JP2008235964A - Information processing apparatus and method, and program, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an apparatus and method for highly correctly extracting information embedded in an image signal etc., and provide an apparatus and method of embedding information to enable such highly correct information extraction. <P>SOLUTION: The information extracting apparatus extracts information from an information-embedded container signal such as an image signal having information embedded per predetermined length segment. First, the apparatus measures mutual correlation as similarity between a local waveform and an embedded waveform of the information-embedded container signal (similarity measuring block 200). The apparatus estimates an embedded phase range of the information in a segment from the distribution of absolute values of mutual correlation in the segment (phase synchronizing block 201). For each segment, the apparatus extracts embedded information from decision based on comparison between the maximum or minimum value in mutual correlation in the estimated phase range and a predetermined value (information extracting block 202). <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an apparatus and method for performing processing for embedding information such as copyright holder information in an image signal or the like and extracting the embedded information using an electronic watermark technique.

  In recent years, the creation of multimedia digital contents has become extremely easy, and the spread of the Internet has made it possible to publish these digital contents. Despite the excellent quality of digital content, there is always the danger of duplicating and using other people's copyrighted work, because the quality is not deteriorated no matter how much it is copied and it is easy to handle. .

  In connection with the distribution of digital content, the copyright owner information and confidential data are illegally leaked in order to deter unauthorized use of digital content as evidence to take legal procedures when a third party misuses it without the author's permission. In order to suppress this, digital watermarks are attracting attention as a technique for embedding accessor information in digital content. Digital watermarking refers to a technique that can be perceived by secret information when a specific operation is performed on digital content, although human beings cannot normally perceive it. In particular, in recent years, digital watermarking using a mechanism of HVS (Human Visual System) has been adopted.

  There are various methods for embedding a digital watermark, but an image signal or audio signal (hereinafter referred to as a container signal) before information is embedded, or an image signal or audio signal in which information is embedded (hereinafter, information embedded) This method is roughly divided into a method for converting a signal into a frequency space and processing it in real space, and each method is further divided into a method using quantization and a method using correlation. Each method has advantages and disadvantages.

  The present invention relates to a method of processing a container signal or a container signal with embedded information in real space and utilizing correlation. In general, this method is more effective for artificial image signals representing characters and figures than natural signals such as photographs and sounds. For example, Patent Documents 1, 2, and 3 can be cited as prior art documents of this type.

Special table 10-513324 gazette Japanese Patent No. 3576993 Japanese Patent No. 3592545

  As a conventional example, the method described in Patent Document 1 will be described. This is because the image is divided into 8 × 8 pixel sections (blocks), and if there is an edge portion in each block, information is embedded by transforming that portion into a predetermined embedding shape (convex shape or concave shape). In this method, the embedded information is extracted by taking the correlation between the embedded image and the embedded shape for each block. The principle of this method will be described with reference to FIG. Although the image is a two-dimensional signal, it will be described as a one-dimensional signal for simplicity.

  In FIG. 11A, the horizontal axis represents the pixel (sample) position, and the vertical axis represents the cross-correlation value between the predetermined embedded waveform and the information-embedded container signal waveform. The information-embedded container signal waveform is divided at regular intervals (here, four sections are shown). For each section, cross-correlation values are obtained at four locations (phases). If the embedding information is “0”, it approaches a predetermined embedding waveform (for example, a convex waveform), and if the embedding information is “1”, it approaches an inversion waveform (concave waveform) of the predetermined embedding waveform. Since the original container signal waveform is deformed, a positive large cross-correlation value is obtained if the embedding information is “0” and a large negative correlation is obtained if the embedding information is “1”. Correlation value comes out.

  In FIG. 11A, since the deformation is performed at the phase of the white circle, the absolute value of the cross-correlation is large at that phase. In this example, “0”, “1”, “1”, and “0” are embedded in order from the head section. However, there is no guarantee that the cross-correlation value at the phase (shaded circle) where the deformation is not performed is a value close to zero. There may be a case where a relatively large cross-correlation value such as “E” appears. This is a case where the original waveform resembles the embedded waveform at that position (phase).

  For example, when information is embedded in image data, the image data in which the information is embedded by the information embedding device 1 is output to a recording medium by the output device 2 as shown in FIG. There is a scene in which the recording medium P is read by an input device 3 such as a scanner and input as image data to the information extraction device 4 to extract information embedded in the image data. Since noise is added to the image data in the output process by the output device 2 and the reading process by the input device 3, the pixel value of the image data input to the information extraction device 4 varies. Accordingly, the correlation value between the image data and the embedded waveform also varies. Further, the image data also undergoes parallel movement during the output process by the output device 2 and the reading process by the input device 3, so that a phase shift also occurs.

  When information is embedded in the same phase in each section, when extracting embedded information, if embedded image data is divided at equal intervals, even if a phase shift occurs, the embedded position Always enter the section. Therefore, if the interval setting at equal intervals is performed with an appropriate phase and the peak of the correlation value can be detected from each interval, the embedded information can be extracted.

  However, if the absolute value of the correlation values of B and C in FIG. 11 becomes smaller than E due to the influence of noise, the phase E that is not embedded is detected in any section setting, and information is erroneously extracted. Resulting in. For example, when section settings such as (b) to (e) in FIG. 11 are performed, incorrect information is extracted in the section marked with X.

  Therefore, there is a method of increasing the probability that correct information can be extracted by repeatedly embedding the same information in a plurality of places and taking a majority vote of the information extracted by the information extracting device. However, even if this method is adopted, the information extraction apparatus often detects a phase in which no information is embedded, and as a result, there is a problem that the reliability of the extracted information cannot be ensured.

  There is also a problem caused by embedding information at the same phase in the section. That is, if the container signal waveform is significantly different from the embedded waveform in the embedded phase, the quality of the image data is greatly degraded if the container signal waveform is modified so as to approach the embedded waveform. Therefore, a method is adopted in which the embedding at such a phase is abandoned and correct information can be extracted from another section by repeatedly embedding the same information in a plurality of places. However, in this method, if the number of sections for which embedding is abandoned increases, the number of extractions of information decreases, so the reliability of majority vote decreases, and the reliability of extracted information cannot be ensured.

  Therefore, an object of the present invention is to improve the above-described problems related to embedding information in an image signal and extracting embedded information.

An information processing apparatus according to the invention of claim 1
The signal before the information is embedded (hereinafter referred to as a container signal) is divided into sections of length T2, and the section is approximated to a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section. An information processing apparatus that embeds information in the container signal by modifying the waveform of the container signal,
From the container signal, while sequentially shifting the cut-out position, cut out a local waveform having a length T1, and measure a similarity measure for measuring the cross-correlation between the cut-out local waveform and the embedded waveform;
For each of the sections, a cross-correlation condition measurement is performed on the plurality of embedded phase candidates in the section by the similarity measurement unit, and one phase that satisfies the determination condition in the plurality of embedded phase candidates is embedded. Phase selection means for selecting as a phase;
Signal modification means for transforming the waveform of the container signal in the embedding phase selected by the phase selection means so as to approach the embedding waveform or its inverted waveform is provided for each section.

An information processing apparatus according to a second aspect of the present invention provides:
The signal before the information is embedded (hereinafter referred to as a container signal) is divided into sections of length T2, and the section is approximated to a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section. An information processing apparatus that embeds information in the container signal by modifying the waveform of the container signal,
From the container signal, the local waveform of length T1 is cut out while sequentially shifting the cut-out position, and the difference between the cut-out local waveform and the embedded waveform (denoted as the first difference) and the inverted waveform of the embedded waveform, Similarity measurement means for measuring the difference (denoted as a second difference) of
For each of the sections, condition determination of the first difference or the second difference measured by the similarity measurement unit is performed for a plurality of embedded phase candidates in the section, and the determination condition in the plurality of embedded phase candidates is satisfied. Phase selection means for selecting one of the phases as an embedding phase;
Signal modification means for transforming the waveform of the container signal in the embedding phase selected by the phase selection means so as to approach the embedding waveform or its inverted waveform is provided for each section.

  According to a third aspect of the present invention, in the information processing apparatus according to the first or second aspect of the present invention, a priority is assigned to the plurality of embedded phase candidates, and the phase selection means A phase having a high priority is preferentially selected as an embedding phase.

An information processing apparatus according to claim 4 is provided.
A signal in which information is embedded by changing the waveform so as to approach a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section of length T2 (hereinafter referred to as an information embedded container signal). ) To extract information embedded for each section of length T2,
A similarity measuring unit that cuts out a local waveform of length T1 while sequentially shifting the cut-out position from the information-embedded container signal, and measures cross-correlation between the cut-out local waveform and the embedded waveform;
From the cross-correlation measured by the similarity measurer, obtain a distribution of the absolute value in the interval, and estimate the information embedded phase range in the interval from the distribution;
For each section, the information embedded phase range estimated by the phase synchronization means is embedded in the section by comparing the maximum or minimum value of the cross-correlation measured by the similarity measurement means with a predetermined value. And an information extracting means for extracting information.

An information processing apparatus according to claim 5 is provided.
A signal in which information is embedded by changing the waveform so as to approach a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section of length T2 (hereinafter referred to as an information embedded container signal). ) To extract information embedded for each section of length T2,
A local waveform having a length of T1 is cut out from the information-embedded container signal while sequentially shifting the cut-out position. The difference between the cut-out local waveform and the embedded waveform (denoted as a first difference) and the inversion of the embedded waveform Similarity measuring means for measuring a difference from a waveform (denoted as a second difference);
A phase synchronization means for obtaining a distribution of the first and second differences measured by the similarity measurement means in the section and estimating an information embedding phase range in the section from the distribution;
For each section, the information embedded phase range estimated by the phase synchronization means is embedded in the section by comparing the minimum value of the first or second difference measured by the similarity measurement means with a predetermined value. And an information extracting means for extracting the extracted information.

An information processing method according to the invention of claim 6 comprises:
The signal before the information is embedded (hereinafter referred to as a container signal) is divided into sections of length T2, and the section is approximated to a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section. An information processing method for embedding information in the container signal by modifying the waveform of the container signal,
From the container signal, while sequentially shifting the cut-out position, a local waveform having a length T1 is cut out, and a similarity measurement step of measuring the cross-correlation between the cut-out local waveform and the embedded waveform;
For each of the sections, the cross-correlation condition measurement performed by the similarity measurement step is performed on a plurality of embedded phase candidates in the section, and one phase that satisfies the determination condition among the plurality of embedded phase candidates is embedded. A phase selection step to select as the phase;
And a signal transformation step for transforming the waveform of the container signal in the embedding phase selected by the phase selection step so as to approach the embedding waveform or its inverted waveform for each section.

An information processing method according to the invention of claim 7 comprises:
A signal before embedding information (hereinafter referred to as a container signal) is divided into sections of length T2, and the section is approximated to a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section. An information processing method for embedding information in the container signal by modifying the waveform of the container signal,
From the container signal, the local waveform of length T1 is cut out while sequentially shifting the cut-out position, and the difference between the cut-out local waveform and the embedded waveform (denoted as the first difference) and the inverted waveform of the embedded waveform, A similarity measurement step of measuring a difference (denoted as a second difference) of
For each of the sections, condition determination of the first difference or the second difference measured by the similarity measurement step is performed for a plurality of embedded phase candidates in the section, and the determination condition in the plurality of embedded phase candidates is satisfied. A phase selection step of selecting the generated one phase as an embedding phase;
And a signal transformation step for transforming the waveform of the container signal in the embedding phase selected by the phase selection step so as to approach the embedding waveform or its inverted waveform for each section.

An information processing method according to the invention described in claim 8 includes:
A signal in which information is embedded by changing the waveform so as to approach a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section of length T2 (hereinafter referred to as an information embedded container signal). ) To extract information embedded for each section of length T2,
A similarity measurement step of cutting out a local waveform of length T1 while sequentially shifting the cut-out position from the information-embedded container signal, and measuring a cross-correlation between the cut-out local waveform and the embedded waveform;
From the cross-correlation measured by the similarity measurement step, obtain a distribution of the absolute value in the section, and estimate the information embedded phase range in the section from the distribution,
For each section, the information embedded phase range estimated by the phase synchronization process is embedded in the section by comparing the maximum value or the minimum value of the cross-correlation measured by the similarity measurement process with a predetermined value. And an information extracting step for extracting information.

An information processing method according to the invention of claim 9 comprises:
A signal in which information is embedded by changing the waveform so as to approach a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section of length T2 (hereinafter referred to as an information embedded container signal). ) To extract information embedded for each section of length T2,
A local waveform having a length of T1 is cut out from the information-embedded container signal while sequentially shifting the cut-out position. The difference between the cut-out local waveform and the embedded waveform (denoted as a first difference) and the inversion of the embedded waveform A similarity measurement step for measuring a difference from a waveform (denoted as a second difference);
A phase synchronization step of obtaining a distribution in the section of the first and second differences measured by the similarity measurement step and estimating an information embedding phase range in the section from the distribution;
For each section, the information embedding phase range estimated by the phase synchronization step is embedded in the section by comparing the minimum value of the first or second difference measured by the similarity measurement step with a predetermined value. And an information extracting step for extracting the extracted information.

  According to the information processing apparatus or method according to the first, second, third, sixth, and seventh aspects of the present invention, the container signal waveform is an embedded waveform or its inverse among a plurality of embedded phase candidates in each section of the container signal. Since it is possible to embed information by selecting a phase close to the waveform, compared to a method of embedding only one specific phase in the section, the section for giving up embedding is reduced, and the reliability of the extracted information on the information extraction side is reduced. Can be improved. Also, it is possible to avoid information embedding at a phase where the difference between the embedded waveform or its inverted waveform and the waveform is large, and to suppress deterioration of the container signal due to embedding.

  According to the information processing apparatus or method according to the inventions of claims 4, 5, 8, and 9, the embedded phase range in the section is estimated from the cross-correlation or difference distribution in the section, and the cross-correlation in the range or Embedded information is extracted based on the difference. Therefore, an information-embedded container signal whose information embedding phase fluctuates, such as a signal in which information is embedded by the information processing apparatus or method according to the inventions of claims 1, 2, 3, 6, 7, Even for information-embedded container signals to which noise has been added in the conversion process as described with reference to FIG. 12, it is possible to extract information with high reliability by avoiding extraction of information at an inappropriate phase.

  The outline of the embodiment of the present invention will be described first with reference to FIGS.

[Overview]
FIG. 1 is a block diagram showing the configuration of the information embedding device. In the information embedding apparatus, the similarity between the waveform of the container signal (for example, an image signal before embedding information) and a predetermined waveform (embedded waveform) used for embedding information is measured (similarity measurement block 100). . Based on the measured similarity, a phase in which information is embedded is selected for each section of the container signal (phase selection block 101). Then, signal transformation processing for embedding information in the phase selected for each section of the container signal is performed (signal transformation block 102). For example, when the information “0” is embedded in the container signal, the signal modification process is performed so that the waveform of the container signal is brought close to a predetermined embedded waveform (for example, the waveform of the container signal and the predetermined embedded waveform are changed in the amplitude direction In this case, when the information “1” is embedded in the container signal, the waveform of the container signal is transformed so as to be close to the inverted waveform of the embedded waveform.

  FIG. 2 is a block diagram illustrating a configuration of the information extraction device. In the information extraction device, for example, as described with reference to FIG. 12, the image signal in which the information is embedded by the information embedding device is output to the recording medium P by the output device 2, and the recording medium P is output by the input device 3 such as a scanner. The embedded information is extracted from the information-embedded container signal such as the image signal in which the read and digitized information is embedded. First, the similarity between the waveform of the information-embedded container signal and the embedded waveform is measured (similarity measurement block 200). The embedding phase of information is estimated from the measured distribution of similarity (phase synchronization block 201). For each section of the information-embedded container signal, the measured similarity is examined within the estimation range of the embedding phase, and embedding information (“0” or “1”) is extracted (information extraction block 202).

(Description of information extraction device)
With reference to FIG. 3, the process of the information extraction apparatus will be described more specifically. FIG. 3A shows an embedded waveform (length T1). The same embedded waveform must be used for the information embedding device and the information extracting device.

  FIG. 3B shows a waveform of a signal in which information is embedded, that is, an information embedded container signal. When an information-embedded container signal is divided into sections of length T2, the middle waveform is the same as the embedded waveform in the left section (information “0” is embedded). In the right section, the middle waveform is the same as the waveform obtained by inverting the embedded waveform in the amplitude direction (information “1” is embedded).

  In the similarity measurement block 200, the local waveform of length T1 is extracted from the waveform of the information-embedded container signal while being gradually shifted, and the similarity between the local waveform and the embedded waveform is measured. Here, a description will be given assuming that cross-correlation (ΣX (n) * Y (n)) is used as the similarity. However, X (n) is the local waveform of the information-embedded container signal, and Y (n) is the embedded waveform.

The difference (Σ | X (n) −Y (n) | or Σ {X (n) −Y (n)} ² ) can also be used as the similarity. The cross-correlation takes a larger absolute value as it is more similar. On the other hand, the difference is that the smaller the difference, the smaller the absolute value. The difference is easier to calculate. However, when using the difference, it is necessary to calculate both the difference from the embedded waveform Y (n) and the difference from the inverted waveform Y ′ (n) of the embedded waveform Y (n).

  FIG. 3C shows a waveform (similarity waveform) in which the cross-correlation values between the information-embedded container signal waveform and the embedded waveform in FIG. 3B are plotted. In the figure, a positive peak appears in the phase A and a negative peak appears in the phase B. Although it is likely that correct information can be extracted simply by detecting such a peak, as described above, an image signal in which information is embedded is output to a recording medium, and the recording medium is read and re-electronicized. In the case of such an information-embedded container signal, the similarity waveform is, for example, as shown in FIG. Due to the effect of noise, the height of the peak is lowered and a phase shift occurs. For this reason, in the second section, since the peak of E is higher than B, E is erroneously extracted (correct answer = “1”, error = “0”). In order to avoid this, the following “phase synchronization” (estimation of embedded phase) is performed.

  The phase synchronization block 201 performs processing for adding and accumulating the absolute values of the cross-correlation values in FIG. As a result, a distribution as shown in FIG. 3E is obtained (the embedding phase is selected on the information embedding apparatus side so as to obtain such a distribution). In this example, since the phase at the substantially center of the section shows a high value, the probability that this phase is an embedded phase is the highest. Therefore, for example, a phase range (from T3 to T3 + T4) having a value of 90 or more when the maximum value of the distribution is 100 is estimated as the embedded phase range.

  In the information extraction block 202, the cross-correlation value of FIG. 3D is checked for only the estimated phase range for each section of length T2, and if the maximum value is equal to or greater than the predetermined value Rd, “0” is set to the minimum. If the value is equal to or less than the predetermined value −Rd, “1” is extracted as embedded information. For example, in the second section of FIG. 3D, B within the estimated phase range is extracted as “1”, but E outside the estimated phase range is not extracted, so the reliability of the extracted information is improved. In practice, there is a method of repeatedly embedding the same information in a plurality of places. Generally, the majority of the extracted information in a plurality of places is taken as final extraction information. This will be described later.

(Description of information embedding device)
The processing of the information embedding device will be described more specifically with reference to FIG.

  In the similarity measurement block 100, the similarity between the waveform of the input container signal and the embedded waveform (FIG. 3A) is measured. This process is the same as the process in the similarity measurement block 200 of the information extraction apparatus, except that the container signal is before information embedding. Here, the cross-correlation is used as the similarity, but as described above, the difference can be used as the similarity.

  FIG. 4 (c) shows a waveform (similarity waveform) in which the cross-correlation values between the waveform of the container signal before embedding information and the embedded waveform are plotted. Since it is a cross-correlation value between the waveform of the container signal before embedding information and the embedded waveform, a peak as large as that in the case of FIG.

  The phase selection block 101 selects an appropriate embedding phase based on the measured similarity (here, the cross-correlation value) for each section of length T2. However, if all phases in the section are handled equally, there is no bias in the distribution of the embedded phase, and “phase synchronization” (embedded phase estimation) in the information extraction device becomes impossible. This is because the phase synchronization block 201 of the information extraction device is premised on the distribution of the embedded phase being biased.

  Two phase selection methods are conceivable for causing a bias in the distribution of the embedded phase. In the first phase selection method, as shown in FIG. 4A, only some phases (circles) in the central portion of the section are candidates for embedding phases, and the absolute value of the cross-correlation value is included in these phases. A phase having a large value (that is, a phase in which the cross-correlation value is equal to or greater than the predetermined value Re in the interval in which the information “0” is embedded, and a phase in which the cross-correlation value is equal to or less than the predetermined value −Re in the interval in which the information “1” is embedded) How to choose. The phase marked with x in FIG. 4A is not to be embedded.

  As shown in FIG. 4 (b), another phase selection method uses a plurality of phases in a section as embedded phase candidates, and gives priority to them in the phase closer to the center of the section (in the figure, A phase with a large absolute value of the cross-correlation value (that is, a phase in which the cross-correlation value is greater than or equal to the predetermined value Re in the section in which the information “0” is embedded, and a cross-correlation value in the section in which the information “1” is embedded). This is a method of selecting the phase with the highest priority among the predetermined phase-phases less than or equal to Re as the embedded phase. The phase marked with x in FIG. 4B is not to be embedded.

  In any phase selection method, if the embedding information is “0” in the left section of FIG. 4C, the phase A, not F, is appropriately selected as the embedding phase. Then, since the distribution of the embedding phase is biased, the “phase synchronization” in the information extracting device functions and the reliability of the extracted information can be improved.

  Conventionally, information is embedded only in one specific phase within a section. When the container signal waveform in a specific phase in a certain section is significantly different from the embedded waveform, the embedding in that section is abandoned. This is because if the container signal waveform is deformed so as to be close to the embedded waveform in such a section, the signal quality is greatly deteriorated. As described above, if the number of sections for which embedding is abandoned increases, the number of information extractions in the information extraction apparatus decreases, leading to a decrease in the reliability of the extracted information.

  However, even if the container signal waveform and the embedded waveform are greatly different in a specific phase in the section, there is a possibility that another phase in which the container signal waveform and the embedded waveform are close in the same section exists. In the present invention, since the above-described embedding phase selection is performed, the probability of giving up embedding in each section is lowered. When the same information is repeatedly embedded in a plurality of places and the majority of the information extracted by the information extraction apparatus is taken, the number of information extraction increases if the probability of abandonment is reduced, and the reliability of the extracted information is improved.

  In the signal transformation block 102, signal transformation processing for embedding information is performed on the phase selected by the phase selection processing. The signal transformation process is as described above. When the majority is used in the information extraction device, the information embedding device repeatedly embeds the same information in a plurality of places.

(When using difference as similarity)
A supplementary explanation will be given for the case where a difference is used as the similarity.

First, the information embedding device will be described. In the similarity measurement block 100, the local waveform of length T1 is cut out from the waveform of the container signal before embedding information while being gradually shifted. Then, the difference between the local waveform and the embedded waveform (Σ | X (n) −Y (n) | or Σ {X (n) −Y (n)} ² ), and the inversion of the local waveform and the embedded waveform The difference (Σ | X (n) −Y ′ (n) | or Σ {X (n) −Y ′ (n)} ² ) from the waveform is measured. X (n) is a local waveform of the container signal, Y (n) is an embedded waveform, and Y ′ (n) is an inverted waveform of the embedded waveform.

  In the phase selection block 101, when the selection method shown in FIG. 4A is used, in the section where the information “0” is embedded, the difference from the embedded waveform among the embedded phase candidates becomes a value equal to or less than a predetermined threshold. Select the phase as the embedding phase. In the section in which the information “1” is embedded, the phase in which the difference from the inverted waveform of the embedded waveform is equal to or less than a predetermined threshold is selected as the embedded phase. In the selection method shown in FIG. 4B, in the section in which the information “0” is embedded, the phase with the highest priority is embedded among the embedded phase candidates whose difference from the embedded waveform is equal to or less than a predetermined threshold. In a section in which information “1” is embedded as a phase selected as a phase, a phase having the highest priority among the embedded phase candidates whose difference from the inverted waveform of the embedded waveform is equal to or less than a predetermined threshold is selected as the embedded phase. .

  Next, the information extraction apparatus will be described. In the similarity measurement block 200, the difference between the waveform of the information-embedded container signal and the difference between the embedded waveform and the inverted waveform of the embedded waveform is measured. In the phase synchronization block 201, a difference distribution is obtained by performing a process of adding and accumulating both differences (both positive values) in a plurality of sections for each same phase. This distribution has a minimum value at the embedding phase. Therefore, for example, when the maximum value of the distribution is 100, a phase range having a value of 10 or less is estimated as an embedded phase range.

  In the information extraction block 202, for each section, two types of differences are checked for only the estimated phase range. If the difference from the embedded waveform is equal to or less than a predetermined threshold, “0” is set and the difference from the inverted waveform of the embedded waveform is set. If the value is equal to or less than a predetermined threshold, “1” is extracted as embedded information. The case of taking a majority vote will be described in the following embodiment.

  Hereinafter, an embodiment of the present invention will be described. In this embodiment, cross-correlation is used as the similarity. The embedded information is an N-bit bit string (bit 0 to bit N-1). For example, the container signal is an image signal, and an N-bit bit string is repeatedly embedded using the raster line as a processing unit.

(Description of information embedding device)
The processing of the similarity measurement block 100 will be described along the flowchart shown in FIG. First, the cutting start point of the container signal waveform is set to the origin 0 (step S300). A local waveform having a length T1 is cut out from the cutout start point from the container signal waveform (step S301). The similarity (cross-correlation) between the cut-out local waveform and the embedded waveform (length T1) is calculated, and the result is output (step S302). The cut start point is advanced by ΔT (step S304), and the process returns to step S301.

  In this way, the similarity between the local waveform of the container signal and the embedded waveform is calculated and output while sequentially shifting the cutout start point. When the extraction of the local waveform is completed up to the end of the container signal waveform (step S303, Yes), the similarity measurement process ends.

  The processing of the phase selection block 101 and the signal transformation block 102 will be described with reference to the flowchart shown in FIG. The “similarity waveform” in the following description means a waveform obtained by plotting the similarity (cross-correlation value) measured as shown in (c) or (d) of FIG.

  First, the cut-out start point of the similarity waveform is set to the origin 0, and the bit number n of the N-bit bit string that is the embedded information is set to 0 (step S310).

  A local waveform having a length T2 is cut out from the cut-out start point from the similarity waveform, and partial waveforms (length T1) at a plurality of embedded phase candidates are cut out from the local waveform (step S311). Here, the embedding phase candidates are those described with reference to FIG.

  If the bit (n) embedded in the section is “0” and the maximum value (cross-correlation value) of the partial waveform in a certain embedded phase candidate is equal to or greater than the predetermined value Re (Yes in step S312), the embedded phase candidate Is selected as the embedding phase, and the section of the container signal waveform having the phase length T1 (the portion corresponding to the partial waveform) is replaced with the embedded waveform (see FIG. 3A) (step S314). If the bit (n) to be embedded in the section is “1” and the minimum value (cross-correlation value) of the partial waveform at a certain embedded phase candidate is less than or equal to the predetermined value −Re (step S313, Yes), the embedded phase A candidate is selected as an embedding phase, and the section of the phase T1 of the phase of the container signal waveform (a portion corresponding to the partial waveform) is replaced with an inverted waveform of the embedded waveform (step S315).

  In FIG. 6, the determination procedure in steps S312 and 313 is shown in a simplified manner. That is, in the case of the phase selection method as shown in FIG. 4A, for example, the determination in step S312 or S313 is performed in order from the left embedded phase candidate, and when the condition is satisfied for a certain embedded phase candidate, Proceed to S315. Further, in the case of the phase selection method as shown in FIG. 4B, the determination in step S312 or S313 is performed in order from the embedded phase candidate with the highest priority, and when the condition is satisfied for a certain embedded phase candidate, step S314 is performed. Or it progresses to S315. If the condition is not satisfied for any embedding phase candidate (the determination result in steps S312 and S313 is No), the section is not embedded.

  Next, the cut-out start point of the similarity waveform is advanced by the length T2 and the bit number n is incremented by 1 (step S317), the process returns to step S311 and the same phase selection and signal transformation are performed for the next section. If the bit number n is increased to N by repeating the above (step S318, Yes), it means that the padding up to the last bit of the N-bit bit string has been completed, so the bit number n is returned to 0 (step S319). Returning to step S311, the embedding process is started again from bit number 0 of the bit string. When the process proceeds to the end of the similarity waveform and there is no partial waveform to be extracted (step S316, Yes), the process ends.

(Description of information extraction device)
The processing of the similarity measurement block 200 will be described along the flowchart shown in FIG. First, the cutting start point of the information-embedded container signal waveform is set to the origin 0 (step S400). A local waveform having a length T1 is cut out from the cut start point from the information embedded container signal waveform (step S401). The similarity (cross-correlation) between the cut-out local waveform and the embedded waveform (length T1) is calculated, and the result is output (step S402). The cut start point is advanced by ΔT (step S404), and the process returns to step S401. In this manner, the similarity between the local waveform of the information-embedded container signal and the embedded waveform is calculated and output while sequentially shifting the cutout start point. When the extraction of the local waveform is completed up to the end of the information-embedded container signal (step S403, Yes), the similarity measurement process ends.

  The processing of the phase synchronization block 201 will be described along the flowchart shown in FIG. First, the cut-off start point of the similarity waveform is set to the origin 0 (step S410). A local waveform having a length T2 is cut out from the cut-out start point from the similarity waveform (step S411). The absolute value of the extracted local waveform is added to the cumulative waveform (step S412). This corresponds to the cumulative addition process for obtaining the distribution described with reference to FIG. The cut start point is advanced by T2 (step S414), the process returns to step S411, and the same processing is repeated. When segmentation of the local waveform is completed up to the end of the similarity waveform (step S413, Yes), the maximum value of the cumulative waveform (corresponding to the distribution of FIG. 3E) is 100%, and the value ranges from 90% to 100%. And T3 and T4 (see FIG. 3E) are determined (step S415).

  The processing of the information extraction block 202 will be described along the flowchart shown in FIG. A counter corresponding to the bit number for taking the majority of the extracted information is used. Here, since it is assumed that a bit string of N bits is repeatedly embedded, N counters (n) (where n = 0, 1,..., N−1) corresponding to the bit numbers are used. .

  First, all the counters (n) are cleared to 0, the bit number n (or counter index) is set to 0, and the cutout start point of the similarity waveform is set to T3 (step S420). T3 is determined in step S415 (see FIG. 3E).

  A local waveform having a length T4 is cut out from the cut-out start point from the similarity waveform (step S421). Here, T4 is set in step S415 (see FIG. 3E). If the maximum value of the cut-out local waveform is equal to or greater than the predetermined value Rd (step S422, Yes), it means that “0” has been extracted as the embedded information, so the counter (n) corresponding to the current bit number n is set to 1. Increment by (step S424). If the minimum value of the local waveform is less than or equal to the predetermined value −Rd (step S423, Yes), it means that “1” has been extracted as the embedded information, and therefore the counter (n) corresponding to the current bit number n is set to 1. Decrement (step S425). If any of the determination results in steps S422 and S423 is No, the counter (n) is not updated.

  While the cut start point is advanced by T2, the bit number n is incremented by 1 (step S427), and the process returns to step S421 to extract the embedded information of the next section and update the counter. When the bit number n is incremented to N in step S427 (step S428, Yes), the bit number n is returned to 0 (step S429), and the process returns to step S421.

  The extraction of the embedded information and the update of the counter (n) are repeated for each section of the length T2, and finally, when the local waveform is cut out to the end of the similarity waveform (Yes in step S426), the extraction information output process is performed. Do. The counter (n) has an initial value of 0, and is incremented by 1 each time “0” is extracted as information of bit number n, and is decremented by 1 each time “1” is extracted. The value of the counter (n) corresponds to the majority result of the extracted information of bit number n.

  First, the bit number n is set to 0 (step S430). Whether the value of the counter (n) corresponding to the bit number n is positive or negative is determined (steps S431 and S433). If the value of the counter (n) is a positive value (step S431, Yes), “0” is output as the extraction information of the bit number n (step S432). If the value of the counter (n) is a negative value (step S433, Yes), “1” is output as the extraction information of the bit number n. When the value of the counter (n) is 0, it means that neither “0” nor “1” can be determined for the extracted information of the bit number by majority, so that the fact is output (step S435). Next, the bit number n is incremented by 1 (step S436), and the process returns to step S431, and the extraction information by majority vote is output for the next bit number. Extraction information is output up to bit number N-1, and when bit number n is incremented to N (step S437, Yes), the process is terminated.

  Next, an embodiment using a difference as the similarity will be described with the aid of the flowchart used in the first embodiment.

(Description of information embedding device)
In the flowchart of FIG. 5 showing the processing flow of the similarity measurement block 100, the processing in step S302 is different from that in the first embodiment. That is, in this embodiment, as the similarity, both the difference between the local waveform of the length T1 cut out from the container signal waveform and the embedded waveform and the difference between the local waveform and the inverted waveform of the embedded waveform are calculated. . As described above, each difference takes a value of 0 or more. The other processing contents are the same as those in the first embodiment.

  In the flowchart of FIG. 6 showing the processing flow of the phase selection block 101 and the signal transformation block 102, the processing of steps S311, S312 and S313 is different from that of the first embodiment. In the present embodiment, the similarity waveform includes a difference waveform between the embedded waveform and an inverted waveform of the embedded waveform. The former is called a similarity waveform (1), and the latter is called a similarity waveform (2).

  In step S311, a local waveform in the section of length T2 from the cutout start point is cut out from the similarity waveform (1) and the similarity waveform (2), and a partial waveform of an embedded phase candidate is further cut out from the cut out local waveform. In the determination step S312 in the case where the information to be embedded in the section is “0”, the determination result is Yes if the minimum value of the partial waveform on the similarity waveform (1) side is equal to or less than a predetermined value. In determination step S313 when the embedding information in the section is “1”, if the minimum value of the partial waveform on the similarity waveform (2) side is equal to or less than a predetermined value, the determination result is Yes. However, as described in the first embodiment, in the case of the phase selection method of FIG. 4A, for example, the determination in steps S312 and S313 is performed from the left embedded phase candidate, and the embedded phase candidate that satisfies the condition is determined. The phase is selected as an embedding phase and the process proceeds to step S314 or S315. In the case of the phase selection method shown in FIG. 4B, the steps S312 and S313 are reversed from the embedded phase candidates having the highest priority, the embedded phase candidates that satisfy the condition are selected as the embedded phases, and the process proceeds to step S314 or S315. move on.

(Description of information extraction device)
In the flowchart of FIG. 7 showing the processing flow of the similarity measurement block 200, the processing in step S402 is different from that in the first embodiment. That is, in this embodiment, as the similarity, the difference between the local waveform of the length T1 cut out from the information-embedded container signal waveform and the embedded waveform, and the difference between the local waveform and the inverted waveform of the embedded waveform are calculated. As described above, each difference takes a value of 0 or more. The other processing contents are the same as those in the first embodiment.

  In the flowchart of FIG. 8 showing the processing flow of the phase synchronization block 201, the processing contents of steps S411, S412 and S415 are different from those of the first embodiment. That is, in step S411, the length from the cut start point is determined based on the similarity waveform (1) which is a difference waveform from the embedded waveform and the similarity waveform (2) which is a difference waveform from the inverted waveform of the embedded waveform. A local waveform in the section T2 is cut out. In step S412, the local waveform on the similarity waveform (1) side and the local waveform on the similarity waveform (2) side are added to the cumulative waveform. In step S415, assuming that the maximum value of the cumulative waveform value is 100%, for example, a phase range that takes a value of 0% to 10% is set as an estimated range, and T3 and T4 (see FIG. 3E) are determined.

  In the flowchart of FIG. 9 showing the processing flow of the information extraction block 202, the contents of steps S421, S422, and S423 are different from those of the first embodiment. That is, in step S421, a local waveform having a length T4 is cut out from the similarity waveform (1) and the similarity waveform (2). In step S422, it is determined whether the minimum value of the local waveform cut out from the similarity waveform (1) is equal to or smaller than a predetermined value. In step S423, it is determined whether the minimum value of the local waveform cut out from the similarity waveform (2) is equal to or smaller than a predetermined value.

[Supplementary explanation of device configuration]
The information processing apparatus that executes the process of the information embedding apparatus and / or the process of the information extraction apparatus described above can be realized by dedicated hardware or by a program using hardware resources of a computer. Is possible. An example of the latter information processing apparatus will be described with reference to FIG.

  In FIG. 10, 400 is a CPU (Central Processing Unit), and 402 is a ROM or HDD (Hard Disk Device) that stores programs and various data of the CPU 400. Reference numeral 401 denotes a signal processing unit capable of high-speed signal processing, which includes a circuit such as an ASIC or a DSP. Reference numeral 403 denotes a RAM, which is used as a working memory for developing data to be processed, an execution program, and the like. Reference numeral 404 denotes an operation unit (switches, buttons, etc.) and a display unit (LEDs, LCDs, etc.) for exchanging information regarding operations and states between the operator and the apparatus. Reference numeral 405 denotes a reading device for various recording media (IC card, CD, DVD, etc.) and an interface (bus, Ethernet, telephone line, wireless, etc.) with other external devices, through which a container signal or information is embedded. Used container signal is input / output. The processing steps of each block shown in FIG. 1 or 2 are executed by the CPU 400 or the signal processing unit 401. In other words, the CPU 400 or the signal processing unit 401 functions as means corresponding to each block shown in FIG. 1 or FIG. The program for that purpose is read from, for example, the ROM or the HDD 402 and loaded into the RAM 403, or is read from various recording media (IC card, CD, DVD, etc.) through the external interface 405 and loaded into the RAM 403.

It is a block diagram which shows the structure of the information embedding apparatus of this invention. It is a block diagram which shows the structure of the information extraction apparatus of this invention. It is a figure for demonstrating the information extraction process in this invention. It is a figure for demonstrating the information embedding process in this invention. It is a flowchart for demonstrating the process of the similarity measurement block of an information embedding apparatus. It is a flowchart for demonstrating the process of the phase selection block of an information embedding apparatus, and a signal deformation | transformation block. It is a flowchart for demonstrating the process of the similarity measurement block of an information extraction device. It is a flowchart for demonstrating the process of the phase-synchronization block of an information extraction device. It is a flowchart for demonstrating the process of the information extraction block of an information extraction device. It is a block diagram for demonstrating the form which implement | achieves the information processing apparatus of this invention using the hardware resource of a computer. It is explanatory drawing of a prior art example. It is explanatory drawing of the processing form from information embedding to information extraction.

Explanation of symbols

100 Similarity Measurement Block 101 Phase Selection Block 102 Signal Deformation Block 200 Similarity Measurement Block 201 Phase Synchronization Block 202 Information Extraction Block

Claims (11)

A signal before embedding information (hereinafter referred to as a container signal) is divided into sections of length T2, and the section is approximated to a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section. An information processing apparatus that embeds information in the container signal by modifying the waveform of the container signal,
From the container signal, while sequentially shifting the cut-out position, cut out a local waveform having a length T1, and measure a similarity measure for measuring the cross-correlation between the cut-out local waveform and the embedded waveform;
For each of the sections, a cross-correlation condition measurement is performed on the plurality of embedded phase candidates in the section by the similarity measurement unit, and one phase that satisfies the determination condition in the plurality of embedded phase candidates is embedded. Phase selection means for selecting as a phase;
An information processing apparatus comprising: a signal deforming unit configured to transform the waveform of the container signal in the embedding phase selected by the phase selecting unit so as to approach the embedded waveform or its inverted waveform for each section. The signal before the information is embedded (hereinafter referred to as a container signal) is divided into sections of length T2, and the section is approximated to a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section. An information processing apparatus that embeds information in the container signal by modifying the waveform of the container signal,
From the container signal, the local waveform of length T1 is cut out while sequentially shifting the cut-out position, and the difference between the cut-out local waveform and the embedded waveform (denoted as the first difference) and the inverted waveform of the embedded waveform, Similarity measurement means for measuring the difference (denoted as a second difference) of
For each of the sections, condition determination of the first difference or the second difference measured by the similarity measurement unit is performed for a plurality of embedded phase candidates in the section, and the determination condition in the plurality of embedded phase candidates is satisfied. Phase selection means for selecting one of the phases as an embedding phase;
An information processing apparatus comprising: a signal deforming unit configured to transform the waveform of the container signal in the embedding phase selected by the phase selecting unit so as to approach the embedded waveform or its inverted waveform for each section.   The priority is given to the plurality of embedded phase candidates, and the phase selection unit preferentially selects a phase having a high priority among the plurality of embedded phase candidates as an embedded phase. 2. The information processing apparatus according to 2. A signal in which information is embedded by changing the waveform so as to approach a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section of length T2 (hereinafter referred to as an information embedded container signal). ) To extract information embedded for each section of length T2,
A similarity measuring unit that cuts out a local waveform of length T1 while sequentially shifting the cut-out position from the information-embedded container signal, and measures cross-correlation between the cut-out local waveform and the embedded waveform;
From the cross-correlation measured by the similarity measurer, obtain a distribution of the absolute value in the interval, and estimate the information embedded phase range in the interval from the distribution;
For each section, the information embedded phase range estimated by the phase synchronization means is embedded in the section by comparing the maximum or minimum value of the cross-correlation measured by the similarity measurement means with a predetermined value. An information processing apparatus comprising information extraction means for extracting information. A signal in which information is embedded by changing the waveform so as to approach a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section of length T2 (hereinafter referred to as an information embedded container signal). ) To extract information embedded for each section of length T2,
A local waveform having a length of T1 is cut out from the information-embedded container signal while sequentially shifting the cut-out position. The difference between the cut-out local waveform and the embedded waveform (denoted as a first difference) and the inversion of the embedded waveform Similarity measuring means for measuring a difference from a waveform (denoted as a second difference);
A phase synchronization means for obtaining a distribution of the first and second differences measured by the similarity measurement means in the section and estimating an information embedding phase range in the section from the distribution;
For each section, the information embedded phase range estimated by the phase synchronization means is embedded in the section by comparing the minimum value of the first or second difference measured by the similarity measurement means with a predetermined value. An information processing apparatus comprising information extracting means for extracting the extracted information. The signal before the information is embedded (hereinafter referred to as a container signal) is divided into sections of length T2, and the section is approximated to a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section. An information processing method for embedding information in the container signal by modifying the waveform of the container signal,
From the container signal, while sequentially shifting the cut-out position, a local waveform having a length T1 is cut out, and a similarity measurement step of measuring the cross-correlation between the cut-out local waveform and the embedded waveform;
For each of the sections, the cross-correlation condition measurement performed by the similarity measurement step is performed on a plurality of embedded phase candidates in the section, and one phase that satisfies the determination condition among the plurality of embedded phase candidates is embedded. A phase selection step to select as the phase;
An information processing method comprising: a signal transformation step for transforming the waveform of the container signal in the embedding phase selected by the phase selection step so as to approach the embedding waveform or its inverted waveform for each section. The signal before the information is embedded (hereinafter referred to as a container signal) is divided into sections of length T2, and the section is approximated to a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section. An information processing method for embedding information in the container signal by modifying the waveform of the container signal,
From the container signal, the local waveform of length T1 is cut out while sequentially shifting the cut-out position, and the difference between the cut-out local waveform and the embedded waveform (denoted as the first difference) and the inverted waveform of the embedded waveform, A similarity measurement step of measuring a difference (denoted as a second difference) of
For each of the sections, condition determination of the first difference or the second difference measured by the similarity measurement step is performed for a plurality of embedded phase candidates in the section, and the determination condition in the plurality of embedded phase candidates is satisfied. A phase selection step of selecting the generated one phase as an embedding phase;
An information processing method comprising: a signal transformation step for transforming the waveform of the container signal in the embedding phase selected by the phase selection step so as to approach the embedding waveform or its inverted waveform for each section. A signal in which information is embedded by changing the waveform so as to approach a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section of length T2 (hereinafter referred to as an information embedded container signal). ) To extract information embedded for each section of length T2,
A similarity measurement step of cutting out a local waveform of length T1 while sequentially shifting the cut-out position from the information-embedded container signal, and measuring a cross-correlation between the cut-out local waveform and the embedded waveform;
From the cross-correlation measured by the similarity measurement step, obtain a distribution of the absolute value in the section, and estimate the information embedded phase range in the section from the distribution,
For each section, the information embedded phase range estimated by the phase synchronization process is embedded in the section by comparing the maximum value or the minimum value of the cross-correlation measured by the similarity measurement process with a predetermined value. And an information extracting method for extracting information. A signal in which information is embedded by changing the waveform so as to approach a predetermined embedded waveform of length T1 (<T2) or its inverted waveform for each section of length T2 (hereinafter referred to as an information embedded container signal). ) To extract information embedded for each section of length T2,
A local waveform having a length of T1 is cut out from the information-embedded container signal while sequentially shifting the cut-out position. The difference between the cut-out local waveform and the embedded waveform (denoted as a first difference) and the inversion of the embedded waveform A similarity measurement step for measuring a difference from a waveform (denoted as a second difference);
A phase synchronization step of obtaining a distribution in the section of the first and second differences measured by the similarity measurement step and estimating an information embedding phase range in the section from the distribution;
For each section, the information embedding phase range estimated by the phase synchronization step is embedded in the section by comparing the minimum value of the first or second difference measured by the similarity measurement step with a predetermined value. And an information extracting process for extracting the extracted information.   A program for causing a computer to execute each step of the information processing method according to claim 6, 7, 8 or 9.   A computer-readable recording medium on which a program for causing a computer to execute each step of the information processing method according to claim 6, 7, 8, or 9 is recorded.

Images