JP2005252491A - Electronic watermark detecting apparatus and method thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an electronic watermark detecting apparatus for detecting watermark information of an image signal to which the watermark has already been embedded and capable of more accurately detecting the watermark information weakened by attacks such as image segmentation and scaling without being attended with an increase in the computation complexity and the circuit scale when an electronic watermark embedding apparatus converts the phase of a particular frequency component of an embedding object image according to the watermark information to generate the image signal to which the watermark has already been embedded. <P>SOLUTION: The electronic watermark detecting apparatus comprises: an extraction unit 12 for extracting the particular frequency component from the image signal to which the watermark has already been embedded; a first orthogonal transforming unit 11 for calculating the orthogonal transformation image of the extracted extraction signal and the image signal to which the watermark has already been embedded, respectively; a composite unit 13 for compositing the calculated orthogonal transformation image of the extraction signal with the orthogonal transformation image of the image signal to which the watermark has already been embedded; a second orthogonal transforming unit 14 for applying orthogonal transformation or inverse orthogonal transformation to the composed composite signal; and an estimate unit 15 for estimating the watermark information on the basis of a peak appearing in the transformed composite signal. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a digital watermark detection apparatus and method effective for preventing illegal copying of a digital moving image signal provided via a recording medium, for example.

  With the widespread use of apparatuses for recording and reproducing digital image data such as digital VTRs or DVDs (Digital Versatile Discs), a large number of digital moving images that can be reproduced by these apparatuses have been provided. In addition, various digital moving images are distributed through digital television broadcasting via the Internet, broadcasting satellites, communication satellites, etc., and users can use high-quality digital moving images.

  A digital moving image can be easily made a high-quality copy at the digital signal level, and there is a possibility that the copy is unlimited without any copy inhibition or copy control. Therefore, in order to prevent illegal copying (copying) of digital moving images or to control the number of generations of copying by authorized users, information for copying control is added to the digital moving images, and this additional information is used. A method of preventing illegal duplication and limiting duplication is considered.

  As a technique for superimposing other additional information on a digital moving image in this way, digital watermarking is known. Digital watermarks are used for digital data such as audio, music, video, still images, etc., content copyright owner and user identification information, copyright owner rights information, content usage conditions, and usage By embedding secret information that is sometimes necessary or information such as the above-mentioned duplication control information (these are called watermark information) so that it is not easy to perceive, and later detecting the watermark information from the content as necessary This is a technology for copyright protection including use control and copy control, and for promoting secondary use.

  As one method of digital watermarking, a method using spread spectrum technology is known. In this method, watermark information is embedded in a digital moving image by the following procedure.

[Step E1]
Spread spectrum by multiplying image signal with PN (Pseudorandom Noise) sequence.

[Step E2]
The image signal after the spread spectrum is subjected to frequency conversion (for example, DCT conversion).

[Step E3]
The watermark information is embedded by changing the value of a specific frequency component.

[Step E4]
Inverse frequency conversion (for example, IDCT conversion) is performed.

[Step E5]
Spectral despreading is performed (multiplying the same PN sequence as in step E1).

  On the other hand, the watermark information is detected from the digital moving image in which the watermark information is embedded in the following procedure.

[Step D1]
The image signal is multiplied by a PN (Pseudorandom Noise) sequence (the same PN sequence as in step E1) to perform spectrum spreading.

[Step D2]
The image signal after the spread spectrum is subjected to frequency conversion (for example, DCT conversion).

[Step D3]
Focusing on the value of a specific frequency component, the embedded watermark information is extracted.

On the other hand, a technique for extracting a specific frequency component signal from an input image signal, performing phase control on the specific frequency component signal, calculating a correlation value of the phase-controlled signal, and detecting watermark information has been proposed (for example, Patent Document 1 and Patent Document 2).
JP 2002-325233 A (refer to claim 2, FIG. 7) JP 2002-218404 A (refer to claim 1, FIG. 1)

  When digital watermarking is applied to prevent unauthorized use, the watermark information is not lost or altered by various operations or intentional attacks that are normally assumed to be applied to digital works. It is necessary to have unique properties (robustness). Attacks that make it impossible to detect watermark information for a digital image in which watermark information is embedded include image clipping and scaling (enlargement / reduction).

  When an image subjected to such an attack is input, the conventional technique first restores the synchronization of the PN sequence by performing a process of estimating the PN sequence used in step E1 at the time of embedding when detecting the watermark information. After that, the processes of steps D1 to D3 are performed to extract the embedded watermark information. However, in order to recover the synchronization of the PN sequence from only the image signal, it is necessary to perform a search for trying processing with a plurality of candidates and adopting those that have been successfully detected. For this reason, there is a problem that the calculation amount and the circuit scale increase. In addition, there is a problem that watermark information is weak in an attacked image, and it is understood that the image is cut out and scaled, and the watermark information cannot be detected even if detection corresponding to it is performed.

  An object of the present invention is to provide a digital watermark detection apparatus and method capable of more accurately detecting watermark information weakened by attacks such as image cutout and scaling without increasing the amount of calculation and the circuit scale. .

  According to the first aspect of the present invention, when the embedded image signal is generated by converting the phase of the specific frequency component in the embedding target image according to the watermark information by the digital watermark embedding device, the watermark of the embedded image signal is generated. In the digital watermark detection apparatus for detecting information, an extraction means for extracting a specific frequency component from the embedded image signal, and a first orthogonal transform for calculating an orthogonal transform image of the extracted extracted signal and the embedded image signal, respectively Means for synthesizing the orthogonal transformation image of the calculated extracted signal and the orthogonal transformation image of the embedded image signal, and a second orthogonal transformation for performing orthogonal transformation or inverse orthogonal transformation on the synthesized signal. And the watermark information based on the peak appearing in the composite signal transformed by the second orthogonal transform means. And estimating means for estimating the a digital watermark detection apparatus characterized by comprising a.

  According to the second aspect of the present invention, when the embedded image signal is generated by converting the phase of the specific frequency component in the embedding target image according to the watermark information by the digital watermark embedding device, the watermark of the embedded image signal is generated. In the digital watermark detection apparatus for detecting information, a first orthogonal transform unit that calculates an orthogonal transform image of the embedded image signal, an extraction unit that extracts a specific frequency component from the calculated orthogonal transform image, and the extracted First orthogonal transforming means for respectively calculating the orthogonal transformation image of the extracted signal and the embedded image signal, and combining means for synthesizing the orthogonal transformation image of the calculated extraction signal and the orthogonal transformation image of the embedded image signal A second orthogonal transform means for performing orthogonal transform or inverse orthogonal transform on the synthesized composite signal, and the second orthogonal transform A digital watermark detection apparatus characterized by comprising an estimation means for estimating the watermark information based on the peak appearing in the converted composite signal by stage.

  The invention according to claim 3 includes first accumulating means for accumulating the embedded image signal to generate a first accumulative signal, and the first accumulating means is the extracting means or the first orthogonal transforming means. The digital watermark detection apparatus according to claim 1, wherein the digital watermark detection apparatus outputs the digital watermark.

  The invention according to claim 4 further includes a second accumulating unit that is arranged at a subsequent stage of the second orthogonal transform unit and accumulates the second orthogonal transform signal to generate a second accumulated signal, and the estimation unit. 3. The digital watermark detection apparatus according to claim 1, wherein the watermark information is estimated from a peak appearing in a second accumulated signal output from the second accumulation unit.

  The invention according to claim 5 is arranged downstream of the second orthogonal transform means, normalizing means for generating a normalized correlation signal by normalizing the second orthogonal transform signal with amplitude, and outputting from the normalization means And a second accumulator for accumulating the normalized normalized correlation signal to generate a second accumulative signal, wherein the estimating means estimates the watermark information from a peak appearing in the second accumulative signal. 3. The digital watermark detection apparatus according to claim 1, wherein the digital watermark detection apparatus is a digital watermark detection apparatus.

  The invention according to claim 6 includes a dividing unit that divides the embedded image signal into at least two image signals, and outputs the divided image signal to the extracting unit or the first orthogonal transform unit, respectively. 3. The digital watermark detection apparatus according to claim 1, wherein the digital watermark detection apparatus is a digital watermark detection apparatus.

  The invention according to claim 7 includes a dividing unit that divides the embedded image signal into at least two images, and a first accumulating unit that accumulates the divided image signals to generate accumulated signals, respectively. 3. The digital watermark detection apparatus according to claim 1, wherein a first accumulated signal of each of the divided images is output to the extraction unit or the first orthogonal transform unit.

  The invention according to claim 8 is characterized in that the estimation means estimates the watermark information by performing level determination based on a threshold value that is changed according to an accumulation period of the second accumulation signal by the second accumulation means. A digital watermark detection apparatus according to claim 4 or 5.

  The invention according to claim 9 is characterized in that the estimation means estimates the watermark information by at least two estimation methods, checks whether or not the detection results by the at least two detection methods match, and if they match, the watermark 8. The digital watermark detection apparatus according to claim 1, wherein it is determined that there is information, and if there is no match, it is determined that there is no watermark information.

  The invention according to claim 10 is characterized in that the estimation means estimates the watermark information by determining a polarity of the peak. 9. The digital watermark detection according to at least one of claims 1 to 7, Device.

  According to an eleventh aspect of the present invention, there is provided thinning means that uses a signal obtained by thinning out pixels of the embedded image signal as a new embedded image signal, and the thinned embedded image signal is used as the extracting means or the first orthogonal signal. 3. The digital watermark detection apparatus according to claim 1, wherein the digital watermark detection apparatus outputs the converted data to a conversion unit.

  The invention according to claim 12 is the digital watermark detection apparatus according to claim 1, wherein the synthesizing unit synthesizes by compressing an amplitude of the orthogonal transform image.

  The invention according to claim 13 is the digital watermark detection apparatus according to claim 1 or 2, wherein the orthogonal transform in the first orthogonal transform unit or the second orthogonal transform unit is a Fourier transform.

  According to the fourteenth aspect of the present invention, when the embedded image signal is generated by converting the phase of the specific frequency component in the embedding target image according to the watermark information by the digital watermark embedding device, the watermark of the embedded image signal is generated. In the digital watermark detection method for detecting information, an extraction step for extracting a specific frequency component from the embedded image signal, and a first orthogonal transform for respectively calculating an orthogonal transform image of the extracted signal and the embedded image signal A synthesis step of synthesizing the orthogonal transformation image of the calculated extracted signal and the orthogonal transformation image of the embedded image signal, and a second orthogonal transformation for performing orthogonal transformation or inverse orthogonal transformation on the synthesized signal And the watermark information based on peaks appearing in the converted combined signal. A digital watermark detection method characterized by comprising the estimation step constant to, a.

  The invention according to claim 15 provides the watermark of the embedded image signal when the digital watermark embedding device generates the embedded image signal by converting the phase of the specific frequency component in the embedding target image according to the watermark information. In the digital watermark detection method for detecting information, a first orthogonal transform step of calculating an orthogonal transform image of the embedded image signal, an extraction step of extracting a specific frequency component from the calculated orthogonal transform image, and the extraction A first orthogonal transform step for calculating an orthogonal transform image of the extracted signal and the embedded image signal, and a combining step for combining the orthogonal transform image of the calculated extracted signal and the orthogonal transform image of the embedded image signal And a second orthogonal transformation step for performing orthogonal transformation or inverse orthogonal transformation on the synthesized signal. When a digital watermark detection method characterized by comprising, an estimation step of estimating the watermark information based on the peak appearing in the converted composite signal.

  According to the sixteenth aspect of the present invention, when the embedded image signal is generated by converting the phase of the specific frequency component in the embedding target image according to the watermark information by the digital watermark embedding device, the watermark of the embedded image signal is generated. In a program for realizing a digital watermark detection method for detecting information by a computer, an extraction function for extracting a specific frequency component from the embedded image signal, and an orthogonal transform image of the extracted signal and the embedded image signal A first orthogonal transform function for respectively calculating the orthogonal transform image of the calculated extracted signal and an orthogonal transform image of the embedded image signal, and orthogonal transform or inverse to the synthesized composite signal A second orthogonal transform function for performing orthogonal transform and a pin appearing in the transformed composite signal; A program of the digital watermark detection method characterized by realizing an estimation function for estimating the watermark information based on the click.

  According to the seventeenth aspect of the present invention, when the embedded image signal is generated by converting the phase of the specific frequency component in the embedding target image according to the watermark information by the digital watermark embedding device, the watermark of the embedded image signal is generated. In a program for realizing a digital watermark detection method for detecting information by a computer, a first orthogonal transform function for calculating an orthogonal transform image of the embedded image signal and extracting a specific frequency component from the calculated orthogonal transform image A first orthogonal transform function for calculating an orthogonal transformation image of the extracted extracted signal and the embedded image signal, respectively, and an orthogonal transformation image of the calculated extracted signal and the orthogonality of the embedded image signal. A synthesis function for synthesizing the transformed image, and orthogonal transformation or inversion to the synthesized signal. A second orthogonal transformation function for converting a program of the digital watermark detection method characterized by realizing an estimation function for estimating the watermark information based on the peak appearing in the converted composite signal.

  According to the present invention, it is possible to accurately detect watermark information without an increase in calculation amount or circuit scale against attacks such as image clipping and scaling.

(Basic configuration of digital watermark embedding device)
First, a digital watermark embedding device will be described.

  FIG. 27 is a block diagram showing a basic configuration of the digital watermark embedding apparatus.

  The digital watermark embedding device receives a digitized image signal of a moving image or a still image as an image signal (hereinafter referred to as “embedding target image signal”) 110 into which watermark information is to be embedded. The embedding target image signal 110 may include both a luminance signal and a color difference signal, but may include only a luminance signal. The embedding target image signal 110 is branched into three and input to the specific frequency component extraction unit 111, the feature amount extraction unit 115, and the watermark information superimposition unit 116.

  The specific frequency component extraction unit 111 includes a frequency domain digital filter, for example, a high-pass filter having a predetermined cut-off frequency, or a band-pass filter having a predetermined passband center frequency. Extract components, eg, relatively high frequency components. Hereinafter, the output signal from the specific frequency component extraction unit 111 is referred to as a “specific frequency component signal”.

  The phase and amplitude of the specific frequency component signal output from the specific frequency component extraction unit 111 are converted by the phase converter 112 and the amplitude converter 113. In the present embodiment, the phase converter 112 is disposed at the front stage and the amplitude converter 113 is disposed at the rear stage. Conversely, the amplitude converter 113 may be disposed at the front stage and the phase converter 112 may be disposed at the rear stage. The watermark information 114 which is digital information to be embedded in the input image signal 110 is given to at least one of the phase converter 112 and the amplitude converter 113.

  The phase converter 112 is configured to perform phase conversion of a specific phase conversion amount determined in advance on the specific frequency component signal. Specifically, the phase converter 112 is realized by a single or a plurality of digital phase shifters, and the phase conversion amount is the phase shift amount of the phase shifter. FIG. 28 is a diagram showing the state of the phase shift by the phase converter 112. In this example, the specific frequency component signal is simply phase-shifted while maintaining the waveform. When the watermark information 114 is input to the phase converter 112, the phase conversion amount (phase shift amount) of the phase converter 112 is controlled according to the watermark information 114.

  The amplitude converter 113 is configured to perform amplitude conversion of a predetermined specific amplitude conversion amount on the input specific frequency component signal. Specifically, the amplitude converter 113 is a single or a plurality of exclusive OR circuits or digital multipliers, and the amplitude conversion amount is a coefficient to be multiplied by the input specific frequency component signal. When the watermark information 114 is input to the amplitude converter 113, the amplitude conversion amount (coefficient) of the amplitude converter 113 is controlled according to the watermark information 114.

  Further, the feature amount extraction unit 115 extracts a feature amount of the embedding target image signal 110, for example, an activity representing the complexity of the image. Information on the feature amount is input to the amplitude converter 113. In the amplitude converter 113, the amplitude conversion amount (coefficient) of the specific frequency component is controlled according to the input feature amount. Specifically, when the feature amount is activity, the coefficient is set to be larger as the activity is larger. The feature extraction unit 115 is not essential and may be omitted.

  The specific frequency component signal that has undergone phase conversion and amplitude conversion by the phase converter 112 and the amplitude converter 113 is supplied as an embedded signal by the watermark information superimposing unit 116 formed of a digital adder and is superimposed on the embedding target image signal 110. . That is, the specific frequency component signal extracted by the specific frequency component extraction unit 111 is subjected to phase conversion and amplitude conversion unique to the digital watermark embedding apparatus by the phase converter 112 and the amplitude converter 113, and the phase conversion amount and amplitude conversion. Since one or both of the quantities are controlled by the watermark information 114, the watermark information superimposing unit 116 embeds the watermark information 114 in the embedding target image signal 110.

  The specific frequency component signal extracted by the specific frequency component extraction unit 111 and converted in phase and amplitude by the phase converter 112 and the amplitude converter 113 may exist in a plurality of channels. The specific frequency component signal is superimposed on the embedding target image signal 110 by the watermark information superimposing unit 116.

  The image signal 100 in which the watermark information is embedded (hereinafter referred to as “embedded image signal”) 100 is recorded on a recording medium by a digital image recording / reproducing apparatus such as a DVD system, or the Internet, broadcast satellite, communication satellite. Or the like via a transmission medium such as

  Hereinafter, in the case of detecting the embedded image signal 100, in particular, the digital watermark detection apparatus for the embedded image signal 100 in which the phase conversion amount (phase shift amount) of the specific frequency component signal is controlled according to the watermark information 114. Each embodiment will be described in order with reference to the drawings.

(First embodiment)
The digital watermark detection apparatus according to the first embodiment will be described with reference to FIGS. 1, 8, 15, and 16.

(1) Configuration of Digital Watermark Detection Device FIG. 1 shows the configuration of a digital watermark detection device according to this embodiment.

  1 includes an embedded image signal 100 in which the phase conversion amount (phase shift amount) of the specific frequency component signal is controlled according to the watermark information 114 by the above-described digital watermark embedding device. Input via a transmission medium. It is assumed that a digital signal “1” or “0” is embedded as watermark information.

  Only a specific component is extracted from the embedded image signal 100 by the extractor 12. This extractor is a digital filter in the same frequency region as the specific frequency component extractor used in the above-described digital watermark embedding device, for example, an HPF (high pass filter) having a predetermined cutoff frequency, or a predetermined passband center frequency. A BPF (band pass filter) having a specific frequency component, for example, a signal having a relatively high frequency component, is extracted from the input moving image signal 100.

  The signal extracted by the extractor 12 is subjected to orthogonal transform processing such as Fourier transform by the first orthogonal transformer 11.

  The signal after the orthogonal transformation process is complex-synthesized by the synthesizer 13. Here, complex synthesis is performed using the orthogonal transformation image of the extracted signal and the orthogonal transformation image of the input image.

  The second orthogonal transformer 14 performs orthogonal transformation (inverse orthogonal transformation) on the composite signal after complex synthesis. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. .

  Note that the signal after the complex synthesis may be subjected to the second orthogonal transformation after being subjected to the amplitude compression process. As an amplitude compression method, the amplitude is fixed to 1 and the exponential logarithm of the amplitude is used. A method is conceivable.

  The combined signal after the second orthogonal transformation is given to the input of the estimator 15. A method in which the estimator 15 estimates the watermark information from the combined signal after the conversion will be described with reference to FIGS. It is assumed that “1” or “0” of the digital signal is embedded as watermark information as described above.

  As shown in FIG. 15, the phase of the combined signal after conversion is phase-shifted, and the correlation with the converted combined signal that has not been phase-shifted is obtained. FIG. 16 shows the relationship between the cross correlation value and the phase shift amount.

  As shown in FIG. 16, when a change in the cross-correlation value is observed, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents watermark information. That is, when the embedded image signal 100 is subjected to a scaling attack, the phase shift amount of the specific frequency component signal becomes a value different from the phase shift amount given to the specific frequency component signal in the digital watermark embedding apparatus. Here, the extractor may extract all frequency components.

  Therefore, in the present embodiment, the phase shift amount is controlled continuously or stepwise by the estimator 15, the peak of the cross-correlation value output is searched for, and the watermark information is estimated from the polarity of the searched peak. The peak of the cross-correlation value detected in this way takes either a positive or negative value depending on the value of the watermark information. For example, in the example of FIG. 16, the watermark information is “1” if it is positive, The watermark information is determined to be “0”. In this way, the watermark information 101 detected by the watermark information estimator 15 is also output for the image subjected to the scaling attack.

  As described above, according to the present embodiment, the specific frequency component signal is extracted from the embedded image signal, and the watermark information is detected by performing the phase-only correlation between the specific frequency component signal and the embedded image signal. In this case, by performing correlation calculation while changing the phase, the peak of the correlation value can be searched, so that the watermark information can be easily detected from the embedded image signal subjected to the scaling attack.

(2) Phase Only Correlation Method The above correlation method is called a “phase only correlation method”, and the method will be described with reference to FIG. Phase only correlation: POC (Phase Only Correlation) is a method for calculating the correlation (similarity) between the input image 101 to be collated with the original registered image 50.

  First, the registered image 50 converted into a digital signal is mathematically processed by Fourier transformation to be decomposed into an amplitude 52 (grayscale data) and a phase 53 (image contour data). Further, the input image 101 converted into a digital signal is mathematically processed by Fourier transform to be decomposed into an amplitude 52 (grayscale data) and a phase 53 (image contour data).

  Second, the phase 53 of the decomposed registered image 50 is amplitude-compressed. This is for the purpose of collating with the phase 53 of the input image 101, that is, for the algorithm that image-processes the correlation using only the phase information without using the amplitude information that does not include the shape information. is there. As an amplitude compression method, for example, the amplitude is fixed to 1. Similarly, the phase 53 of the decomposed input image 101 is also subjected to amplitude compression.

  Third, a composite image 54 is created from two pieces of phase information of the registered image 50 and the input image 101, and a correlation image 55 is obtained by performing inverse Fourier transform on the composite image.

  The POC is completely different from the conventional two-dimensional correlation method and feature extraction method using amplitude information, and has a feature that it is resistant to disturbance and has no large error.

(3) Computational amount in orthogonal transformation Here, the computational amount in orthogonal transformation when FFT (Fast Fourier Transform) is used for orthogonal transformation of this embodiment is demonstrated. Assume that the input image has N rows and M columns.

The calculation amount of the two-dimensional FFT is as follows.

In this embodiment, since three orthogonal transformations (first orthogonal transformation and second orthogonal transformation of two images) are used, the following calculation amount is required.

(4) Procedure of Unauthorized Copy Prevention Method Next, the procedure of the unauthorized copy prevention method according to the present embodiment will be described using the flowchart shown in FIG.

  Only the specific component of the embedded image signal is extracted in the extraction step S32. This extraction step is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted in the extraction step S31 is subjected to orthogonal transformation processing such as Fourier transformation in the first orthogonal transformation step S32.

  The signal after the orthogonal transformation process is complex-synthesized by the synthesis step S33. Here, complex synthesis is performed between the extracted signals.

  Orthogonal transformation (inverse orthogonal transformation) is performed on the signal after complex synthesis in the second orthogonal transformation step S34. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. . Note that the second orthogonal transform may be performed after the amplitude compression processing is performed on the signal after the complex synthesis.

  The data after the second orthogonal transformation is given to the input of the estimation step S35. In the watermark information estimation step S35, the watermark information is estimated and detected by searching for the peak of the cross-correlation value while shifting the phase as shown in FIGS. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information.

(Second Embodiment)
Next, a digital watermark detection apparatus according to the second embodiment of the present invention will be described with reference to FIG. 2, FIG. 9, and FIG.

(1) Configuration of Digital Watermark Detection Device FIG. 2 shows a configuration of a digital watermark detection device according to this embodiment.

  In the digital watermark detection apparatus shown in FIG. 2, the phase conversion amount (phase shift amount) of the specific frequency component signal is controlled according to the watermark information 114 by the digital watermark embedding apparatus as in the digital watermark detection apparatus shown in FIG. The embedded image signal 100 is input via a recording medium or a transmission medium.

  The embedded image signal 100 is subjected to orthogonal transform processing such as Fourier transform by the first orthogonal transformer 11.

  Only a specific component is extracted from the signal after the orthogonal transformation process by the extractor 12. This extractor is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding device described above, for example, an HPF (high pass filter) having a predetermined cutoff frequency, or a predetermined pass band. A BPF (band pass filter) having a center frequency, which extracts a signal having a specific frequency component, for example, a relatively high frequency component, from the input moving image signal 100.

  The signal extracted by the extractor 12 is complex-synthesized by the combiner 13. Here, complex synthesis is performed using the orthogonal transformation image of the extracted signal and the orthogonal transformation image of the input image.

  The second orthogonal transformer 14 performs orthogonal transformation (inverse orthogonal transformation) on the signal after complex synthesis. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. .

  Note that the signal after the complex synthesis may be subjected to the second orthogonal transformation after being subjected to the amplitude compression process. As an amplitude compression method, the amplitude is fixed to 1 and the exponential logarithm of the amplitude is used. A method is conceivable.

  The data after the second orthogonal transformation is given to the input of the estimator 15. The watermark information estimator 15 estimates and detects watermark information by searching for the peak of the cross-correlation value while shifting the phase. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information. In this way, the watermark information 101 detected by the watermark information estimator 15 is also output for the image subjected to the scaling attack. Here, the extractor may extract all frequency components.

  As described above, according to the present embodiment, the specific frequency component signal is extracted after performing the orthogonal transformation on the embedded image signal, and the phase-only correlation between the specific frequency component signal and the orthogonal transformation image of the embedded image signal is performed. Detect watermark information. Since the extraction process and the Odaka conversion process are linear processes, the processing result is not changed even if the order is changed, and the processing amount can be reduced by using a processing order with a small amount of calculation.

(2) Procedure of Unauthorized Copy Prevention Method Next, the procedure of the unauthorized copy prevention method according to the present embodiment will be described using the flowchart shown in FIG.

  The embedded image signal is subjected to orthogonal transform processing such as Fourier transform in the first orthogonal transform step S31.

  Only a specific component is extracted from the signal after the orthogonal transformation process in the extraction step S32. This extraction step is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted by the extraction step S32 is complex-synthesized by the synthesis step S33. Here, complex synthesis is performed between the orthogonally transformed signals.

  Orthogonal transformation (inverse orthogonal transformation) is performed on the signal after complex synthesis in the second orthogonal transformation step S34. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. . Note that the second orthogonal transform may be performed after the amplitude compression processing is performed on the signal after the complex synthesis.

  The data after the second orthogonal transformation is given to the input of the estimation step S35. In the watermark information estimation step S35, the watermark information is estimated and detected by searching for the peak of the cross-correlation value while shifting the phase. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information. As shown in FIG. 12, it is also conceivable that the input image signal is processed in the division step S30 so that the image signal divided into at least two is used as the input of the subsequent stage (here, the first orthogonal transformation step S31).

(Third embodiment)
Next, a digital watermark detection apparatus according to the third embodiment of the present invention will be described with reference to FIGS. 3, 10, and 24 to 26.

(1) Configuration of Digital Watermark Detection Device FIG. 3 shows the configuration of the digital watermark detection device according to this embodiment.

  In the digital watermark detection apparatus of FIG. 3, as with the digital watermark detection apparatus of FIG. 1, the phase conversion amount (phase shift amount) of the specific frequency component signal is controlled according to the watermark information 114 by the digital watermark embedding apparatus. An image signal 100 is input via a recording medium or a transmission medium.

  Only a specific component is extracted from the embedded image signal 100 by the extractor 12. This extractor is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted by the extractor 12 is subjected to orthogonal transform processing such as Fourier transform by the first orthogonal transformer 11.

  The signal after the orthogonal transformation process is complex-synthesized by the synthesizer 13. Here, complex synthesis is performed using the orthogonal transformation image of the extracted signal and the orthogonal transformation image of the input image.

  The second orthogonal transformer 14 performs orthogonal transformation (inverse orthogonal transformation) on the signal after complex synthesis. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. .

  Note that the signal after the complex synthesis may be subjected to the second orthogonal transformation after being subjected to the amplitude compression process. As an amplitude compression method, the amplitude is fixed to 1 and the exponential logarithm of the amplitude is used. A method is conceivable.

  The data after the second orthogonal transformation is given to the input of the second accumulator 18. The second accumulator 18 accumulates the input signal over a specific accumulation period and outputs a second accumulated signal. The accumulation period is selected, for example, 15 seconds, 30 seconds, and 1 minute. The second accumulator 18 is reset when it accumulates over an accumulation period and outputs a second accumulation signal.

  The second accumulated signal is provided to the input of the estimator 15. The watermark information estimator 15 estimates and detects watermark information by searching for the peak of the cross-correlation value while shifting the phase. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information. In this way, the watermark information 101 detected by the watermark information estimator 15 is also output for the image subjected to the scaling attack. Here, the extractor may extract all frequency components.

  As described above, according to the present embodiment, it is considered that a robust correlation result is calculated in the input image by accumulating the correlation signal, and a long and high detection rate is realized in the input image by extending the accumulation period. It becomes possible to do.

(2) Accumulation pattern of second accumulator As an accumulation pattern of the second accumulator, for example, when adding in all fields (or frames) as shown in FIG. 24, as shown in FIG. 25 and FIG. There may be a case where addition / subtraction is performed periodically by interweaving addition / subtraction for each field (or for each frame). This accumulated pattern is determined so as to increase the correlation of the watermark signal with the input signal in relation to the embedding pattern on the digital watermark signal embedding side.

(3) Procedure of Unauthorized Copy Prevention Method Next, the procedure of the unauthorized copy prevention method according to the present embodiment will be described using the flowchart shown in FIG.

  The embedded image signal is subjected to orthogonal transform processing such as Fourier transform in the first orthogonal transform step S31.

  Only a specific component is extracted from the signal after the orthogonal transformation process in the extraction step S32. This extraction step 32 is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted by the extraction step S32 is complex-synthesized by the synthesis step S33. Here, complex synthesis is performed between the orthogonally transformed signals.

  Orthogonal transformation (inverse orthogonal transformation) is performed on the signal after complex synthesis in the second orthogonal transformation step S34. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. . Note that the second orthogonal transform may be performed after the amplitude compression processing is performed on the signal after the complex synthesis.

  The data after the second orthogonal transformation is given to the input of the second accumulation step S38. In the second accumulation step S38, the input signal is accumulated over the specific accumulation period that is the accumulation period of the first accumulation step S37, and the second accumulation signal is output. The second accumulation step S38 is reset when it accumulates over the accumulation period and outputs the second accumulation signal.

  The second accumulated signal is given to the input of the estimation step S35. In the watermark information estimation step S35, the watermark information is estimated and detected by searching for the peak of the cross-correlation value while shifting the phase. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information.

(Fourth embodiment)
Next, a digital watermark detection apparatus according to the fourth embodiment of the present invention will be described with reference to FIGS.

(1) Configuration of Digital Watermark Detection Device FIG. 4 shows the configuration of the digital watermark detection device according to this embodiment.

  In the digital watermark detection apparatus of FIG. 4, as with the digital watermark detection apparatus of FIG. 1, the phase conversion amount (phase shift amount) of the specific frequency component signal is controlled according to the watermark information 114 by the digital watermark embedding apparatus. An image signal 100 is input via a recording medium or a transmission medium.

  Only a specific component is extracted from the embedded image signal 100 by the extractor 12. This extractor is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted by the extractor 12 is subjected to orthogonal transform processing such as Fourier transform by the first orthogonal transformer 11.

  The signal after the orthogonal transformation process is complex-synthesized by the synthesizer 13. Here, complex synthesis is performed using the orthogonal transformation image of the extracted signal and the orthogonal transformation image of the input image.

  The second orthogonal transformer 14 performs orthogonal transformation (inverse orthogonal transformation) on the signal after complex synthesis. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. . Note that the signal after the complex synthesis may be subjected to the second orthogonal transformation after being subjected to the amplitude compression process. As an amplitude compression method, the amplitude is fixed to 1 and the exponential logarithm of the amplitude is used. A method is conceivable.

  The data after the second orthogonal transformation is input to the normalizer 19. The signal normalized by the normalizer 19 is given to the input of the second accumulator 18.

  The second accumulator 18 accumulates the input signal over a specific accumulation period and outputs a second accumulated signal. The accumulation period is selected, for example, 15 seconds, 30 seconds, and 1 minute. The second accumulator 18 is reset when it accumulates over an accumulation period and outputs a second accumulation signal.

  The second accumulated signal is provided to the input of the estimator 15. The watermark information estimator 15 estimates and detects watermark information by searching for the peak of the cross-correlation value while shifting the phase. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information. In this way, the watermark information 101 detected by the watermark information estimator 15 is also output for the image subjected to the scaling attack. Here, the extractor may extract all frequency components.

  As described above, according to the present embodiment, it is considered that a correlation result that is robust to the input image is calculated by normalizing the correlation signal and then accumulated. By increasing the accumulation period, the input image is robust. A high detection rate can be realized.

(2) Procedure of Unauthorized Copy Prevention Method Next, the procedure of the unauthorized copy prevention method according to the present embodiment will be described using the flowchart shown in FIG.

  Only the specific component is extracted from the embedded image signal in the extraction step S32. This extraction step is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted in the extraction step S32 is subjected to orthogonal transformation processing such as Fourier transformation in the first orthogonal transformation step S31.

  Only a specific component is extracted from the signal after the orthogonal transformation process in the extraction step S32. This extraction step is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted by the extraction step S32 is complex-synthesized by the synthesis step S33. Here, complex synthesis is performed between the extracted signals.

  Orthogonal transformation (inverse orthogonal transformation) is performed on the signal after complex synthesis in the second orthogonal transformation step S34. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. . Note that the second orthogonal transform may be performed after the amplitude compression processing is performed on the signal after the complex synthesis.

  The data after the second orthogonal transformation is input to the normalization step S39.

  The signal normalized by the normalization step S39 is given to the input of the second accumulation step S38. The second accumulation step S38 accumulates the input signal over a second period longer than the first period, which is the accumulation period of the first accumulation step S37, and outputs a second accumulation signal. The second accumulating step S38 is reset when accumulating over the second period and outputting the second accumulating signal. The second accumulated signal is given to the input of the estimation step S35. In the watermark information estimation step S35, the watermark information is estimated and detected by searching for the peak of the cross-correlation value while shifting the phase. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information.

(Fifth embodiment)
Next, a digital watermark detection apparatus according to the fifth embodiment of the present invention will be described with reference to FIGS. 5, 12, and 23 to 26.

(1) Configuration of Digital Watermark Detection Device FIG. 5 shows a configuration of a digital watermark detection device according to this embodiment.

  In the digital watermark detection apparatus of FIG. 5, similarly to the digital watermark detection apparatus of FIG. 1, the phase conversion amount (phase shift amount) of the specific frequency component signal is controlled according to the watermark information 114 by the digital watermark embedding apparatus. An image signal 100 is input via a recording medium or a transmission medium.

  The embedded image signal 100 is input to the first accumulator 17. The first accumulator 17 accumulates the input signal over a short period in which the characteristics of the image such as several lines, one field, several fields, one frame, several frames, etc. do not change greatly, and outputs a first accumulated signal. The first accumulator 17 is reset when it accumulates over the accumulation period and outputs the first accumulation signal.

  The signal accumulated in the first accumulator 17 is extracted by the extractor 12 only with a specific component. This extractor is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted by the extractor 12 is subjected to orthogonal transform processing such as Fourier transform by the first orthogonal transformer 11.

  The signal after the orthogonal transformation process is complex-synthesized by the synthesizer 13. Here, complex synthesis is performed using the orthogonal transformation image of the extracted signal and the orthogonal transformation image of the input image.

  The second orthogonal transformer 14 performs orthogonal transformation (inverse orthogonal transformation) on the signal after complex synthesis. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. . Note that the signal after the complex synthesis may be subjected to the second orthogonal transformation after being subjected to the amplitude compression process. As an amplitude compression method, the amplitude is fixed to 1 and the exponential logarithm of the amplitude is used. A method is conceivable.

  The data after the second orthogonal transformation is given to the input of the estimator 15. The watermark information estimator 15 estimates and detects watermark information by searching for the peak of the cross-correlation value while shifting the phase. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information. In this way, the watermark information 101 detected by the watermark information estimator 15 is also output for the image subjected to the scaling attack. Here, the extractor may extract all frequency components.

  As described above, according to the present embodiment, by accumulating the input images, for example, by adjusting the embedding side and the detection side so that the input image and the embedded signal (watermark signal) can be separated, It is considered that a robust correlation result is calculated, and it is possible to achieve a high and high detection rate in the input image.

(2) Input Image Accumulation Method Here, an example of the input image accumulation method will be described with reference to FIG.

  Assuming that the input image is composed of an embedded image 60 and an embedded non-image 61, the difference between the embedded image 60 and the embedded non-image 61 is obtained from the embedded signal alone when the input images are accumulated. An embedded image 62 is calculated. By using this method with the embedded image 62 as an input, it is considered that a robust correlation result is calculated for the input image, and a robust and high detection rate can be realized for the input image.

(3) Accumulation pattern of the first accumulator As an accumulation pattern of the first accumulator, for example, when adding in all fields (or frames) as shown in FIG. 24, as shown in FIG. 25 and FIG. There may be a case where addition / subtraction is performed periodically by interweaving addition / subtraction for each field (or for each frame). This accumulated pattern is determined so as to increase the correlation of the watermark signal with the input signal in relation to the embedding pattern on the digital watermark signal embedding side.

(4) Procedure of Unauthorized Copy Prevention Method Next, the procedure of the unauthorized copy prevention method according to the present embodiment will be described using the flowchart shown in FIG.

  Only the specific component is extracted from the embedded image signal in the extraction step S32. This extraction step is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted in the extraction step S32 is subjected to orthogonal transformation processing such as Fourier transformation in the first orthogonal transformation step S31.

  The signal after the orthogonal transformation process is complex-synthesized by the synthesis step S33. Here, complex synthesis is performed between the extracted signals.

  Orthogonal transformation (inverse orthogonal transformation) is performed on the signal after complex synthesis in the second orthogonal transformation step S34. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. . Note that the second orthogonal transform may be performed after the amplitude compression processing is performed on the signal after the complex synthesis.

  The data after the second orthogonal transformation is given to the input of the estimation step S35. In the watermark information estimation step S35, the watermark information is estimated and detected by searching for the peak of the cross-correlation value while shifting the phase. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information.

(Sixth embodiment)
Next, a digital watermark detection apparatus according to the sixth embodiment of the present invention will be described using FIG. 6, FIG. 13, and FIG.

(1) Configuration of Digital Watermark Detection Device FIG. 6 shows a configuration of a digital watermark detection device according to this embodiment.

  In the digital watermark detection apparatus of FIG. 6, as with the digital watermark detection apparatus of FIG. 1, the phase conversion amount (phase shift amount) of the specific frequency component signal is controlled according to the watermark information 114 by the digital watermark embedding apparatus. An image signal 100 is input via a recording medium or a transmission medium.

  The embedded image signal 100 is input to the divider 19. The divider 19 divides the input image into at least two images, and the divided image is extracted only by a specific component by the extractor 12. This extractor is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted by the extractor 12 is subjected to orthogonal transform processing such as Fourier transform by the first orthogonal transformer 11.

  The signal after the orthogonal transformation process is complex-synthesized by the synthesizer 13. Here, complex synthesis is performed using the orthogonal transformation image of the extracted signal and the orthogonal transformation image of the input image.

  The second orthogonal transformer 14 performs orthogonal transformation (inverse orthogonal transformation) on the signal after complex synthesis. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. . Note that the signal after the complex synthesis may be subjected to the second orthogonal transformation after being subjected to the amplitude compression process. As an amplitude compression method, the amplitude is fixed to 1 and the exponential logarithm of the amplitude is used. A method is conceivable.

  The data after the second orthogonal transformation is given to the input of the estimator 15. The watermark information estimator 15 estimates and detects watermark information by searching for the peak of the cross-correlation value while shifting the phase. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information.

  In this way, the watermark information 101 detected by the watermark information estimator 15 is also output for the image subjected to the scaling attack. Here, the extractor may extract all frequency components.

  As described above, according to the present embodiment, it is considered that the amount of processing such as two-dimensional orthogonal transformation can be reduced by dividing the input image. Therefore, it is possible to reduce the calculation amount of the phase-only correlation. In the case of dividing for each row, the number of divisions can be considered up to the number of rows of the image, and when one-dimensional processing (one-dimensional phase-only correlation) is accumulated for the number of rows by dividing each row. Conceivable.

  Similarly, a method such as division for each column is also conceivable. In the case of division for each column, the maximum number of divisions is considered up to the number of columns of the image, and one-dimensional processing is performed by dividing each column. A case where (one-dimensional phase-only correlation) is accumulated for the number of columns is also conceivable.

(2) Flow of the phase only correlation method The flow of the phase only correlation method when the input image is divided will be described with reference to FIG.

  Here, considering the case where the image is divided into two, first, the input image 101 is divided into two divided images 56. Then, the phase only correlation process is performed on each divided image 56. In order to calculate the correlation (similarity) between the input image 56 to be collated with the original registered image 50, the digital signal image is mathematically processed by Fourier transform to obtain the amplitude 52 (grayscale data). Decompose into phase 53 (image contour data).

  As an amplitude compression method, if the amplitude is fixed to 1, the correlation information is image-processed using only the phase information without using the amplitude information that does not include the shape information among the two pieces of information (two phase information). A composite image 54 is created from the image, and a correlation image 55 is obtained by performing inverse Fourier transform on the composite image). A correlation image for the input image 101 is calculated by accumulating the two obtained correlation images 55.

(3) Computational amount in orthogonal transformation Here, the computational amount in orthogonal transformation when FFT (Fast Fourier Transform) is used for orthogonal transformation of this embodiment is demonstrated. It is assumed that the input image has N rows and M columns, and the amount of calculation when one screen is divided (N rows are divided into Q and P rows) is as follows.

In this equation, if one screen is divided into N and one-dimensional processing is performed for the number of rows (Q = 128, P = 1 (N = 128), M = 512), the amount of calculation is as follows: become.

  When one screen is not divided (Q = 1, P = 128 (N = 128), M = 512), the calculation amount is as follows.

In this way, the amount of calculation can be reduced by increasing the number of divisions.

(4) Procedure of Unauthorized Copy Prevention Method Next, the procedure of the unauthorized copy prevention method according to the present embodiment will be described using the flowchart shown in FIG.

  The embedded image signal is divided into at least two images by the dividing step S30.

  In the division step S30, the image signal is divided into at least two images, and only the specific component is extracted from the divided image signal in the extraction step S32. This extraction step is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted in the extraction step S32 is subjected to orthogonal transformation processing such as Fourier transformation in the first orthogonal transformation step S31.

  The signal after the orthogonal transformation process is complex-synthesized by the synthesis step S33. Here, complex synthesis is performed between the extracted signals.

  Orthogonal transformation (inverse orthogonal transformation) is performed on the signal after complex synthesis in the second orthogonal transformation step S34. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. . Note that the second orthogonal transform may be performed after the amplitude compression processing is performed on the signal after the complex synthesis.

  The data after the second orthogonal transformation is given to the input of the estimation step S35. In the watermark information estimation step S35, the watermark information is estimated and detected by searching for the peak of the cross-correlation value while shifting the phase. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information.

(Seventh embodiment)
Next, a digital watermark detection apparatus according to the seventh embodiment of the present invention will be described with reference to FIGS.

(1) Configuration of Digital Watermark Detection Device FIG. 7 shows the configuration of a digital watermark detection device according to this embodiment.

  In the digital watermark detection apparatus of FIG. 7, as with the digital watermark detection apparatus of FIG. 1, the phase conversion amount (phase shift amount) of the specific frequency component signal is controlled according to the watermark information 114 by the digital watermark embedding apparatus. An image signal 100 is input via a recording medium or a transmission medium.

  The embedded image signal 100 is input to the divider 19. The divider 19 divides the input image into at least two images, and the divided images are input to the first accumulator 17.

  The first accumulator 17 accumulates the input signal over a short specific period in which the characteristics of the image such as several lines, one field, several fields, one frame, several frames, etc. do not change greatly, and outputs the first accumulated signal. . The first accumulator 17 is reset when it accumulates over the accumulation period and outputs the first accumulation signal.

  The signal accumulated in the first accumulator 17 is extracted by the extractor 12 only with a specific component. This extractor is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted by the extractor 12 is subjected to orthogonal transform processing such as Fourier transform by the first orthogonal transformer 11.

  The signal after the orthogonal transformation process is complex-synthesized by the synthesizer 13. Here, complex synthesis is performed using the orthogonal transformation image of the extracted signal and the orthogonal transformation image of the input image. The second orthogonal transformer 14 performs orthogonal transformation (inverse orthogonal transformation) on the signal after complex synthesis. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. . Note that the signal after the complex synthesis may be subjected to the second orthogonal transformation after being subjected to the amplitude compression process. As an amplitude compression method, the amplitude is fixed to 1 and the exponential logarithm of the amplitude is used. A method is conceivable.

  The data after the second orthogonal transformation is given to the input of the estimator 15. The watermark information estimator 15 estimates and detects watermark information by searching for the peak of the cross-correlation value while shifting the phase. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information. In this way, the watermark information 101 detected by the watermark information estimator 15 is also output for the image subjected to the scaling attack.

  Here, the extractor may extract all frequency components, and when the orthogonal transformation / synthesis process performs two-dimensional processing (such as two-dimensional Fourier transform) on the two-dimensional image, the two-dimensional image is extracted. A case where one-dimensional processing (one-dimensional Fourier transform) is accumulated by the number of rows by dividing every row is also conceivable.

  As described above, according to the present embodiment, it is considered that the calculation amount of the phase-only correlation can be reduced by dividing the input image, and the embedding side / detection so as to separate the input image and the embedded signal (watermark signal). By adjusting and accumulating the sides, it is considered that a robust correlation result is calculated for the input image, and a robust and high detection rate can be realized for the input image.

  Next, the procedure of the unauthorized copy prevention method according to the present embodiment will be described using the flowchart shown in FIG.

  The embedded image signal is divided into at least two images by the dividing step S30. In the division step S30, the image signal is divided into at least two images, and the divided image signal is input to the first accumulation step S37.

  In the first accumulation step S37, for example, the input signal is accumulated over a short specific period in which the characteristics of the image such as several lines, one field, several fields, one frame, several frames, etc. do not change greatly, and the first accumulated signal is output. . The first accumulating step S37 is reset when accumulating over a specific period and outputting the first accumulating signal.

  Only the specific components of the signal accumulated in the first accumulation step S37 are extracted in the extraction step S32. This extraction step is a digital filter in the same frequency domain as the specific frequency component extractor used in the paired digital watermark embedding apparatus described above.

  The signal extracted in the extraction step S32 is subjected to orthogonal transformation processing such as Fourier transformation in the first orthogonal transformation step S31.

  The signal after the orthogonal transformation process is complex-synthesized by the synthesis step S33. Here, complex synthesis is performed between the extracted signals.

  Orthogonal transformation (inverse orthogonal transformation) is performed on the signal after complex synthesis in the second orthogonal transformation step S34. The orthogonal transformation here needs to be paired with the transformation in the first orthogonal transformation, and when the Fourier transformation is used in the first orthogonal transformation, the second orthogonal transformation performs Fourier transformation or inverse Fourier transformation. . Note that the second orthogonal transform may be performed after the amplitude compression processing is performed on the signal after the complex synthesis.

  The data after the second orthogonal transformation is given to the input of the estimation step S35. In the watermark information estimation step S35, the watermark information is estimated and detected by searching for the peak of the cross-correlation value while shifting the phase. When the change of the correlation value is seen, a peak appears at a position of a certain phase shift amount, and the polarity of this peak represents the watermark information.

(Specific example 1 of watermark information estimator)
Next, a specific example of the watermark signal estimator 15 in the digital watermark detection apparatus will be described with reference to FIGS.

  As shown in FIG. 17, the estimator 15 includes a threshold setting unit 31, a watermark detector 32, and a watermark determination unit 33 in this example.

  The threshold setting unit 31 obtains information on the accumulation period from the second accumulator 18 in the previous stage, changes the threshold for determination of detection of watermark information according to the accumulation period, and gives the watermark determination unit 33 with the threshold for determination of watermark information as shown in FIG. . On the other hand, the watermark detector 32 receives the accumulated signal from the second accumulator 18 in the previous stage, detects watermark information, and outputs the watermark information and level (absolute value of the peak amplitude of the accumulated signal) to the watermark determiner 33. To do.

  The watermark determination unit 33 compares the threshold value provided from the threshold setting unit 31 with the level provided from the watermark detector 32. That is, if the level is equal to or higher than the threshold value, the watermark determination unit 33 determines that the watermark information has been detected, and outputs the watermark information input from the watermark detector 32. On the other hand, if the level is less than the threshold value, the watermark determination unit 33 determines that the watermark information is not embedded, and outputs information “no watermark”. The threshold value is basically set such that the threshold value decreases as the cumulative period increases, but the reverse may be possible. The determination by the watermark determination unit 33 may be performed at a predetermined interval (for example, 15 seconds, 30 seconds, 1 minute, etc.) with a threshold corresponding to the period, or continuously changing while accumulating. You may carry out with the threshold to do.

  As described above, in this embodiment, the probability that watermark information can be detected is increased by lowering the determination threshold value for detecting watermark information when the accumulation period is lengthened. Therefore, the calculation amount and circuit scale required for watermark information detection are increased. It is possible to improve detection performance without increasing.

(Specific example 2 of watermark information estimator)
Next, another specific example of the watermark signal estimator 15 in the digital watermark detection apparatus will be described with reference to FIG.

  As shown in FIG. 19, the estimator 15 includes at least two watermark detectors 41A and 41B having different watermark detection methods and a watermark determiner 42. The watermark detectors 41A and 41B each independently detect watermark information. The watermark determination unit 42 determines whether the detection results of the watermark detectors 41A and 41B match.

  The watermark detector 41 </ b> A receives the signal accumulated by the second accumulator 18, detects watermark information using the first detection method, and outputs it to the watermark determiner 42. Similarly, the watermark detector 41B detects watermark information using the second detection method and outputs the watermark information to the watermark determiner 42. The watermark determination unit 42 compares whether the watermark information from the two watermark detectors 41A and 41B match. If they match, the watermark determination unit 42 determines that the digital watermark has been detected and outputs the watermark information as it is. On the other hand, if they do not match, it is determined that the digital watermark is not embedded, and information “no watermark” is output.

  For example, if “A” is detected by the first detection method by the watermark detector 41A and “A” is also detected by the second detection method by the watermark detector 41B, the two detection results match. Finally, watermark information “A” is detected. On the other hand, if “B” is detected by the first detection method and “C” is detected by the second detection method, the final detection of the watermark information cannot be performed because the two detection results are different. Is determined not to be embedded. Even when there are three or more detection methods, the same idea as in this embodiment can be applied.

  As described above, according to the present embodiment, by comparing the detection results of watermark information by a plurality of detection methods, it is possible to accurately detect watermark information and to reduce the probability of erroneous detection.

(Specific example of decimation of input image)
Next, a specific example of thinning out an input image in the digital watermark detection apparatus will be described with reference to FIG.

  The input image 101 in this example is an example in the case where the thinned image 57 thinned out for each column (shaded area in FIG. 22) is used as a new input image in the subsequent stage in the first processing. Although it is considered that the accuracy of the calculated correlation coefficient is reduced by thinning out images in this way, it is sufficient for detecting a digital watermark, and the amount of calculation can be effectively reduced. In this example, thinning is performed for each column. However, various thinning methods such as thinning for each row, thinning for several columns, thinning for several rows, and the like can be considered.

  The present invention is not limited to the above-described embodiment, and can be embodied by modifying the components without departing from the scope of the invention in the implementation stage. In addition, various inventions can be formed by appropriately combining a plurality of components disclosed in the embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, constituent elements over different embodiments may be appropriately combined.

  The present invention is suitable for an apparatus for recording and reproducing digital image data such as a digital VTR or a DVD.

It is a block diagram which shows the structure of the digital watermark detection apparatus which concerns on the 1st Embodiment of this invention. It is a block diagram which shows the structure of the digital watermark detection apparatus which concerns on 2nd Embodiment. It is a block diagram which shows the structure of the digital watermark detection apparatus which concerns on 3rd Embodiment. It is a block diagram which shows the structure of the digital watermark detection apparatus which concerns on 4th Embodiment. It is a block diagram which shows the structure of the digital watermark detection apparatus which concerns on 5th Embodiment. It is a block diagram which shows the structure of the digital watermark detection apparatus which concerns on 6th Embodiment. It is a block diagram which shows the structure of the digital watermark detection apparatus which concerns on 7th Embodiment. It is a flowchart which shows the procedure of the digital watermark detection method which concerns on 1st Embodiment. It is a flowchart which shows the procedure of the digital watermark detection method which concerns on 2nd Embodiment. It is a flowchart which shows the procedure of the digital watermark detection method which concerns on 3rd Embodiment. It is a flowchart which shows the procedure of the digital watermark detection method which concerns on 4th Embodiment. It is a flowchart which shows the procedure of the digital watermark detection method which concerns on 5th Embodiment. It is a flowchart which shows the procedure of the digital watermark detection method which concerns on 6th Embodiment. It is a flowchart which shows the procedure of the digital watermark detection method which concerns on 7th Embodiment. It is a figure explaining the phase shift of a specific frequency component signal. It is a figure which shows the peak search of the correlation value in a digital watermark detection apparatus, and the example of watermark information detection. It is a block diagram which shows the 1st specific example of the watermark estimator contained in a digital watermark detection apparatus. It is a block diagram which shows the 2nd specific example of the watermark estimator contained in a digital watermark detection apparatus. It is a block diagram which shows the 3rd specific example of the watermark estimator contained in a digital watermark detection apparatus. It is a figure for demonstrating the concept of a phase only correlation method. It is a figure for demonstrating the concept of the division | segmentation of an input image. It is a figure for demonstrating the concept of the thinning-out of an input image. It is a figure for demonstrating the concept of accumulation of an input image. It is a figure of the 1st example of the accumulation pattern of the 2nd accumulator. It is a figure of the 2nd example of the accumulation pattern of a 2nd accumulator. It is a figure of the 3rd example of the accumulation pattern of a 2nd accumulator. It is a block diagram which shows the basic composition of a digital watermark embedding apparatus. It is explanatory drawing about the phase shift of the specific frequency component signal by a phase converter.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Divider 11 1st orthogonal transformer 12 Extractor 13 Synthesizer 14 2nd orthogonal transformer 15 Estimator 100 Image signal with embedded watermark signal 101 Watermark signal

Claims (17)

An electronic watermark detection apparatus for detecting the watermark information of the embedded image signal when the phase of a specific frequency component in the image to be embedded is converted according to the watermark information and the embedded image signal is generated by the electronic watermark embedding apparatus In
Extracting means for extracting a specific frequency component from the embedded image signal;
First orthogonal transform means for respectively calculating orthogonal transform images of the extracted extracted signal and the embedded image signal;
Combining means for combining the orthogonal transformation image of the calculated extracted signal and the orthogonal transformation image of the embedded image signal;
Second orthogonal transformation means for performing orthogonal transformation or inverse orthogonal transformation on the synthesized synthesized signal;
Estimating means for estimating the watermark information based on a peak appearing in the combined signal transformed by the second orthogonal transform means;
An electronic watermark detection apparatus comprising: An electronic watermark detection apparatus for detecting the watermark information of the embedded image signal when the phase of a specific frequency component in the image to be embedded is converted according to the watermark information and the embedded image signal is generated by the electronic watermark embedding apparatus In
First orthogonal transform means for calculating an orthogonal transform image of the embedded image signal;
Extraction means for extracting a specific frequency component from the calculated orthogonal transform image;
First orthogonal transform means for respectively calculating orthogonal transform images of the extracted extracted signal and the embedded image signal;
Combining means for combining the orthogonal transformation image of the calculated extracted signal and the orthogonal transformation image of the embedded image signal;
Second orthogonal transformation means for performing orthogonal transformation or inverse orthogonal transformation on the synthesized synthesized signal;
Estimating means for estimating the watermark information based on a peak appearing in the combined signal transformed by the second orthogonal transform means;
An electronic watermark detection apparatus comprising: First accumulating means for accumulating the embedded image signal to generate a first accumulated signal;
3. The digital watermark detection apparatus according to claim 1, wherein the first accumulated signal is output to the extraction unit or the first orthogonal transform unit. Further comprising second accumulating means arranged at a subsequent stage of the second orthogonal transform means and accumulating the second orthogonal transform signal to generate a second accumulated signal;
The digital watermark detection apparatus according to claim 1, wherein the estimation unit estimates the watermark information from a peak appearing in a second accumulated signal output from the second accumulation unit. A normalizing unit disposed at a subsequent stage of the second orthogonal transform unit, and normalizing the second orthogonal transform signal with an amplitude to generate a normalized correlation signal;
Second accumulating means for accumulating the normalized correlation signal output from the normalizing means to generate a second accumulated signal;
The digital watermark detection apparatus according to claim 1, wherein the estimation unit estimates the watermark information from a peak appearing in the second accumulated signal. Dividing means for dividing the embedded image signal into at least two image signals;
The digital watermark detection apparatus according to claim 1 or 2, wherein the divided image signal is output to the extraction unit or the first orthogonal transform unit, respectively. Dividing means for dividing the embedded image signal into at least two images;
First accumulating means for accumulating the divided image signals to generate accumulated signals, respectively;
The digital watermark detection apparatus according to claim 1 or 2, wherein a first accumulated signal of each of the divided images is output to the extraction unit or the first orthogonal transform unit. 6. The electronic device according to claim 4, wherein the estimation unit estimates the watermark information by performing level determination based on a threshold value that is changed according to an accumulation period of the second accumulation signal by the second accumulation unit. Watermark detection device. The estimation means estimates the watermark information by at least two estimation methods;
The detection result by the at least two detection methods is checked for coincidence or disagreement, and if they coincide, it is determined that the watermark information is present, and if they do not coincide, it is determined that the watermark information is absent. 7. The digital watermark detection apparatus according to at least one of items 7. The digital watermark detection apparatus according to claim 1, wherein the estimation unit estimates the watermark information by determining a polarity of the peak. A thinning means for setting a signal obtained by thinning out pixels of the embedded image signal as a new embedded image signal;
The digital watermark detection apparatus according to claim 1 or 2, wherein the thinned embedded image signal is output to the extraction unit or the first orthogonal transform unit. The digital watermark detection apparatus according to claim 1, wherein the synthesizing unit compresses and synthesizes the amplitude of the orthogonal transform image. The digital watermark detection apparatus according to claim 1, wherein the orthogonal transform in the first orthogonal transform unit or the second orthogonal transform unit is a Fourier transform. A digital watermark detection method for detecting the watermark information of the embedded image signal when the phase of a specific frequency component in the image to be embedded is converted according to the watermark information and the embedded image signal is generated by the digital watermark embedding device In
An extraction step of extracting a specific frequency component from the embedded image signal;
A first orthogonal transform step for calculating an orthogonal transform image of each of the extracted extracted signal and the embedded image signal;
A synthesis step of synthesizing the orthogonal transformation image of the calculated extracted signal and the orthogonal transformation image of the embedded image signal;
A second orthogonal transform step for performing orthogonal transform or inverse orthogonal transform on the synthesized composite signal;
An estimation step of estimating the watermark information based on a peak appearing in the converted combined signal;
An electronic watermark detection method comprising: A digital watermark detection method for detecting the watermark information of the embedded image signal when the phase of a specific frequency component in the image to be embedded is converted according to the watermark information and the embedded image signal is generated by the digital watermark embedding device In
A first orthogonal transform step of calculating an orthogonal transform image of the embedded image signal;
An extraction step of extracting a specific frequency component from the calculated orthogonal transform image;
A first orthogonal transform step for calculating an orthogonal transform image of each of the extracted extracted signal and the embedded image signal;
A synthesis step of synthesizing the orthogonal transformation image of the calculated extracted signal and the orthogonal transformation image of the embedded image signal;
A second orthogonal transform step for performing orthogonal transform or inverse orthogonal transform on the synthesized composite signal;
An estimation step of estimating the watermark information based on a peak appearing in the converted combined signal;
An electronic watermark detection method comprising: A digital watermark detection method for detecting the watermark information of the embedded image signal when the phase of a specific frequency component in the image to be embedded is converted according to the watermark information and the embedded image signal is generated by the digital watermark embedding device In a program for realizing by a computer,
An extraction function for extracting a specific frequency component from the embedded image signal;
A first orthogonal transform function for respectively calculating an orthogonal transform image of the extracted extracted signal and the embedded image signal;
A synthesis function for synthesizing the orthogonal transformation image of the calculated extracted signal and the orthogonal transformation image of the embedded image signal;
A second orthogonal transform function for performing orthogonal transform or inverse orthogonal transform on the synthesized composite signal;
An estimation function for estimating the watermark information based on a peak appearing in the converted combined signal;
An electronic watermark detection method program characterized by realizing the above. A digital watermark detection method for detecting the watermark information of the embedded image signal when the phase of a specific frequency component in the image to be embedded is converted according to the watermark information and the embedded image signal is generated by the digital watermark embedding device In a program for realizing by a computer,
A first orthogonal transform function for calculating an orthogonal transform image of the embedded image signal;
An extraction function for extracting a specific frequency component from the calculated orthogonal transform image;
A first orthogonal transform function for respectively calculating an orthogonal transform image of the extracted extracted signal and the embedded image signal;
A synthesis function for synthesizing the orthogonal transformation image of the calculated extracted signal and the orthogonal transformation image of the embedded image signal;
A second orthogonal transform function for performing orthogonal transform or inverse orthogonal transform on the synthesized composite signal;
An estimation function for estimating the watermark information based on a peak appearing in the converted combined signal;
An electronic watermark detection method program characterized by realizing the above.

Images