation angle of the image is detected on the basis of the amplitude of each frequency component, and a non-sigma for identifying the sth (s = 1, 2,... N / M) region of the image. Detecting a small rotation angle of the image by detecting a position of a zero frequency component;
Means for repeating N for k,
K according to the distribution over N k of non-zero frequency components for detecting the rotation angle of the image and identifying the s th (s = 1, 2,..., N) region of the image. Means for determining the correspondence with the data number j and the large rotation angle;
Means for determining a rotation angle of the image based on the small rotation angle and the large rotation angle;
Based on the amplitude of each frequency component, in the frequency domain, a first predetermined number of positions on the circumference, and the rotation angle detected above the first predetermined number out of the rotation angles detected as rotation references. Detecting for each circle of different radii whether there is a non-zero frequency component at any of a small second predetermined number of positions;
Detecting the jth data inserted in the image based on a combination of the positions of the non-zero frequency components detected to be in any of the second predetermined number of positions;
Means for repeating N for k,
Means for restoring data based on the N jth data and the rank j;
An electronic watermark detection apparatus comprising: Means for obtaining one or more image blocks from the image;
Means for transforming the data of the one or more image blocks from a spatial domain to a frequency domain;
Means for obtaining the amplitude of each frequency component based on each frequency component of the one or more image blocks;
Means for detecting the rotation angle of the image by detecting the position of a non-zero frequency component for detecting the rotation angle of the image based on the amplitude of each frequency component in the frequency domain;
Means for correcting the rotation angle of the image by the rotation angle;
Obtaining one or more image blocks from the j-th (j = 1, 2,..., N) region of the image whose rotation angle is corrected;
Converting the data of the one or more image blocks from a spatial domain to a frequency domain;
Obtaining the amplitude of each frequency component based on each frequency component of the one or more image blocks;
Whether or not there is a non-zero frequency component at any one of the second predetermined number smaller than the first predetermined number of the first predetermined number of positions on the circumference in the frequency domain. Detecting for each circle of radius;
Detecting the jth data inserted in the image based on a combination of the positions of the non-zero frequency components detected to be in any of the second predetermined number of positions;
Means for repeating for N j;
Means for restoring data based on the N jth data and the rank j;
An electronic watermark detection apparatus comprising:   The program for making a computer perform the electronic watermark insertion method of any one of Claims 1 thru | or 3.   The program for making a computer perform the electronic watermark detection method of any one of Claims 5 thru | or 9.

Images