JP3950126B2 - Digital watermark detection method and apparatus - Google Patents

Description

  The present invention relates to a digital watermark detection method and apparatus effective for preventing illegal duplication 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 necessary at times, 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 watermark information from the content as necessary This technology protects copyrights including usage control and copy control, and promotes secondary usage.

  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] The image signal is multiplied by a PN (Pseudorandom Noise) sequence to perform spread spectrum.

  [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] Perform spectrum despreading (multiply 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.

  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, scaling (enlargement / reduction), rotation, and the like.

  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 that attempts to process with a plurality of candidates and employs those that have been successfully detected. There is a problem of increasing.

  It is an object of the present invention to provide a digital watermark detection method and apparatus capable of detecting embedded watermark information without increasing the amount of calculation and the circuit scale against attacks such as image clipping, scaling, and rotation. To do.

  In order to solve the above problems, the present invention extracts a specific frequency component signal from an input image signal in an electronic watermark detection method for detecting watermark information from an image signal in which the watermark information is embedded. The phase conversion of the component signal is performed by at least one conversion means, the cross-correlation operation is performed between the specific frequency component signal subjected to the phase conversion and the input image signal, and the mutual change with respect to the change in the phase conversion amount of the phase conversion is performed. The watermark information is extracted from the polarity of the peak of the correlation value change.

  An electronic watermark detection apparatus according to the present invention includes an extraction unit that extracts a specific frequency component signal from an input image signal, at least one conversion unit that performs phase conversion of the extracted specific frequency component signal, and the conversion unit. Performs cross-correlation operation between the specific frequency component signal that has undergone phase conversion and the input image signal, and extracts the watermark information from the polarity of the peak of the cross-correlation value change with respect to the change in the phase conversion amount of the phase conversion Correlation calculating means.

  The digital watermark detection apparatus may further include an amplitude limiter that limits the amplitude of the specific frequency signal inserted between the extraction unit and the correlation calculation unit. Further, at least one characteristic of the extraction unit and the conversion unit may be randomized by randomization information. Furthermore, you may have further the nonlinear filter inserted between the said conversion means and the said correlation calculating means.

  According to the present invention, it is possible to provide a digital watermark detection method and apparatus capable of detecting embedded watermark information without increasing the amount of computation or the circuit scale against attacks such as image clipping, scaling, and rotation. .

  Embodiments of the present invention will be described below with reference to the drawings.

[First Embodiment]
(Basic configuration of digital watermark embedding device)
FIG. 1 is a block diagram showing a basic configuration of a digital watermark embedding apparatus according to an embodiment of the present invention. A digital image signal of a moving image or a still image is input to the digital watermark embedding device as an image signal 10 to be embedded with watermark information (referred to as an embedding target image signal) 10. The embedding target image signal 10 may include both a luminance signal and a color difference signal, but may include only a luminance signal.

  The embedding target image signal 10 is branched into three and input to the specific frequency component extracting unit 11, the feature amount extracting unit 15, and the watermark information superimposing unit 16. The specific frequency component extraction unit 11 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 11 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 11 are converted by the phase converter 12 and the amplitude converter 13. In the present embodiment, the phase converter 12 is arranged at the front stage and the amplitude converter 13 is arranged at the rear stage. Conversely, the amplitude converter 13 may be arranged at the front stage and the phase converter 12 may be arranged at the rear stage. The watermark information 14 which is digital information to be embedded in the input image signal 10 is given to at least one of the phase converter 12 and the amplitude converter 13.

  The phase converter 12 is configured to perform phase conversion of a predetermined specific phase conversion amount on the specific frequency component signal. Specifically, the phase converter 12 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. 2 is a diagram showing the state of the phase shift by the phase converter 12. In this example, the specific frequency component signal is simply phase-shifted while maintaining the waveform. When the watermark information 14 is input to the phase converter 12, the phase conversion amount (phase shift amount) of the phase converter 12 is controlled according to the watermark information 14.

  The amplitude converter 13 is configured to perform amplitude conversion of a predetermined specific amplitude conversion amount on the input specific frequency component signal. Specifically, the amplitude converter 13 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 14 is input to the amplitude converter 13, the amplitude conversion amount (coefficient) of the amplitude converter 13 is controlled according to the watermark information 14.

  Further, in the present embodiment, the feature amount extraction unit 15 extracts a feature amount of the embedding target image signal 10, for example, an activity representing the complexity of the image. Information on the feature amount is input to the amplitude converter 13. In the amplitude converter 13, 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 15 is not essential and may be omitted.

  The specific frequency component signal subjected to the phase conversion and the amplitude conversion by the phase converter 12 and the amplitude converter 13 is supplied as an embedded signal by the watermark information superimposing unit 16 composed of a digital adder and is superimposed on the embedding target image signal 10. . That is, the specific frequency component signal extracted by the specific frequency component extraction unit 11 is subjected to phase conversion and amplitude conversion specific to the digital watermark embedding device by the phase converter 12 and the amplitude converter 13, and the phase conversion amount and amplitude conversion. Since one or both of the amounts are controlled by the watermark information 14, the watermark information superimposing unit 16 embeds the watermark information 14 in the embedding target image signal 10. The specific frequency component signal extracted by the specific frequency component extraction unit 11 and converted in phase and amplitude by the phase converter 12 and the amplitude converter 13 may exist in a plurality of channels. The specific frequency component signal is superimposed on the embedding target image signal 10 in the watermark information superimposing unit 16.

  The image signal (embedded image signal) 17 in which the watermark information is embedded in this way is recorded on a recording medium by a digital image recording / reproducing apparatus such as a DVD system, or a transmission medium such as the Internet, a broadcasting satellite, and a communication satellite. Is transmitted through.

(Basic configuration of digital watermark detection device)
Next, a basic configuration of a digital watermark detection apparatus corresponding to the digital watermark embedding apparatus shown in FIG. 1 will be described using FIG. 3 and FIG. 3 is provided with an embedded image signal 17 generated by the digital watermark embedding apparatus shown in FIG. 1 as an input embedded image signal 20 via a recording medium or a transmission medium. The embedded image signal 20 is branched into three and is given to one input of the specific frequency component extraction unit 21, the feature amount extraction unit 24, and the correlation calculator 25.

  The specific frequency component extraction unit 21 includes the same HPF or BPF as the specific frequency component extraction unit 11 used in the digital watermark embedding apparatus shown in FIG. 1, and the specific frequency component extraction unit 11 extracts from the embedded image signal 10. The same specific frequency component as the frequency component to be extracted is extracted from the embedded image signal 20.

  The phase and amplitude of the specific frequency component signal output from the specific frequency component extraction unit 21 are converted by the phase converter 22 and the amplitude converter 23. In the present embodiment, the phase converter 22 is arranged at the front stage and the amplitude converter 23 is arranged at the rear stage, but conversely, the amplitude converter 23 may be arranged at the front stage and the phase converter 22 may be arranged at the rear stage.

  The phase converter 22 is configured to perform phase conversion of a predetermined specific phase conversion amount on the specific frequency component signal. Specifically, the phase converter 22 is realized by a digital phase shifter as described later, and the phase conversion amount (phase shift amount) given by the phase converter 12 used in the digital watermark embedding apparatus shown in FIG. The same phase conversion amount (phase shift amount) is given.

  In the amplitude converter 23, the input specific frequency component signal is multiplied by a coefficient corresponding to the feature amount extracted from the embedded image signal 20 by the feature amount extraction unit 24, for example, the activity representing the complexity of the image. It is done.

  The specific frequency component signal whose phase and amplitude have been converted by the phase converter 22 and the amplitude converter 23 is given to the other input of the correlation calculator 25 and correlated with the embedded image signal 20 (more specifically, cross-correlation). An operation is performed, and the embedded watermark information 26 is detected. That is, when a change in the cross-correlation value with respect to the phase shift amount is observed, a peak appears at the position of the phase shift amount corresponding to the phase conversion amount of the phase converter 22, and the polarity of this peak represents the watermark information. The peak of the cross-correlation value takes either a positive value or a negative value depending on the watermark information. For example, the watermark information is determined to be “1” when positive, and the watermark information is determined to be “0” when negative. In this way, the watermark information 26 determined from the correlation calculator 25 is output.

  FIG. 4 shows a modified embodiment of the digital watermark detection apparatus of FIG. 3, and has a configuration suitable when the embedded image signal 20 has undergone scaling. When the embedding target image signal 20 is subjected to scaling, 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.

  Therefore, in the present embodiment, the phase shift amount of the phase converter 22 is controlled continuously or stepwise according to the phase shift amount information 27. Accordingly, the watermark information estimator 28 arranged on the output side of the correlation calculator 25 searches for the peak of the cross-correlation value output from the correlation calculator 23 as shown in FIG. Watermark information is estimated from the polarity of. In this example, since the cross-correlation value is positive, the watermark information is estimated (determined) as “1”.

  In the digital watermark embedding apparatus shown in FIG. 1, a plurality of phase shifters having different phase shift amounts are used as the phase converter 12 as will be described later, and the amplitude converter 13 has a plurality of amplitude conversion elements corresponding to the respective phase shifters. Suppose that it is prepared. In such a case, the phase converter 22 of FIG. 3 or FIG. 4 may be configured by a plurality of phase shifters, but is configured by a single phase shifter having a variable phase shift amount, and the phase shift amount is shown in FIG. The watermark information estimator 28 may search for the peak of the cross-correlation value output from the correlation calculator 25 while changing according to the phase shift amount information as shown in FIG. In this case, for example, as shown in FIG. 6, it is possible to detect the peak corresponding to the phase shift amount of the phase shifter when the watermark information is embedded, and to estimate each watermark information.

(Specific configuration example of digital watermark detection apparatus 1)
FIG. 7 shows a more specific embodiment of the digital watermark embedding apparatus according to the present invention. The correspondence relationship with FIG. 1 showing the basic configuration of the digital watermark embedding apparatus will be described. The high-pass filter (HPF) 31 is supplied to the specific frequency component extraction unit 11 by n phase shifters (PS) 32-1 to 32-n. , N exclusive OR circuits (EXOR) 33-1 to 33-n and multipliers (MPY) 34-1 to 34-n are connected to the amplitude converter 13, and the activity calculation circuit 35 is connected to the phase converter 12. The digital adder 36 corresponds to the feature amount extraction unit 15 and the watermark information superimposition unit 16, respectively.

  The specific frequency component signal output from the high-pass filter 31 is subjected to phase shifts of different shift amounts determined in advance by the phase shifters 32-1 to 32-n, and then the exclusive OR circuits 33-1 to 33-n. Given to one input of each. Each bit of the n-bit watermark information 14 (CCI) is given to the other input of each of the exclusive OR circuits 33-1 to 33-n. The outputs of the exclusive OR circuits 33-1 to 33-n are multiplied by the activity obtained by the activity calculation circuit 35 by the multipliers 34-1 to 24-n.

  The embedding signal output from the multipliers 34-1 to 34-n is added to the embedding target image signal 10 by the adder 36, whereby the watermark information 14 is embedded in the embedding target image signal 10, and the embedded image signal 17 is embedded. Is generated.

(Specific configuration example 2 of digital watermark detection apparatus)
FIG. 8 is a digital watermark embedding device which is a modification of FIG. 7, and a 3-input multiplier is substituted for the exclusive OR circuits 33-1 to 33-n and multipliers 34-1 to 34-n of FIG. (MPY) 37-1 to 37-n are used. The first input of the multipliers 37-1 to 37-n is supplied with the phase-shifted specific frequency component signals from the phase shifters 32-1 to 32-n, respectively, and the n-bit watermark information 14 is supplied to the second input. Each bit of (CCI) is given, and the activity obtained by the activity calculation circuit 35 is commonly given to the third input. Even with such a configuration, functions equivalent to those of the digital watermark embedding apparatus shown in FIG. 7 can be obtained.

(Specific configuration example of digital watermark detection apparatus)
FIG. 9 shows a more specific embodiment of the digital watermark detection apparatus according to the present invention, and corresponds to the digital watermark embedding apparatus of FIG. The correspondence relationship with FIG. 4 showing the basic configuration of the digital watermark detection apparatus will be described. The high-pass filter (HPF) 41 sends n phase shifters (PS) 42-1 to 42-n to the specific frequency component extraction unit 21. Are the phase converter 22, the n first multipliers (MPY) 43-1 to 43-n are the amplitude converter 23, the activity calculation circuit 44 is the feature quantity extraction unit 24, and the n second multipliers. (MPY) 45-1 to 45-n and cumulative adders 46-1 to 46-n correspond to the correlation calculator 25, and the CCI estimator 47 corresponds to the watermark information estimator 27, respectively.

  The specific frequency component signal output from the high-pass filter 41 is subjected to a phase shift of a predetermined shift amount that is the same as the shift amount of the phase shifters 32-1 to 32-n in FIG. 7 by the phase shifters 42-1 to 42-n. The first multipliers 43-1 to 43-n multiply the activity obtained by the activity calculation circuit 44 with the activity.

  The output signals from the first multipliers 43-1 to 43-n are multiplied by the embedded image signal 20 by the second multipliers 45-1 to 45-n, and output signals from the multipliers 45-1 to 45-n. Are cumulatively added by the cumulative adders 45-1 to 46-n and then input to the CCI estimator 47 to generate each bit of the watermark information 26 (CCI).

(Operation example 1 of digital watermark embedding / detecting device)
Next, a specific operation example when embedding 2-bit watermark information in the digital watermark embedding apparatus of FIG. 7 or FIG. 8 and detecting it by the digital watermark detection apparatus of FIG. 9 will be described with reference to FIGS. I will explain. In the digital watermark embedding apparatus of FIG. 7 or FIG. 8, the specific frequency component signal of FIG. 10B is extracted from the embedding target image signal 10 of FIG. The phase is shifted by a predetermined shift amount by the two phase shifters 32-1 and 32-2. For these phase shift signals, the factors expressing the 0th and 1st bits of the watermark information 14 (CCI) are the exclusive OR circuits 33-1 and 33-2 in FIG. 7 or the multiplier 37 in FIG. -1 and 37-2, respectively. For example, if the watermark information 14 is “0”, it is multiplied by −1, and if it is “1”, it is multiplied by +1. FIGS. 10C and 10D show the phase shift signals output from the exclusive OR circuits 33-1 and 33-2 or the multipliers 37-1 and 37-2 when the watermark information is (1, 1). Indicates.

  Furthermore, the activity obtained by the activity calculation circuit 35 is multiplied by the multipliers 34-1 and 34-2 as necessary for the phase shift signal, and then added to the embedding target image signal 10 by the adder 36. As a result, the embedded image signal 17 shown in FIG. In FIG. 10 (e), the solid line indicates the embedded image signal 17, and the embedding target image signal of FIG. 10 (a) and the phase shift signal of FIGS. 10 (c) and 10 (d) indicated by the broken line are added and synthesized. It has a waveform.

  On the other hand, when the watermark information is detected by the digital watermark detection apparatus of FIG. 9 from the embedded image signal in which the watermark information is embedded as shown in FIG. 10, first, the embedded image signal 20 (FIG. 10) of FIG. The specific frequency component signal shown in FIG. 11B is extracted by the high pass filter 41 from (e) corresponding to the embedded image signal 17). When the embedded image signal 20 is not scaled, the phase shifters 42-1 and 42-2 make the phase shifters 32-1 and 32-2 of FIG. 7 as shown in FIGS. The phase is shifted by a predetermined shift amount that is the same as the shift amount.

  Next, after the activity is multiplied by the first multipliers 44-1 and 44-2 on the phase shift signal of FIGS. 11 (b) and 11 (c) as necessary, the embedding of FIG. 11 (a) is completed. The image signal 20 is multiplied by the second multipliers 45-1 and 45-2 and further cumulatively added by the cumulative adders 46-1 and 46-2. The watermark information is determined from the correlation value peak. For example, if the cross-correlation value peak is positive, the watermark information is determined to be +1 (“1”), and if the cross-correlation value peak is negative, the watermark information is determined to be −1 (“0”).

  On the other hand, in order to cope with the case where scaling is performed on the embedded image signal 20, the phase shift amounts of the phase shifters 42-1 and 42-2 are controlled as described in FIG. The amount of phase shift is searched. That is, with the control of the phase shift amount, the CCI estimator 47 searches for the peak of the cross correlation value, and the watermark information 26 is estimated from the peak position.

  For example, when the embedded information 14 (CCI) is (1, 1), there are two positive peaks of cross-correlation values other than the origin (point where the phase shift amount is zero) as shown in FIG. Information is determined. Further, when the watermark information 14 (CCI) is (1, −1), the positive peak of the cross-correlation value exists near the origin as shown in FIG. 13, and the negative peak is far from the positive peak from the origin. The watermark information is determined by being present there.

(Operation example 2 of digital watermark embedding / detecting device)
Next, another operation example of the digital watermark embedding device of FIG. 7 or FIG. 8 and the digital watermark detection device of FIG. 9 will be described with reference to FIGS. In this digital watermark embedding apparatus, the polarity of the phase shift signal is inverted by any one of line, per line, per field, per field, per frame, per frame, or an appropriate combination thereof. In the following description, the case where the watermark information is 2 bits will be described.

  First, for the N (N = 1, 2,...) Line of the embedding target image signal 10, processing as shown in FIG. 14 is performed by the digital watermark embedding apparatus. The specific frequency component signal extracted by the high-pass filter 31 from the Nth line signal shown in FIG. 14A of the embedding target image signal 10 is a predetermined shift determined in advance by the two phase shifters 32-1 and 32-2. The factors representing the 0th bit and the 1st bit of the watermark information 14 (CCI) with respect to these phase-shifted signals are phase-shifted by the amount, and the exclusive OR circuits 33-1 and 33-2 in FIG. 8 multipliers 37-1, 37-2 respectively. For example, if the watermark information 14 is “0”, it is multiplied by −1, and if it is “1”, it is multiplied by +1. FIGS. 14B and 14C show the phase shift signals output from the exclusive OR circuits 33-1 and 33-2 or the multipliers 37-1 and 37-2 when the watermark information is (1, 1). Indicates.

  Furthermore, the activity obtained by the activity calculation circuit 35 is multiplied by the multiplier 34-1 and 34-2 or the multipliers 37-1 and 37-2 as necessary for the phase shift signal, and then the adder. By being added to the embedding target image signal 10 in 36, the embedding target image signal shown in FIG. 14 (a) and the phase shift signal shown in FIGS. 14 (b) and 14 (c) are added and synthesized. An embedded image signal 17 having a waveform indicated by a solid line is generated.

  Next, a process as shown in FIG. 15 is performed by the digital watermark embedding apparatus on the (N + 1) th line of the embedding target image signal 10. First, a specific frequency component signal extracted by the high-pass filter 31 from the signal of the (N + 1) th line shown in FIG. 15A of the embedding target image signal 10 is predetermined by the two phase shifters 32-1 and 32-2. Are shifted by the amount of shift, and the factors expressing the 0th and 1st bits of the watermark information 14 (CCI) for these phase shift signals are the exclusive OR circuits 33-1 and 33-2 in FIG. Alternatively, they are multiplied by the multipliers 37-1 and 37-2 in FIG.

  In this case, contrary to the case of the signal on the Nth line, for example, if the watermark information 14 is “0”, +1 is multiplied, and if it is “1”, 11 is multiplied. Therefore, the phase shift signals output from the exclusive OR circuits 33-1 and 33-2 or the multipliers 37-1 and 37-2 when the watermark information is (1, 1) are shown in FIG. Unlike c), the polarities are reversed as shown in FIGS.

  Furthermore, the activity obtained by the activity calculation circuit 35 is multiplied by the multiplier 34-1 and 34-2 or the multipliers 37-1 and 37-2 as necessary for the phase shift signal, and then the adder. By being added to the embedding target image signal 10 in 36, the embedding target image signal shown in FIG. 15 (a) and the phase shift signal shown in FIGS. 15 (b) and 15 (c) are added and synthesized. An embedded image signal 17 having a waveform indicated by a solid line is generated.

  In the above description, the polarity of the phase shift signal is inverted at the Nth and N + 1th lines of the embedding target image signal, that is, for each line. However, every line, every field, every field, every frame, and every frame The polarity of the phase shift signal may be inverted every time.

  On the other hand, in the digital watermark detection apparatus of FIG. 9, the polarity of the phase shift signal is inverted by any one of line, multiple lines, fields, fields, frames, frames, or an appropriate combination thereof. Correspondingly, the polarity is appropriately reversed during cumulative addition.

  For example, when the polarity of the phase shift signal is inverted for each line as described with reference to FIGS. 14 and 15, the cross-correlation value for the Nth line of the embedded image signal 20 is as shown in FIG. As shown in FIG. 17, a positive peak appears, but the cross-correlation value for the (N + 1) -th line of the embedded image signal 20 appears as a negative peak with respect to the phase shift amount as shown in FIG.

  Therefore, the cross-correlation values output from the multipliers 45-1 and 46-2 are inverted in polarity for each line and cumulatively added by the cumulative adders 46-1 and 46-1. In this case, the cross-correlation value after the cumulative addition appears as a positive peak continuously as shown in FIG. 18, whereby the watermark information is determined to be (1, 1).

  In this way, the polarity inversion of the phase shift signal is combined at the time of embedding the watermark information, and when detecting the watermark information, the cross-correlation value is inverted correspondingly and cumulatively added to thereby add the watermark on the image. While making the information inconspicuous, the falsification of the watermark information can be prevented more effectively.

(Operation example 3 of digital watermark embedding / detecting device)
Next, another operation example of the digital watermark embedding apparatus of FIG. 7 or FIG. 8 and the digital watermark detection apparatus of FIG. 9 will be described with reference to FIGS. This is a method in which the phase shift amount is horizontally reversed for each line in the digital watermark embedding apparatus. In the following description, the case where the watermark information is 2 bits will be described.

  First, the N-th (N = 1, 2,...) Line of the embedding target image signal 10 is processed by the digital watermark embedding apparatus as shown in FIG. The specific frequency component signal extracted by the high-pass filter 31 from the N-th line signal shown in FIG. 19A of the embedding target image signal 10 is advanced to the right by the two phase shifters 32-1 and 32-2, that is, the phase is advanced. The phase is shifted in the direction by a predetermined shift amount. The factors expressing the 0th bit and the 1st bit of the watermark information 14 (CCI) for these phase shift signals are the same as in the exclusive OR circuit 33-1 and 33-2 in FIG. Multiplied by multipliers 37-1 and 37-2, respectively. 19B and 19C show the phase shift signals output from the exclusive OR circuits 33-1 and 33-2 or the multipliers 37-1 and 37-2 when the watermark information is (1, 1). Indicates.

  Furthermore, the activity obtained by the activity calculation circuit 35 is multiplied by the multiplier 34-1 and 34-2 or the multipliers 37-1 and 37-2 as necessary for the phase shift signal, and then the adder. By being added to the embedding target image signal 10 in 36, the embedding target image signal shown in FIG. 19 (a) and the phase shift signal shown in FIGS. 19 (b) and 19 (c) are added and synthesized. An embedded image signal 17 having a waveform indicated by a solid line is generated.

  On the other hand, for the (N + 1) th line of the embedding target image signal 10, processing as shown in FIG. 20 is performed by the digital watermark embedding apparatus. The specific frequency component signal extracted by the high-pass filter 31 from the signal of the (N + 1) th line shown in FIG. 20A of the embedding target image signal 10 is delayed to the left by the two phase shifters 32-1 and 32-2, that is, the phase is delayed. The phase is shifted in the direction by a predetermined shift amount. The factors expressing the 0th bit and the 1st bit of the watermark information 14 (CCI) for these phase shift signals are the same as in the exclusive OR circuit 33-1 and 33-2 in FIG. Multiplied by multipliers 37-1 and 37-2, respectively. 20B and 20C show the phase shift signals output from the exclusive OR circuits 33-1 and 33-2 or the multipliers 37-1 and 37-2 when the watermark information is (1, 1). Indicates.

  Furthermore, the activity obtained by the activity calculation circuit 35 is multiplied by the multiplier 34-1 and 34-2 or the multipliers 37-1 and 37-2 as necessary for the phase shift signal, and then the adder. By being added to the embedding target image signal 10 in 36, the embedding target image signal shown in FIG. 20 (a) and the phase shift signal shown in FIGS. 20 (b) and 20 (c) are added and synthesized. An embedded image signal 17 having a waveform indicated by a solid line is generated.

  On the other hand, in the digital watermark detection apparatus shown in FIG. 9, the cumulative information of the cross-correlation values is simply added for each line, and the peak information is searched to detect the watermark information. However, as described in the operation example 2, the polarity of the phase shift signal is reversed by any one of line, multiple lines, fields, fields, frames, frames, or an appropriate combination thereof. If so, polarity inversion is performed for each cumulative addition unit.

  FIG. 21 shows the cross-correlation value after cumulative addition for each line in this case. When searching for the phase shift amount, when the phase shift amount is searched in the positive direction and the search is performed in the negative direction, cross-correlation values that are approximately the same shape, that is, line-symmetric with respect to the center, are obtained. By utilizing such a property of the cross-correlation value, the watermark information can also be detected by searching only in one direction (for example, right direction) at the time of searching.

(Operation example 4 of digital watermark embedding / detecting device)
Next, still another operation example of the digital watermark embedding apparatus of FIG. 7 or FIG. 8 and the digital watermark detection apparatus of FIG. 9 will be described with reference to FIGS. In the example described below, a calibration signal is embedded together when watermark information is embedded, and the method is used for detection of watermark information. A specific operation example is given below.

  (1) When embedding N-bit watermark information in the digital watermark embedding apparatus, an N + 1-bit phase shift signal is generated, and one bit other than N bits used for watermark information embedding is used as a calibration signal. Are always embedded at a level of +1 (or -1). This calibration signal is a signal that serves as a reference when watermark information is detected.

  On the other hand, in the digital watermark detection apparatus, the watermark information is detected based on the known relative relationship between the cross-correlation value at each position and the cross-correlation value corresponding to the calibration signal. For example, if it is assumed that the calibration signal is embedded at a level of +1 (or −1), the cross-correlation value at the position corresponding to the calibration signal and the cross-correlation value at other embedding positions are as shown in FIG. In the case of the same polarity, +1 (or -1) is set.

  (2) In the digital watermark embedding apparatus, the calibration signal may be embedded at a position where the phase shift amount is minimum or maximum. In this case, the digital watermark detection apparatus detects a calibration signal having a minimum or maximum phase shift amount, and determines watermark information embedded in a relative relationship with the calibration signal.

  (3) In the digital watermark embedding apparatus, the calibration signal is embedded as a predetermined value (for example, +1 or −1) at a position where the phase shift amount is minimum (or maximum), and a plurality of phase shift amounts are set at equal intervals. Then, ternary information {+1, 0, −1} is embedded in each phase shift position.

  Specifically, for example, when the ternary information is {+1}, a phase shift signal multiplied by a positive multiplier is added to the embedding target image signal 10, and when {−1}, a phase shift multiplied by a negative multiplier is added. The signal is added to the embedding target image signal 10, and in the case of {0}, nothing is added to the embedding target image signal 10.

  On the other hand, the digital watermark detection apparatus obtains a cross-correlation value at the phase shift position of the watermark information estimated from the calibration signal. As shown in FIG. 23, when the cross-correlation value is near 0, the ternary information is determined to be {0}, and when it is not near 0, the ternary information is determined by the relative relationship with the cross-correlation value of the calibration signal. {+1, -1} is determined. The following is an application example of (3).

  (3-1) In the digital watermark embedding apparatus, as shown in FIG. 24, binary numbers are encoded in advance as ternary numbers as watermark information, and these are embedded as ternary information as described above. In the electronic watermark detection apparatus, the original binary watermark information is detected by decoding the ternary number detected as the ternary information.

  (3-2) In the digital watermark embedding apparatus as described above, binary numbers are encoded in advance as ternary numbers as watermark information, and these are embedded as ternary information as described above. As shown in FIG. A combination in which the embedded information is all 0 is not used (forbidden). In the electronic watermark detection apparatus, the original binary watermark information is detected by decoding the ternary number detected as the ternary information.

  (3-3) Three values are used for CCI carry.

  (3-4) Embedding and detecting watermark information with Copy Free (replicatable) as +1, Copy Once (replicatable only once) as 0, and Never Copy (replicatable as impossible) as -1. In this case, when watermark information is remarked, −1 is embedded at 0, so that it is possible to save time and effort for cancellation.

(Operation example 5 of digital watermark embedding / detecting device)
Next, another operation example of the digital watermark embedding apparatus in FIG. 7 or FIG. 8 and the digital watermark detection apparatus in FIG. 9 will be described. In the digital watermark embedding apparatus, a plurality of phase shift amounts are set at arbitrary intervals. Do not break the relative relationship. In this case, the digital watermark detection apparatus counts the cross-correlation value peaks as shown in FIG. 26, and determines, for example, the innermost side closest to the origin as bit 0, the outer side as bit 1,.

(Operation example 6 of digital watermark embedding / detecting device)
Next, another operation example of the digital watermark embedding apparatus of FIG. 7 or FIG. 8 and the digital watermark detection apparatus of FIG. 9 will be described with reference to FIG. In the digital watermark embedding apparatus, the outermost position of the existing embedding position where the watermark information is embedded is obtained, and information for Remark is additionally written on the outermost position.

  On the other hand, the digital watermark detection apparatus searches until the peak of the cross-correlation value disappears, and determines the information after the remark from the information embedded on the outermost side.

(Operation example 7 of digital watermark embedding / detecting device)
Next, the operations of the digital watermark embedding apparatus and the digital watermark detection apparatus of FIG. 3 or 4 when the phase converter 12 is controlled using the watermark information 14 in the digital watermark embedding apparatus of FIG. 1 will be described.

  The phase converter 12 includes, for example, four phase shifters having different phase shift amounts and switches that select these phase shifters, and specific frequency component signals from the specific frequency component extraction unit 11 are parallel to these phase shifters. Shall be entered. If the phase shift amounts of the four phase shifters are θ1, θ2, θ3, and θ4, for example, the 0th bit of the watermark information 14 is represented by the presence or absence of superposition of specific frequency component signals that have undergone a phase shift of θ1 and θ2. The first bit of the watermark information 14 is represented by whether or not the specific frequency component signal subjected to the phase shift of θ3 and θ4 is superimposed. Specifically, for example, the specific frequency component signal is superimposed on the embedding target image signal 10 according to the following combinations (a-1) and (a-2) or (b-1) and (b-2).

  (A-1) If the 0th bit = “1”, only the specific frequency component signal subjected to the phase shift of θ1 is superimposed, and the specific frequency component signal subjected to the phase shift of θ2 is not superimposed. (A-2) If the first bit = “1”, only the specific frequency component signal subjected to the phase shift of θ3 is superimposed, and the specific frequency component signal subjected to the phase shift of θ4 is not superimposed.

  (B-1) If the 0th bit = “1”, only the specific frequency component signal subjected to the phase shift of θ1 is superimposed, and the specific frequency component signal subjected to the phase shift of θ2 is not superimposed. (B-2) If the first bit = “1”, only the specific frequency component signal subjected to the phase shift of θ4 is superimposed, and the specific frequency component signal subjected to the phase shift of θ3 is not superimposed.

  On the other hand, in the digital watermark detection apparatus shown in FIG. 3, when scaling with respect to the input embedded image signal 20 is not considered, the phase shift amounts of the four phase shifters constituting the phase converter 22 are set to the phase in the digital watermark embedding apparatus. It is set in the same manner as the phase shift amounts θ1, θ2, θ3, θ4 of the phase shifter of the converter 12, and the watermark information is determined from the cross-correlation values at θ1, θ2, θ3, θ4.

  FIG. 28 shows the cross-correlation value when the watermark information is (1, 1), and FIG. 29 shows the cross-correlation value when the watermark information is (1, -1). The watermark information can be determined from the cross-correlation values at the phase shift amounts θ1, θ2, θ3, and θ4.

  When considering the scaling for the embedded image signal 20, the phase shift amount may be changed and the phase shift amount given by the digital watermark embedding apparatus may be searched to determine the cross-correlation value as described above.

  Second Embodiment Next, another embodiment of the present invention will be described with reference to FIGS. The present embodiment inserts an amplitude limiter into the digital watermark embedding device and the digital watermark detection device to limit the amplitude of the signal to be superimposed on the embedding target image signal 10, thereby reducing the watermark information at a low level of the embedding target image signal 10. Therefore, it is possible to more effectively prevent image quality degradation by embedding uniformly over a wide level range from high to low.

  In the digital watermark embedding device shown in FIG. 30, an amplitude limiter 18 is inserted between the specific frequency component extraction unit 11 and the phase and amplitude conversion unit (in this example, the phase converter 12). In the digital watermark detection apparatus shown in FIG. 31, an amplitude limiter 28 is inserted between the specific frequency component extraction unit 21 and the conversion unit (phase converter 22 in this example) corresponding to FIG.

  In the digital watermark embedding apparatus shown in FIG. 32, an amplitude limiter 18 is inserted between the phase and amplitude conversion unit (in this example, the amplitude converter 13) and the watermark information superimposing unit 16. In the digital watermark detection apparatus shown in FIG. 33, an amplitude limiter 28 is inserted between the phase and amplitude converter (in this example, the amplitude converter 23) and the correlation calculator 25 in correspondence with FIG.

[Third Embodiment]
Next, still another embodiment of the present invention will be described with reference to FIGS. This embodiment provides a digital watermark technique that is more resistant to attacks by embedding and detecting watermark information depending on randomized information, thereby preventing watermark information from being detected without knowing the randomized information. .

  The randomized information may be generated inside the digital watermark embedding device or the detection device, or may be input from the outside of the device if safety is maintained. The randomized information may be constant or may be changed as needed during the digital watermark embedding process or the detection process. For example, different randomized information is used in the left half (first half of one horizontal scanning period) and right half (second half of one horizontal scanning period) in the same line of the image signal, or different randomized information is used for each line. Changes may be made.

  The digital watermark embedding apparatus shown in FIG. 34 is an example in which, when the filters constituting the specific frequency component extraction unit 11 have different characteristics depending on the parameters, the parameters are given by the secret randomized information 19.

  In the digital watermark detection apparatus shown in FIG. 35, the parameters of the filter constituting the specific frequency component extraction unit 21 are given by the randomized information 39 corresponding to FIG. The randomized information 29 is the same as the randomized information 19 used in the digital watermark embedding apparatus shown in FIG. 34, and the randomized information 29 is correctly generated only by the digital watermark detection apparatus generated inside the apparatus or given from the outside of the apparatus. 26 can be detected.

  FIGS. 36 and 37 are examples of filters used in the specific frequency component extraction units 11 and 21 in FIGS. 34 and 35, and the continuous pixel values {... P (h−1) of the embedding target image signal (original signal). ), P (h), p (h + 1),...} Are multiplied by a coefficient, and the sum of the multiplication results is taken as a filter output. Since the coefficients can be randomized within a certain range, these coefficients are used as the randomized information 19.

  In the digital watermark embedding apparatus shown in FIG. 38, the phase shift amount of the phase shifter constituting the phase converter 12 is randomized according to the randomized information 19, thereby making the shape of the peak of the autocorrelation value difficult to see. . In this case, it is desirable to change the randomized phase shift amount frequently. For example, the phase shift amount is set to a different value between the left half and the right half of the screen. The screen may be divided into a plurality of strip-shaped regions extending in the vertical direction, and the phase shift amount may be set to a different value for each region.

  In the digital watermark detection apparatus shown in FIG. 39, the phase shift amount of the phase shifter constituting the phase conversion unit 22 is randomized according to the randomization information 29, corresponding to the digital watermark embedding apparatus of FIG. The randomized information 29 is the same as the randomized information 19 used in the digital watermark embedding apparatus of FIG. 38, and the randomized information 29 is correctly generated only by the digital watermark detection apparatus generated inside the apparatus or given from the outside of the apparatus. 26 can be detected.

  FIG. 40 is an example of a phase shifter having a variable phase shift amount used in the phase converters 11 and 21 of FIGS. 38 and 39. A plurality of phase shift elements are connected in series, and each tap (input of the phase shift element is connected). Output) is selected by the selector according to the randomization information.

  In the digital watermark embedding apparatus shown in FIG. 41, an amplitude modulator 51 is inserted between the phase and amplitude conversion unit (in this example, the amplitude converter 13) and the watermark information superimposing unit 16, and the embedded signal is converted into randomized information 19. According to the amplitude modulation.

  In the digital watermark detection apparatus shown in FIG. 42, an amplitude modulator 61 is provided between the phase and amplitude converter (in this example, the amplitude converter 23) and the correlation calculator 25 in correspondence with the digital watermark embedding apparatus shown in FIG. The embedded signal is amplitude-modulated according to the randomized information 29. The randomized information 29 is the same as the randomized information 19 used in the digital watermark embedding apparatus shown in FIG. 41, and the randomized information 29 is generated correctly inside the apparatus or only by the digital watermark detection apparatus provided from the outside of the apparatus. 26 can be detected.

  In the digital watermark embedding apparatus shown in FIG. 43, an embedding signal and an embedding target image are inserted by inserting a non-linear filter 52 between the phase and amplitude converting section (in this example, the amplitude converter 13) and the watermark information superimposing section 16. By reducing the correlation with the signal 10, no peak appears in the autocorrelation value.

  The digital watermark detection apparatus shown in FIG. 44 corresponds to the digital watermark embedding apparatus shown in FIG. 43 between the phase and amplitude conversion unit (in this example, the amplitude converter 23) and the correlation calculator 25. A non-linear filter 62 having a reverse characteristic of the non-linear filter 52 used in the embedding apparatus is inserted.

  As the non-linear filter 52, a filter using a trigonometric function or amplitude modulation by a higher order expression can be used. For example, it is a nonlinear filter that outputs sin (x), x2, etc., where x is an input signal. FIG. 45 shows a specific example of the nonlinear filter 52.

  FIG. 45A shows a squarer that outputs x2 when the input signal is x by giving the same input signal to two inputs of the multiplier. The part overflowed by the square is discarded. FIG. 45B shows a nonlinear filter in which the relationship between the input value and the output value is implemented in a tabular form so that complicated nonlinear transformation can be expressed. For example, by using this nonlinear conversion table as a sine table, a nonlinear filter that outputs sin (ax) with respect to the input signal x can be realized.

1 is a block diagram showing a basic configuration of a digital watermark embedding device according to an embodiment of the present invention. The figure explaining the phase shift of the specific frequency component signal by the phase converter in the same embodiment 1 is a block diagram showing the basic configuration of a digital watermark detection apparatus according to the embodiment FIG. 2 is a block diagram showing another basic configuration of the digital watermark detection apparatus according to the embodiment. The figure which shows the operation example of the peak search of a cross correlation value and the watermark information detection in the digital watermark detection apparatus based on the embodiment The figure which shows the operation example of the peak search of a cross correlation value and the watermark information detection in the digital watermark detection apparatus based on the embodiment FIG. 3 is a block diagram showing a more specific configuration example of the digital watermark embedding apparatus according to the embodiment. FIG. 3 is a block diagram showing another more specific configuration example of the digital watermark embedding apparatus according to the embodiment. FIG. 3 is a block diagram showing a more specific configuration example of the digital watermark detection apparatus according to the embodiment. Waveform diagram of each part showing the operation of the digital watermark embedding apparatus of FIG. 7 or FIG. Waveform diagram of each part showing the operation of the digital watermark detection apparatus of FIG. The figure which shows the operation | movement of the peak search of a cross correlation value and watermark information detection in case the watermark information is (1, 1) in the digital watermark detection apparatus of FIG. The figure which shows the operation | movement of the peak search of a cross correlation value and watermark information detection in case the watermark information is (1, 1) in the digital watermark detection apparatus of FIG. Waveform diagram of each part showing processing for the image signal of the Nth line in the digital watermark embedding apparatus of FIG. 7 or FIG. Waveform diagram of each part showing processing for the image signal of the (N + 1) -th line of the digital watermark embedding apparatus of FIG. 7 or FIG. 14 is a diagram for explaining a cross-correlation value in the digital watermark detection apparatus of FIG. 9 for the embedded image signal obtained by the process of FIG. 15 is a diagram for explaining a cross-correlation value in the digital watermark detection apparatus of FIG. 9 for the embedded image signal obtained by the process of FIG. The figure which shows the watermark information detection operation | movement in the digital watermark detection apparatus of FIG. 9 with respect to the embedded image signal obtained by the process of FIG. Waveform diagram of each part showing other processing for the image signal of the Nth line of the digital watermark embedding apparatus of FIG. 7 or FIG. Waveform diagram of each part showing other processing for the image signal of the (N + 1) th line of the digital watermark embedding apparatus of FIG. 7 or FIG. 9 is a diagram for explaining a cross-correlation value in the digital watermark detection apparatus of FIG. 9 for an embedded image signal obtained by the processing of FIGS. 19 and 20. 9 is a diagram showing a cross-correlation value and watermark information detection operation in the digital watermark detection apparatus in FIG. 9 when a calibration signal is embedded together with watermark information in the digital watermark embedding apparatus in FIG. 7 or FIG. 9 is a diagram showing another example of the cross-correlation value and watermark information detection operation in the digital watermark detection apparatus of FIG. 9 when the calibration signal is embedded together with the watermark information in the digital watermark embedding apparatus of FIG. 7 or FIG. The figure which shows the table which encodes the binary number which is watermark information to a ternary number expression in the digital watermark embedding apparatus of FIG. 7 or FIG. The figure which shows the other table which encodes the binary number which is watermark information to a ternary number expression in the digital watermark embedding apparatus of FIG. 7 or FIG. 9 is a diagram showing a watermark information detection operation in the digital watermark detection apparatus of FIG. 9 when a plurality of phase shift amounts are set at arbitrary intervals while maintaining a relative relationship in the digital watermark embedding apparatus of FIG. 7 or FIG. The figure which shows the watermark information detection operation | movement in the digital watermark detection apparatus of FIG. 9 at the time of adding the information for remarks in the digital watermark embedding apparatus of FIG. 7 or FIG. In the digital watermark embedding apparatus in FIG. 1, the watermark information (1, 1) is embedded depending on whether or not a specific frequency component signal subjected to a fixed phase shift by four phase shifters is superimposed. Diagram showing watermark information detection operation The digital watermark detection apparatus of FIG. 3 when the watermark information (1, −1) is embedded depending on whether or not a specific frequency component signal subjected to a fixed phase shift by four phase shifters in the digital watermark embedding apparatus of FIG. Showing watermark information detection operation 1 is a block diagram showing a basic configuration of a digital watermark embedding apparatus using an amplitude limiter according to an embodiment of the present invention. 2 is a diagram showing a basic configuration of a digital watermark detection apparatus using an amplitude limiter according to the embodiment. FIG. 1 is a block diagram showing a basic configuration of a digital watermark embedding apparatus using an amplitude limiter according to an embodiment of the present invention. 2 is a diagram showing a basic configuration of a digital watermark detection apparatus using an amplitude limiter according to the embodiment. FIG. 1 is a block diagram showing a basic configuration of a digital watermark embedding apparatus using randomized information according to an embodiment of the present invention. 1 is a block diagram showing a basic configuration of a digital watermark detection apparatus using randomized information according to the embodiment The figure which shows the specific structural example of the specific frequency component extraction part in FIG.34 and FIG.35. The figure which shows the other specific structural example of the specific frequency component extraction part in FIG.34 and FIG.35. 1 is a block diagram showing a basic configuration of a digital watermark embedding apparatus using randomized information according to an embodiment of the present invention. 1 is a block diagram showing a basic configuration of a digital watermark detection apparatus using randomized information according to the embodiment The figure which shows the specific structural example of the phase conversion part in FIG.38 and FIG.39. 1 is a block diagram showing a basic configuration of a digital watermark embedding apparatus using randomized information according to an embodiment of the present invention. 1 is a block diagram showing a basic configuration of a digital watermark detection apparatus using randomized information according to the embodiment 1 is a block diagram showing a basic configuration of a digital watermark embedding apparatus using randomized information according to an embodiment of the present invention. 1 is a block diagram showing a basic configuration of a digital watermark detection apparatus using randomized information according to the embodiment 43 is a block diagram illustrating a specific configuration example of the nonlinear filter in FIGS.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Embedded object image signal 11 ... Specific frequency component extraction part 12 ... Phase converter 13 ... Amplitude converter 14 ... Watermark information 15 ... Feature-value extraction part 16 ... Watermark information superimposition part 17 ... Embedded image signal 18 ... Amplitude limiter 19 ... Randomized information 20 ... Embedded image signal 21 ... Specific frequency component extraction unit 22 ... Phase converter 23 ... Amplitude converter 24 ... Feature amount extraction unit 25 ... Correlation calculator 26 ... Watermark information 27 ... Watermark information estimator 28 ... Amplitude limiter 29 ... Randomized information 31 ... High-pass filter 32-1 to 32-n ... Phase shifter 33-1 to 33-n ... Exclusive OR circuit 34-1 to 34-n ... Multiplier 35 ... Activity calculation circuit 36: Adders 37-1 to 37-n ... Multipliers 41 ... High-pass filters 42-1 to 42-n ... Phase shifters 43-1 to 43 n: first multiplier 44 ... activity calculation circuit 45-1 to 45-n ... second multiplier 46-1 to 46-n ... cumulative adder 47 ... watermark information estimator 51 ... amplitude modulator 52 ... nonlinear filter 61 ... Amplitude modulator 62 ... Nonlinear filter

Claims (5)

In a digital watermark detection method for detecting watermark information from an image signal in which watermark information is embedded,
Extract a specific frequency component signal from the input image signal,
Perform phase conversion of the extracted specific frequency component signal by at least one conversion means,
Performs cross-correlation operation between the specific frequency component signal that has undergone phase conversion and the input image signal, and extracts the watermark information from the polarity of the peak of the cross-correlation value change with respect to the change in the phase conversion amount Digital watermark detection method. In a digital watermark detection apparatus for detecting watermark information from an image signal in which watermark information is embedded,
Extraction means for extracting a specific frequency component signal from the input image signal;
At least one conversion means for performing phase conversion of the extracted specific frequency component signal;
A cross-correlation operation between the specific frequency component signal phase-converted by the converting means and the input image signal is performed, and the peak polarity of the change of the cross-correlation value with respect to the change of the phase conversion amount of the phase conversion An electronic watermark detection apparatus comprising correlation calculation means for extracting watermark information.   3. The digital watermark detection apparatus according to claim 2, further comprising an amplitude limiter that limits an amplitude of the specific frequency signal inserted between the extraction unit and the correlation calculation unit.   3. The digital watermark detection apparatus according to claim 2, wherein at least one of the characteristics of the extraction unit and the conversion unit is randomized by randomized information.   The digital watermark detection apparatus according to claim 2, further comprising a non-linear filter inserted between the conversion unit and the correlation calculation unit.

Images