JP2010141591A - Digital watermark embedding method and device, and digital watermark detecting method and device - Google Patents

Abstract

Watermark information can be stably detected even when a time lag occurs between embedding and detection of a digital watermark.
Digital watermark embedding generates an embedded signal from a specific watermark signal and watermark information, detects a scene change from an input image signal, and controls the polarity based on the detected scene change information. A digital watermark embedded image is generated by synthesizing an embedded signal with the input image. Digital watermark detection calculates the correlation between an input image signal and a specific watermark signal, detects a scene change from the input image signal, and accumulates the correlation signal while controlling the polarity based on the detected scene change information Then, the embedded information is detected from the peak appearing in the accumulated correlation signal.
[Selection] Figure 8

Description

    The present invention relates to a digital watermark embedding method and apparatus, and a digital watermark detection method and apparatus.

  A digital moving image can be easily produced as a high-quality copy (copy), and may be copied without restriction unless some kind of copy inhibition or copy control is performed. Therefore, in order to prevent illegal copying of a digital moving image or to control the number of generations of copying by a regular user, a method for adding information for copying control to a digital moving image has been considered.

  As a technique for superimposing other additional information on a digital moving image, a digital watermark 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 Information such as secret information and duplication control information that are sometimes necessary is embedded in a state that cannot be recognized by human perception. Later, by detecting watermark information from the content as necessary, copyright protection including use control and copy control can be performed, or secondary use can be promoted.

  Various methods have been proposed as a digital watermarking method, and a method using a spread spectrum technique is known as one of them. A technique has been proposed in which a watermark signal is created and embedded using a specific frequency component signal extracted from an image signal of an original image in which watermark information is embedded (for example, Patent Document 1).

Furthermore, in recent years, the problem of illegal copying, such as shooting a theater screen with a video camera, has become apparent, and in such shot images, in addition to geometric deformation, the time caused by the rotational speed deviation of the projector is shown. It is known that a gap has occurred. Therefore, in order to use digital watermarks as a measure for preventing unauthorized copying by private shooting in a movie theater, not only geometrical deformation but also stable performance against time shift is required.
JP 2006-254147 A

  In the above prior art, at the time of detection, watermark information is detected by accumulating signals for a long time with a specific polarity inversion pattern at the time of embedding. Therefore, in order to accurately detect watermark information, it is necessary to synchronize the polarity inversion pattern between embedding and detection. There is a problem that the accuracy of watermark information detection is not stable due to accumulation of accumulated pattern errors between embedding and detection due to time lag.

  The present invention has been made to solve the above-described problem, and is a case where a time lag has occurred by preventing accumulation of accumulated pattern errors between embedding and detection due to a time lag. However, an object of the present invention is to provide a digital watermark detection apparatus, method and program capable of detecting watermark information more stably.

  In order to solve the above problems, the present invention provides an embedded signal generation that generates an embedded signal in accordance with the watermark information in an electronic watermark embedding apparatus that generates an embedded image in which watermark information is embedded in an original image. A scene change detector for detecting a scene change frame in which a scene change occurs before the display timing of the frame in the original image, the scene change frame as a head of an arbitrary pattern cycle, and the original image A polarity controller for obtaining a time-series polarity change pattern when embedding the embedded signal in an image, and changing the polarity of the embedded signal for each frame of the original image according to the polarity change pattern; And an embedded signal superimposing device for obtaining the embedded image by synthesizing the electronic watermark.

  In addition, in an electronic watermark detection apparatus that detects watermark information from an input image in which watermark information is embedded in an original image, an embedded signal generator that generates an embedded signal in accordance with the watermark information, and the input image Among them, a scene change detector that detects a scene change frame in which a scene change occurs before the display timing of the frame, a correlation calculation unit that obtains a correlation value indicating a degree of correlation between the input image and the watermark signal, An arbitrary pattern starting from the scene change frame, a polarity controller for obtaining a time-series polarity change pattern of polarity when the embedded signal is embedded in the original image as the input image signal, and the correlation value A superimposer that cumulatively adds according to the polarity change pattern, and an estimator that estimates the watermark information from the signals cumulatively added by the superimposer Provides a digital watermark detection apparatus characterized by comprising a.

  According to the present invention, watermark information can be detected more stably even when a time lag occurs.

  Hereinafter, a digital watermark embedding device and a digital watermark detection device according to an embodiment of the present invention will be described with reference to the drawings. Here, the same code | symbol is attached | subjected to the mutually same structure and duplication description is abbreviate | omitted.

  The digital watermark embedding device is information such as identification information of the copyright owner or user of the content, copyright holder rights information, content usage conditions, secret information necessary for the use, and copy control information (hereinafter referred to as watermark information). Embedded in the content. The digital watermark detection apparatus detects watermark information from an input image (an image in which watermark information is embedded) input via a recording medium or a transmission medium. The image includes a moving image and a still image (frame).

[First Embodiment]
In order to effectively separate the image signal component and the digital watermark signal component, it is necessary to superimpose / accumulate while inverting the polarity (sign) of the signal in the same pattern as in embedding at the time of detection.

  In embedding a digital watermark, an image pattern (embedded signal) associated with watermark information to be embedded is generated. Next, a “+” or “−” sign is assigned to each frame of the original image. Finally, the image pattern of the embedded signal generated earlier is embedded in each frame. At that time, the frame assigned with “−” is embedded by inverting the sign of the embedded signal. Also, it is embedded as it is in the frame to which the sign “+” is assigned.

  At the time of detection, a sign “+” or “−” is assigned to each frame of the input image in the same order as at the time of embedding. At this time, in a frame to which “−” is assigned, all the pixel values are made negative and then added. When an average image obtained by cumulative addition is created, an embedded image pattern (embedded signal) emerges. In the present embodiment, in order to reduce the amount of memory and the amount of processing, watermark information is detected from a signal obtained by performing cumulative addition of correlation values between an input image signal and a watermark signal. An average image is generated in which the embedded signal embedded with the polarity inverted in the frame to which “-” is assigned is re-inverted at the time of detection. Therefore, in the average image, the embedded image pattern is accumulated. On the other hand, in the image signal of the original image, the polarity of the pixel value is inverted for the first time at the time of detection in a frame to which “−” is assigned. Therefore, the image signal component of the original image is canceled and becomes a value close to 0.

  In order to effectively separate the image signal component of the original image and the signal component of the digital watermark, it is necessary to synchronize the polarity inversion pattern between detection and embedding. In the present embodiment, in order to detect the embedded signal more stably, the polarity inversion pattern is synchronized at the scene change point between detection and embedding. Thereby, even when a time shift occurs between the original image and the input image, it is possible to prevent the shift from propagating for a long time. In the case of processing using a linear correlation value such as cross-correlation, the superimposed signal and the accumulated signal match. However, in the case of processing using a nonlinear correlation value such as phase-only correlation, the superimposed signal And the accumulated signal do not match.

  A synchronization example at the time of embedding / detecting a digital watermark using scene change information will be described with reference to FIG. In FIG. 1, the original image and the input image to be detected in which watermark information is embedded have the same scene change points A, B, C, and D. In this case, a time lag occurs. For example, scene change points A, B, C, and D are input at times t1, t2, t3, and t4 that occur in the original image, and scene change points A, B, C, and D are input. It is assumed that there is a time lag between times t′1, t′2, t′3, and t′4 occurring in the image. Even in that case, the time-series positions where scene changes (locations where the amount of histogram change between frames before and after the display time exceeds a certain threshold) occur substantially coincide between the original image and the input image. . Therefore, by embedding watermark information in the original image in a time-series pattern using scene change information and accumulating in the same pattern as in embedding using scene change points at the time of detection, Thus, the patterns can be synchronized. In this embodiment, since the accumulated pattern is reset at the scene change position, even if there is a time lag between the original image and the input image, stable performance is achieved without accumulating accumulated pattern errors. It becomes possible to do. In addition, it is not limited to the pattern illustrated below, Various patterns may be sufficient.

  The digital watermark embedding device of this embodiment detects a scene change of an image in which watermark information is to be embedded. The polarity at the time of embedding an embedded signal in the original image is controlled using the detected scene change. As shown in FIG. 1, it is assumed that four scene changes (A, B, C, D) are detected when a scene change of the original image is detected. In the embedding, the embedding pattern after the scene change is determined in advance. Here, after the scene change, the embedding signal is embedded in each frame by repeating the polarity of the pattern A “+ − + −− +” for each frame. Shall.

  The digital watermark detection apparatus according to the present embodiment detects a scene change point of an input image in which watermark information is embedded, and generates a polarity inversion pattern when accumulating the image signal of the input image using the detected scene change. Control. As shown in FIG. 1, in the scene change detection of the input image, it is assumed that the scene change at the same position (four locations) as when the digital watermark is embedded is detected. In the detection, the detection pattern after the scene change is determined in advance. Here, after the scene change, the signal is cumulatively added by repeating the pattern A “+ − + −− +” for each frame. It is assumed that watermark information is detected from an average image generated by cumulative addition.

(Digital watermark embedding device)
First, the digital watermark embedding device of this embodiment will be described. The digital watermark embedding device can be used, for example, by a producer who creates a video, a screener who shows a video in a movie theater, and the like, and can create a video in which a digital watermark is embedded.

  FIG. 2 is a diagram showing a configuration of the digital watermark embedding apparatus 1 of the present embodiment.

  The digital watermark embedding apparatus 1 receives a digitized image signal as an image in which watermark information is to be embedded (hereinafter referred to as “original image 100”). An embedded image 102 that is an image in which watermark information 101 is embedded in the original image 100 by a watermark signal 103 is generated and output. Note that the image signal of the original image 100 may include both a luminance signal and a color difference signal, or may be only a luminance signal.

  The digital watermark embedding apparatus 1 includes an embedded signal generator 11, an embedded signal superimposer 12, a scene change detector 13, and a polarity controller 14.

  The embedded signal generator 11 generates a signal embedded in the original image 100 (hereinafter referred to as “embedded signal 106”) from the input watermark signal 103 in accordance with the watermark information 101 to be embedded in one frame. The watermark information 101 is a signal sequence of “1” or “0” of a digital signal, and has a plurality of bits of data. The watermark information 101 indicates information such as identification information of the copyright owner or user of the content, rights information of the copyright owner, usage conditions of the content, secret information necessary at the time of use, and copy control information. The sign of the watermark signal 103 is determined according to whether each bit is “1” or “0”. Then, the phase of the watermark signal 103 is shifted by the phase corresponding to the position to be embedded in the frame of the original image 100 of each of the plurality of bits. The embedded signal 106 is a signal per frame on which a signal corresponding to each bit is superimposed.

  The watermark signal 103 is various general digital watermark signals. For example, there are various signals such as a signal such as random noise, a specific watermark image signal unique to each frame, and a specific frequency component signal of the original image. . Random noise is characterized in that the embedded watermark signal is difficult to perceive. The specific watermark signal 103 that is unique for each frame uses a specific watermark image signal that is resistant to various attacks so that the watermark information is not lost or altered by the attack (hereinafter referred to as robust). ) Digital watermark signal of nature. On the other hand, when the specific frequency component signal extracted from the original image is used as the watermark signal 103, the watermark information is detected by cross-correlation with respect to the correlation result of the phase-only correlation between the specific frequency component signal and the input image signal. . In this case, since the correlation value peak can be searched for by performing the correlation calculation while changing the phase, it is robust against a spatial attack.

  The embedded signal generator 11 first sets an embedded position in the frame of the original image 100 for each bit of the watermark information 101. Each bit of the watermark information 101 is embedded at a position in a different frame. The setting of the embedding position of the watermark information 101 is controlled by a phase shift amount by a single or a plurality of digital phase shifters. An embedded signal 106 obtained by multiplying the amplitude of the watermark signal 103 whose phase is shifted by a coefficient corresponding to the value of each bit (“1” or “0”) is generated. The control processing of the amplitude of the frequency component is performed by a single or a plurality of exclusive OR circuits or digital multipliers.

  The scene change detector 13 detects a scene change from the original image 100. An example of a specific method for detecting a scene change point will be described later. In addition, you may utilize well-known detection methods other than the method mentioned later.

  The polarity controller 14 uses the scene change information of the original image 100 to initialize the polarity inversion pattern of the embedded signal 106 when the embedded signal 106 is superimposed on the original image 100. As described above, the accumulated pattern is reset at the position where the scene change occurs. For example, when the polarity of the embedding signal 106 is reversed every three frames, control is performed such as embedding the frame displayed immediately after the scene change with a positive polarity. Alternatively, when the polarity is inverted with a predetermined pattern, adjustment is made so that the closest polarity is switched to the polarity pattern when the embedded signal 106 is embedded in the frame displayed immediately after the scene change. There are various possibilities.

  The embedded signal superimposing unit 12 includes a digital adder, and superimposes the embedded signal 106 on the original image 100 while changing the polarity according to the inversion pattern obtained by the polarity controller 14 to generate an embedded image 102. Here, the embedded signal superimposing unit 12 makes the original image 100 while inverting the polarity of the embedded signal 106 with a predetermined pattern in order to facilitate separation of the image signal component and the digital watermark signal component at the time of detection. Superimposing.

  FIG. 3 is a diagram showing a method of embedding the watermark information 101 in the original image 100 by the digital watermark embedding apparatus 1 of the present embodiment.

  First, the scene change detector 13 detects a scene change in the original image 100 (S11). Next, the polarity controller 14 determines a polarity inversion pattern for superimposing the embedded signal 106 on the original image 100 based on the scene change information of the original image 100 (S12). The embedded signal generator 11 generates an embedded signal 106 based on the watermark signal 103 and the watermark information 101 (S13). Next, the embedded signal superimposing unit 12 superimposes the signal obtained by inverting the polarity of the embedded signal 106 with the polarity inversion pattern obtained by the polarity controller 14 on the original image 100 to obtain the embedded image 102 (S14). ).

  Next, a method in which the embedded signal generator 11 generates the embedded signal 106 will be described.

  FIG. 4 is a diagram illustrating an example of the phase shift. 4 and 5 exemplarily show one line of the two-dimensional image data as an analog signal. In both figures, the horizontal axis indicates the position in the image, and the vertical axis indicates the value of the image signal (for example, the luminance value).

  In FIG. 4, the phase of the watermark signal 103 input to the embedded signal generator 11 is shifted while maintaining its waveform. The embedding position of the watermark information 101 in the two-dimensional image data of the original image 100 is controlled by the phase shift amount. The embedded signal generator 11 uses a predetermined shift amount. The shift amount is set in advance according to the information amount of the watermark information 101 to be embedded, and is given to the embedded signal generator 11. The sign and size of the coefficient to be multiplied by the watermark signal 103 whose phase has been shifted are determined according to the activity representing the complexity of the original image 100. For example, the larger the activity, the larger the coefficient is set. Note that the coefficient differs depending on whether the bit value is “1” or “0”.

  FIG. 5A shows an image signal of the original image 100. FIG. 5B shows the input watermark signal 103. FIGS. 5C and 5D show two different phase shifts generated by the embedded signal generator 11 (two phase shifters) shifting the watermark signal 103 by a predetermined shift amount. It illustrates the signal. Next, the generated phase shift signal is multiplied by a coefficient representing the 0th bit or the 1st bit of the watermark information 101. For example, if the watermark information 101 is “0”, the phase shift signal is multiplied by −1, and if it is “1”, the phase shift signal is multiplied by +1. The reason why the embedded signal generator 11 is the two phase shifters is because an example in which the generated phase shift signal is 2 bits is shown.

  FIGS. 5C and 5D are diagrams showing examples of phase shift signals when the watermark information 101 to be embedded is 2-bit information (1, 1). There are four combinations of 2-bit information: (0, 0) (0, 1) (1, 0) (1, 1). As shown in FIG. 5 (c), as a signal corresponding to the first bit, a watermark signal is set as a + sign corresponding to 1, and a signal whose phase is shifted by α in the X direction is generated. Further, as shown in FIG. 5D, as a signal corresponding to the second bit, a watermark signal is set as a positive sign corresponding to 1, and a signal obtained by shifting a signal shifted in the X direction by 2α is generated. . Since the phase of the image corresponds to the position of the image, the phase shift represents the movement of the position within the screen. In FIGS. 5C and 5D, the signal position differs between the watermark signal and the phase shift signal 1 due to the phase shift, and the position of the leftmost peak of the signal differs. The difference in peak position is caused by the phase being shifted. Note that the positive / negative relationship between the signal and the sign and the phase shift amount are examples, and various modifications may be made. Further, although the phase shift amount is exemplified as a one-dimensional direction shift, it indicates a position shift on a two-dimensional screen.

  After that, the embedded image signal (embedded image 102) generated by the watermark signal superimposing unit 12 adding the phase shift signal multiplied by the bit representation factor to the original image 100 is shown in FIG. Shown in solid line. The broken line in FIG. 5 (e) shows the embedding target image signal before being superimposed, and the phase shift signal shown in FIGS. 5 (c) and 5 (d).

  Next, details of how the scene change detector 13 detects a scene change in the original image 100 will be described.

  FIG. 6 is a diagram showing the configuration of the scene change detector 13. The scene change detector 13 includes a histogram calculator 20, a buffer 21, and a determiner 22.

  The histogram calculator 20 calculates a histogram for each frame of the original image 100. As the histogram, for example, a luminance histogram, an edge histogram, or a direction histogram may be used, or a plurality of histograms may be used. The luminance histogram is a histogram showing the distribution of how many pixels having which luminance value exist in the image. An edge histogram is a histogram that uses the direction of an edge. The edge direction and intensity at each pixel in the image are calculated. It is a histogram obtained by dividing the direction of the edge into a plurality of classes and making the number of pixels assigned to the same class become the frequency of the class. A direction histogram is a histogram using the gradient of pixel values between adjacent pixels. It is a histogram which uses the gradient direction of the brightness | luminance between the adjacent pixels of each pixel, and the magnitude | size of gradient as a frequency.

  Note that an optimal histogram can be used as appropriate in accordance with an assumed application (system to be applied). For example, when a large luminance change is assumed at a point other than the scene change point, the direction histogram and the edge histogram are considered to be effective. In addition, when a large luminance change other than the scene change point is not expected, it is considered effective to use the luminance histogram. Although three types of histograms are described above, other histograms can be used in the same manner.

  The buffer 21 accumulates histograms of a plurality of past frames calculated by the histogram calculator 20.

  The determiner 22 calculates a similarity between the histogram of the current frame obtained by the histogram calculator 20 and the histograms of a plurality of past frames held by the buffer 21. A location where the calculated similarity is small (a location where the amount of histogram change is large) is determined as a scene change.

  In this embodiment, the magnitude of the histogram change amount is used to detect the scene change. However, the scene change may be detected based on the magnitude of the correlation value change amount due to the image correlation such as cross-correlation or phase-only correlation. . In that case, a portion where the amount of change in image correlation is large is detected as a scene change.

  Next, the operation of the determiner 22 will be described.

  FIG. 7 is a diagram illustrating a method for detecting a scene change from a histogram change in the original image 100. 7A to 7C show how the histogram changes between different frames.

  The result of multiplying the histogram of the past frame and the histogram of the current frame for each direction is shown as the similarity calculation result. If the histogram of the past frame and the histogram of the current frame are similar, the multiplication result is a large value. On the other hand, when the similarity between the histogram of the past frame and the histogram of the current frame is small, the multiplication result is a small value. If the multiplication result is a small value, it can be determined that a scene change has occurred in the current frame. FIG. 7 shows an example in which four gradient directions (up, down, left and right) are used as components using a direction histogram. FIG. 7 shows an example in which the histogram frequencies of the past and current frames are values normalized to 1, respectively.

  FIG. 7A shows a comparison between the histogram of the past frame An and the histogram of the current frame An + 1. Since the histogram of the past frame An and the histogram of the current frame An + 1 are similar, the multiplication result is a large value. Therefore, it can be determined that no scene change has occurred in A (An, An + 1).

  FIG. 7B compares the histogram of the past frame Bn and the histogram of the current frame Bn + 1. Since the similarity between the histogram of the past frame Bn and the histogram of the current frame Bn + 1 is low, the multiplication result is a small value. Therefore, it can be determined that a scene change has occurred in B (Bn, Bn + 1).

  FIG. 7C compares the histogram of the past frame Cn with the histogram of the current frame Cn + 1. Since the similarity between the histogram of the past frame Cn and the histogram of the current frame Cn + 1 is low, the multiplication result is a small value. Therefore, it can be determined that a scene change has occurred in C (Cn, Cn + 1).

  Note that although B (Bn, Bn + 1) and C (Cn, Cn + 1) are both determined as scene change points because the multiplication result is small, B (Bn, Bn + 1) and C (Cn, Cn + 1) are shown. And the method of changing the histogram is different. Therefore, by determining in advance a change tendency of a histogram used when determining a scene change, the polarity can be controlled by a specific scene change. By doing so, it is considered that the scene change matching rate between embedding and detection is improved, and more stable synchronization between embedding and detection becomes possible.

  Note that the threshold value of the amount of change in the histogram that is determined to be a scene change may be changed according to the interval (frequency) at which a scene change was detected in the past in the input image. Specifically, the scene change detector 13 is provided with a counter for calculating the frequency of past scene changes of the corresponding content. When there are few scene changes, the number of scene changes is adjusted by lowering the scene change determination threshold. Similarly, when there are many scene changes, the number of scene changes is adjusted by raising the scene change determination threshold. By changing the threshold value, an appropriate scene change corresponding to the content can be detected.

  It is also conceivable to effectively cope with various time lags by embedding at different positions in the image with different cumulative patterns.

(Digital watermark detection device)
FIG. 8 is a diagram showing the configuration of the digital watermark detection apparatus 11 of the present embodiment. The digital watermark detection apparatus 11 includes a scene change detector 44, a polarity controller 45, a correlation calculator 41, a correlation accumulator 42, and a watermark information estimator 43. The digital watermark detection apparatus 11 separates an embedded signal component from the input image 104 (an image in which the digital watermark is embedded), and detects the watermark information 105.

  The scene change detector 44 detects a scene change in the input image 104. A scene change in the input image 104 is detected using the same method as the scene change detector 13 of FIG. Shooting a screen projection on a film projector with a video camera causes a time lag in embedding and detection, but this is due to deviations in the projector's rotational speed (e.g., typically less than 0.2% in the specification). to cause. Even when the frequency at which scene changes occur varies depending on the content, it is considered that a scene change will occur to the extent that a temporal shift due to the rotational speed deviation does not accumulate. Note that it is preferable to use the same detection method as the scene change detection method used when embedding a digital watermark and the scene change detection method used when detecting a digital watermark.

  The correlation calculator 41 calculates the degree of correlation between the input image 104 and the watermark signal 103 (hereinafter referred to as a cross correlation value). Here, as a correlation calculation method, a general correlation method such as cross-correlation or phase-only correlation is used.

  The polarity controller 45 determines a time-series pattern for inverting the polarity during accumulation from the scene change information using the same method as the polarity controller 14 of FIG. As shown in FIG. 1, the polarity controller 45 obtains a pattern for initializing the polarity inversion pattern in the frame immediately after the scene change in the detected input image 104. According to the obtained pattern, the correlation accumulator 42 cumulatively adds the cross-correlation values of each frame. In addition to initializing the accumulated pattern, there are various cases such as adjustment to be the closest polarity switching pattern at the scene change position. Synchronization can be achieved by determining the accumulated pattern by a method corresponding to the polarity inversion pattern determination method performed by the embedding device 1.

  The correlation accumulator 42 cumulatively adds the cross-correlation values between the input image 104 and the watermark signal 103 based on the polarity inversion pattern determined by the polarity controller 45.

  The digital watermark estimator 43 estimates the watermark information 105 from the signal cumulatively added by the correlation accumulator 42.

  When embedding and embedding with different patterns at different positions in the image space, detection is performed with all possible accumulated patterns, and the accumulated pattern that is most likely to be estimated by the digital watermark estimator 43 is used. The result of is the final estimation result.

  FIG. 9 is a diagram illustrating the operation of the digital watermark detection apparatus 11 of the present embodiment.

  First, the scene change detector 44 detects a scene change in the input image 104 (S21). Next, the polarity controller 14 determines an inversion pattern of a signal when accumulating the correlation values based on the detected scene change (S22). The correlation value calculator 41 calculates a cross-correlation value between the image signal of the input image 104 and the watermark signal 103 (S23). Next, the correlation value accumulator 42 performs cumulative addition while inverting the cross-correlation value with the polarity inversion pattern obtained in S22 (S24). Next, the watermark information estimator 43 detects the watermark information 105 embedded in the input image 104 from the cumulatively added signal, outputs it, and ends the processing (S25). A detailed method for detecting the watermark information 105 will be described later.

  FIG. 10 is a diagram illustrating an example of the configuration of the correlation calculator 41.

  The correlation calculator 41 includes a first orthogonal transformer 50, a combined amplitude adjuster 51, and a second orthogonal transformer 52.

  The first orthogonal transformer 50 performs orthogonal transformation processing such as orthogonal transformation on the watermark signal 103. Similarly, the first orthogonal transformer 50 performs orthogonal transformation processing such as orthogonal transformation on the input image 104.

  The synthesizer / amplitude adjuster 51 complex synthesizes two signals after orthogonal transformation.

  The second orthogonal transformer 52 performs orthogonal transformation (inverse orthogonal transformation) on the composite signal after complex synthesis. Here, the orthogonal transformation needs to be paired with the transformation in the first orthogonal transformation. When the first orthogonal transformer 50 uses the Fourier transformation, the second orthogonal transformer 52 performs the Fourier transformation or the inverse. Perform Fourier transform.

  FIG. 11 is a diagram illustrating the relationship between the cross-correlation value and the phase shift amount. As shown in FIG. 11, the peak of the cross-correlation value appears at a certain phase shift amount position. The polarity (positive / negative) of this peak represents the watermark information 105. The digital watermark estimator 43 controls the phase shift amount continuously or stepwise, searches for the peak of the cross-correlation value that is output along with it, and estimates and detects the watermark information 105 from the polarity of the searched peak. To do. The peak of the cross-correlation value takes either positive or negative value according to the watermark information 105. FIG. 11 illustrates the case where the peak is positive, and the watermark information 105 is “1”. On the other hand, when the peak is negative, the watermark information 105 is determined to be “0”.

  The above correlation method is called a “phase-only correlation method” when the amplitude component after the first orthogonal transform is compressed to give a specific gravity to the phase (limited), and amplitude information is used. Unlike the conventional two-dimensional correlation method and feature extraction method, it has a feature that it is resistant to disturbances and does not easily cause a large error.

  FIG. 12 is a diagram for explaining the operation of the digital watermark detection apparatus 11. Since one line of the signal of the two-dimensional image data is exemplarily shown, the one-dimensional signal is drawn.

  FIG. 12A shows an image signal of the embedded image 102. FIG. 12B shows the input watermark signal. FIGS. 12C and 12D are phase shift signals obtained by phase-shifting the watermark signal 103 by a predetermined shift amount. A correlation value between the input image signal and the phase shift signal is obtained, and the watermark information 105 is determined from the peak of the correlation value. For example, if the correlation value peak is positive, the watermark information 105 is determined to be +1 (“1”), and if the correlation value peak is negative, the watermark information 105 is determined to be −1 (“0”). The The watermark signal 43 detects the peak of the correlation value while controlling the phase shift amount. The watermark information 105 is estimated from the peak position.

  FIG. 13 is a diagram showing the correlation value peak search and watermark information detection operations when the watermark information 105 is (1, 1). As shown in FIG. 13, when there are two positive peaks of the correlation value other than the origin (a point where the phase shift amount is zero), (1, 1) as the watermark information 105 is detected.

  FIG. 14 is a diagram showing a correlation value peak search and watermark information detection operation when the watermark information 105 is (1, −1). As shown in FIG. 14, since the positive peak of the correlation value exists near the origin and the negative peak exists far from the positive peak from the origin, the watermark information 105 is (1, −1). Detected.

  When an embedded image is created by adding a watermark signal to an original image, one of each line, each line, each field, each field, each field, each frame, and each frame, or an appropriate one of these A method of inverting the polarity of the phase shift signal in combination or a method of changing the phase shift amount may be used.

[Second Embodiment]
The digital watermark embedding / detection device of this embodiment is different from the digital watermark embedding / detection device of the first embodiment in that a watermark signal is generated from an original image / input image.

(Digital watermark embedding device)
FIG. 15 is a diagram showing a configuration of the digital watermark embedding apparatus 2 of the present embodiment. In the digital watermark embedding apparatus 1 of FIG. 2, an example in which the watermark signal 103 is input has been described. The digital watermark embedding device 2 extracts a specific frequency component from the image signal of the original image 100 and uses the specific frequency component as a watermark signal. The other configurations perform the same operations as those shown in FIG.

  The watermark signal generator 10 generates a watermark signal 103 from the original image 100.

  FIG. 16 is a diagram illustrating a configuration of the watermark signal generator 10. The watermark signal generator 10 includes an enlargement / reduction unit 30 and a frequency component extractor 31.

  The enlargement / reduction unit 30 reduces (or enlarges) the original image 100. Here, the enlargement / reduction unit 30 does not have to perform a signal obtained by multiplying the original image 100 by 1, that is, enlargement or reduction processing. By using a plurality of enlargement / reduction ratios of the enlargement / reduction unit 30, it is possible to embed with different accumulated patterns at the same position in the image.

  The frequency component extractor 31 extracts a specific frequency component (hereinafter referred to as “specific frequency component”) from the reduced (or enlarged) signal of the original image 100. The frequency component extractor 31 has a frequency domain digital filter for extracting a specific frequency component, for example, a low pass filter or a high pass filter having a predetermined cutoff frequency, or a band pass filter having a predetermined pass band center frequency. . Here, the filter for extracting the specific frequency component to be used needs to extract a specific frequency component equivalent to that at the time of detection. The frequency component extractor 31 may extract all frequency components of the signal of the original image 100 as specific frequency components.

  The specific frequency component signal extracted by the frequency component extractor 31 may exist in a plurality of channels. In this case, the specific frequency component signals of the plurality of channels are superimposed on the original image 100 in the embedded signal superimposing unit 12. .

(Digital watermark detection device)
FIG. 17 is a diagram illustrating a configuration of the digital watermark embedding device 12 according to the present embodiment. The digital watermark embedding device 12 generates a watermark signal 103 from the input image 104 in the same manner as the digital watermark embedding device 12.

  The watermark signal generator 40 generates a watermark signal 103 from the input image 104.

  FIG. 18 is a diagram illustrating a configuration of the watermark signal generator 40. The watermark signal generator 40 includes an embedded enlargement / reduction unit 60 and a frequency component extractor 61.

  The embedded enlargement / reduction unit 60 reduces (or enlarges) the input image 104. Here, the embedded enlargement / reduction unit 60 does not have to perform a signal obtained by multiplying the original image 100 by 1, that is, enlargement or reduction processing.

  The frequency component extractor 61 extracts a specific frequency component (hereinafter referred to as “specific frequency component”) from the reduced (or enlarged) signal of the input image 104. The frequency component extractor 61 has a frequency domain digital filter for extracting a specific frequency component, for example, a low pass filter or a high pass filter having a predetermined cutoff frequency, or a band pass filter having a predetermined pass band center frequency. . Here, the filter for extracting the specific frequency component to be used needs to extract a specific frequency component equivalent to that at the time of embedding. The frequency component extractor 61 may extract all frequency components of the signal of the input image 104 as specific frequency components.

  As described above, the watermark signal is created using the specific frequency component signal extracted from the image signal of the original image in which the watermark information is embedded, and the embedding method is geometric transformation (image cutout, physical scale conversion such as scaling). Etc.)

[Third Embodiment]
The present embodiment is different from the first embodiment in that a histogram is calculated from an image obtained by reducing an original image / input image (with a reduced resolution).

(Digital watermark embedding device)
FIG. 19 is a diagram showing a configuration of the histogram calculator 13 of the present embodiment. The scene change detector 13 further includes a reduced image generator 24.

  The reduced image generator 24 generates an image obtained by reducing the original image 100.

  The histogram calculator 21 calculates a histogram of the reduced image.

(Digital watermark detection device)
FIG. 20 is a diagram illustrating a configuration of the histogram calculator 44 of the present embodiment. The scene change detector 44 further includes a reduced image generator 74.

  The reduced image generator 74 generates an image obtained by reducing the original image 100.

  The histogram calculator 20 calculates a reduced image histogram of the original image 100 generated by the reduced image generator 74.

  Note that there are various ways of creating a reduced image, such as when a digital filter is used or when an image is thinned out in a horizontal direction or a vertical direction.

  In addition, the calculation of the histogram may be performed not on the entire screen of the image but on the center of the screen. In this way, not only the processing amount is reduced, but also the matching rate of scene change positions in embedding and detection is expected to be improved. This is because there is a high possibility that much information of the original image is included in the center of the screen of the original image 100. For example, when a captured image of a screening screen is input to the detection device, there may be components other than the original image such as a dark screen around the screen and a human shadow in the theater. The part is less likely to capture components other than this type of original image, and it is considered that scene change detection can be performed more accurately.

  According to the digital watermark embedding / detecting apparatus of this embodiment, since the histogram is calculated from the reduced image, the processing amount can be further reduced.

  Note that these original image and input image reduction processes are performed so that the image sizes at the time of histogram calculation in the digital watermark embedding device and the detection device are equal, so that more stable embedding and detection synchronization can be achieved. It is done. For example, if the histogram calculation at the time of detection is 720 × 480 (pixel unit; hereinafter omitted) resolution which is SDTV, even when the resolution of the embedded image is HDTV (1920 × 1080), It is considered that it is better to perform the histogram calculation at SDTV (720 × 480) resolution.

[Fourth Embodiment]
The digital watermark embedding / detecting apparatus according to the present embodiment is different from the digital watermark embedding / detecting apparatus according to the first to third embodiments in that the polarity controller 45 determines the polarity inversion pattern.

  The polarity controller 14 of the digital watermark embedding device and the polarity controller 45 of the digital watermark detection device according to the present embodiment calculate the occurrence frequency of scene changes in the original image and the input image. On the contrary, if the frequency of occurrence of scene changes is high, the polarity inversion cycle is shortened.

  FIG. 21 shows an example in which the synchronization rate of embedding and detection for each polarity inversion pattern is compared when a frame rate deviation corresponding to a rotation speed deviation of 0.2% of the projector occurs (one frame in 50/3 seconds). . The vertical axis represents the synchronization rate of detection and embedding polarity. The horizontal axis represents the frequency of occurrence of a scene change during 1 second. Pattern 1 shows a pattern in which the polarity is inverted every frame, pattern 2 every 2 frames, pattern 3 every 3 frames, pattern 4 every 4 frames, and pattern 5 every 5 frames.

  It can be seen that the shorter the polarity reversal period, the sharper the synchronization rate decreases as the accumulated time becomes longer, and conversely, the longer the polarity reversal period, the slower the synchronization decrease rate. Therefore, if the frequency of past scene changes of the corresponding content is calculated and the number of scene changes is small, the error due to frame deviation seems to increase, so the polarity inversion period is lengthened and conversely there are many scene changes. It is conceivable to shorten the polarity inversion period. This is expected to improve robustness against frame deviation.

  According to the digital watermark embedding / detecting apparatus of the present embodiment, it is possible to reduce the influence of content by changing the embedding / detecting pattern and the scene change determination threshold according to the scene change frequency. In this case, it is necessary to change the embedding / detection pattern and the scene change determination threshold in the same manner between embedding and detection.

  According to the digital watermark embedding / detecting device of each of the above embodiments, when the information (codec information or the like) of the original image of the input image cannot be used when detecting the digital watermark (such as an image taken by a video camera). Even if it exists, it becomes possible to initialize a cumulative pattern regularly by embedding and detection. As a result, even when a time lag occurs between the original image and the input image, detection can be performed with stable accuracy.

  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.

FIG. 6 is a diagram for explaining a synchronization example of digital watermark embedding / detection using scene change information. 1 is a diagram illustrating a configuration of a digital watermark embedding device according to a first embodiment. 5 is a flowchart illustrating a digital watermark embedding method according to the first embodiment. The figure explaining the phase shift of a watermark signal. The wave form diagram for demonstrating operation | movement of a digital watermark embedding apparatus. The figure which shows the structure of a scene change detector. The figure explaining a histogram change amount. FIG. 3 is a diagram illustrating a configuration of a digital watermark detector according to the first embodiment. 5 is a flowchart illustrating a digital watermark detection method according to the first embodiment. The figure which shows the structure of a correlation calculator. The figure which shows the relationship between a cross correlation value and phase shift amount. The wave form diagram for demonstrating operation | movement of a digital watermark detection apparatus. The figure which shows the operation | movement of the peak search of a correlation value in case watermark information is (1, 1) and watermark information detection. The figure which shows the operation | movement of the peak search of a correlation value and watermark information detection in case watermark information is (1, -1). The figure which shows the electronic watermark embedding apparatus structure of 2nd Embodiment. The figure which shows the structure of a watermark signal generator. The figure which shows the structure of the digital watermark detection apparatus of 2nd Embodiment. The figure which shows the structure of the watermark signal generator of the digital watermark detection apparatus of 2nd Embodiment. The figure which shows the structure of the scene change detector of 3rd Embodiment. The figure which shows the structure of the scene change detector of 3rd Embodiment. The figure which shows the example which compared the synchronous ratio of embedding and detection when a frame deviation arises.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Original image 101 Watermark information 102 Embedded image 103 Watermark signal 104 Input image 105 Watermark information 106 Embedded signal 1 and 2 Digital watermark embedding apparatus 10 Watermark signal generator 11 Embedded signal generator 12 Embedded signal superimposer 13 Scene change Detector 14 Polarity controller 20 Histogram calculator 21 Buffer 22 Judgment unit 11, 12 Digital watermark detection device 30 Embedding enlargement / reduction unit 31 Frequency component extractor 40 Watermark signal generator 41 Correlation calculator 42 Correlation accumulator 43 Watermark information estimation Device 44 scene change detector 45 polarity controller 50 first orthogonal transformer 51 synthesis / amplitude adjuster 52 second orthogonal transformer 60 embedded enlargement / reduction unit 61 frequency component extractor

Claims (17)

In a digital watermark embedding device that generates an embedded image in which watermark information is embedded in an original image,
An embedded signal generator for generating an embedded signal according to the watermark information;
A scene change detector for detecting a scene change frame in which the scene change occurs before the display timing of the frame in the original image;
A polarity controller that obtains a time-series polarity change pattern when the embedded signal is embedded in the original image, with the scene change frame at the beginning of an arbitrary pattern period;
An embedded signal superimposer that obtains the embedded image by changing the polarity of the embedded signal for each frame of the original image according to the polarity change pattern, and combining the original image with the original image;
An electronic watermark embedding device comprising:
The polarity controller sets a pattern that repeats a predetermined first pattern as the arbitrary pattern, and assigns the first polarity change pattern of the first pattern to the scene change frame to assign the polarity. 2. The digital watermark embedding apparatus according to claim 1, wherein a change pattern is obtained.
The polarity controller uses the change timing of the polarity that is scheduled to be assigned to the scene change frame when the polarity is determined for each frame according to the arbitrary pattern and the closest time-series position as the starting point, as the starting point. The digital watermark embedding apparatus according to claim 1, wherein a polarity change pattern is obtained.
The polarity controller obtains the occurrence frequency of the scene change of the original image detected by the scene change detector, and when the occurrence frequency is low, a pattern whose polarity is inverted in a long cycle is used as the arbitrary pattern. 4. The digital watermark according to claim 2, wherein the polarity change pattern is obtained, and when the occurrence frequency is high, the polarity change pattern is obtained by using a pattern whose polarity is inverted in a short cycle as the arbitrary pattern. Implantation device.
A histogram calculator for calculating a histogram of the frame of the original image;
The scene change detector detects the scene change frame when the amount of change in the histogram exceeds a predetermined reference before and after the display timing of the frame. The electronic watermark embedding apparatus according to claim 1.
6. The electronic apparatus according to claim 5, wherein the scene change detector obtains the occurrence frequency of the scene change of the original image detected by the scene change detector and changes the reference according to the occurrence frequency. Watermark embedding device.
7. The digital watermark according to claim 6, wherein when the occurrence frequency is low, a threshold value that is a small value is used as the reference, and when the occurrence frequency is high, a threshold value that is a large value is used as the reference. Implantation device.
In an electronic watermark detection apparatus for detecting the watermark information from an input image in which watermark information is embedded in an original image,
An embedded signal generator for generating an embedded signal according to the watermark information;
A scene change detector for detecting a scene change frame in which a scene change occurs before the display timing of the frame in the input image;
Correlation calculating means for obtaining a correlation value indicating the degree of correlation between the input image and the watermark signal;
A polarity controller that obtains an arbitrary pattern starting from the scene change frame, a time-series polarity change pattern of polarity when the embedded signal is embedded in the input image signal in an original image,
A superimposer that cumulatively adds the correlation value according to the polarity change pattern;
An estimator for estimating the watermark information from the signal cumulatively added by the superimposer;
An electronic watermark detection apparatus comprising:
The polarity controller obtains the polarity change pattern by assigning a first polarity change pattern of the first pattern to the scene change frame as a pattern that repeats the first pattern as the arbitrary pattern. The digital watermark detection apparatus according to claim 8.
The polarity controller uses the change timing of the polarity that is scheduled to be assigned to the scene change frame when the polarity is determined for each frame according to the arbitrary pattern and the closest time-series position as the starting point, as the starting point. 9. The digital watermark detection apparatus according to claim 8, wherein a polarity change pattern is obtained.
The polarity controller obtains the occurrence frequency of the scene change of the input image detected by the scene change detector, and when the occurrence frequency is low, a pattern whose polarity is inverted in a long cycle is set as the arbitrary pattern. 11. The polarity change pattern according to claim 9, wherein the polarity change pattern is obtained, and when the occurrence frequency is high, the polarity change pattern is obtained by using a pattern whose polarity is inverted in a short cycle as the arbitrary pattern. The electronic watermark detection apparatus according to the item.
A histogram calculator for calculating a histogram of the frame of the original image;
12. The scene change detector detects the scene change frame when the amount of change in the histogram exceeds a predetermined reference before and after the display timing of the frame. The electronic watermark embedding apparatus according to claim 1.
The said scene change detector calculates | requires the occurrence frequency of the said scene change of the said input image which the said scene change detector detected, and changes the said reference | standard according to the said occurrence frequency. Digital watermark detection device.
14. The digital watermark detection apparatus according to claim 13, wherein when the occurrence frequency is low, the reference threshold value is reduced, and when the occurrence frequency is high, the reference threshold value is increased. .
A histogram calculator for calculating a histogram of the frame of the original image;
15. The scene change detector detects the scene change frame when the amount of change in the histogram exceeds a predetermined reference before and after the display timing of the frame. The electronic watermark embedding apparatus according to claim 1.
In a digital watermark embedding method for generating an embedded image in which watermark information is embedded in an original image,
An embedded signal generating step for generating an embedded signal in accordance with the watermark information;
A scene change detection step for detecting a scene change frame in which a scene change occurs before the display timing of the frame in the original image;
A polarity control step for obtaining a time-series polarity change pattern when embedding the embedded signal in the original image, with the scene change frame at the beginning of an arbitrary pattern cycle;
An embedding signal superimposing step of obtaining the embedded image by changing the polarity of the embedding signal for each frame of the original image according to the polarity change pattern, and combining the original image with the original image,
An electronic watermark embedding method comprising:
In a digital watermark detection method for detecting the watermark information from an input image in which watermark information is embedded in an original image,
An embedded signal generating step for generating an embedded signal in accordance with the watermark information;
A scene change detection step for detecting a scene change frame in which a scene change occurs before the display timing of the frame in the input image;
Correlation calculating means for obtaining a correlation value indicating the degree of correlation between the input image and the watermark signal;
A polarity control step for obtaining an arbitrary pattern starting from the scene change frame, a time-series polarity change pattern of polarity when the embedded signal is embedded in the original image as the input image signal,
A superimposing step of accumulating the correlation value according to the polarity change pattern;
An estimation step of estimating the watermark information from the signal cumulatively added in the superposition step;
An electronic watermark detection method comprising:

Images