JP4740094B2 - Information embedding device, information extracting device, information embedding method, information extracting method, information embedding program, and information extracting program - Google Patents

Description

  The present invention relates to an information embedding device, an information extracting device, an information embedding method, an information extracting method, an information embedding program, and an information extracting program. More specifically, the present invention relates to copyright holder information and content information using digital watermark technology. The present invention relates to an information embedding device, an information extracting device, an information embedding method, an information extracting method, an information embedding program, and an information extracting program for embedding embedded data such as accessor information inseparably and extracting the embedded data.

  In recent years, copyright owner information for preventing unauthorized use of digital content data or accessor information for preventing unauthorized leakage of confidential data is inseparably embedded in the content data body (hereinafter also referred to as a container signal). Digital watermarking technology is attracting attention.

  In this digital watermark technology, an embedding process for adding a predetermined change expressed by information to be embedded in content data such as an image and sound to an unperceivable level, and an extraction process for detecting the predetermined change from the embedded data and extracting the information And is composed of.

  If the container signal is propagated between the embedding process and the extracting process, the embedded data may be lost due to various external changes (attack). Therefore, the same data is repeatedly embedded in one container signal. If the same data is embedded in n locations and the probability of data loss at each location is p, the probability of data loss is “p ＾ n”, which is the principle that defects decrease as n increases. For example, if a data string in which 8-bit data (hereinafter referred to as a symbol) as shown in the embedded data string in FIG. 10 is embedded is embedded, even if a part of the data string is erroneously extracted (underline in the error data string in FIG. 10) The correct symbol (0xC4) can be obtained if the majority is taken with the same phase data.

  However, the type of change (attack) on the container signal is not limited to the data error of the embedded data string, but is also considered to be a cut data string (data cut) or an inverted data string (data rotation) as shown in FIG. It is done. For this reason, as shown in the cut data string in FIG. 10, when the first 3 bits of the embedded data string are cut, the embedding phase is shifted, and an erroneous symbol (0x26) may be extracted.

  Further, as shown in the inverted data sequence in FIG. 10, if the sequence of the embedded data sequence is inverted, the embedding direction may be out of order and an erroneous symbol (0x23) may be extracted.

  In order to cope with these problems, two types of methods have been conventionally used. One is a method for extracting information on the basis of the characteristic originally provided in the container signal, and the other is a method for extracting information by applying a reference process to the container signal.

  The former extracts information on the basis of image feature points such as banknotes and securities (see Patent Document 1), so that correct information can be extracted even if there is clipping, parallel movement, rotation, or the like.

  The latter includes an object such as a frame line (see Patent Document 2) or an image obtained by synthesizing a predetermined high-frequency component into an image and performing alignment based on this (see Patent Document 3). In addition, the embedded data itself includes a reference, and there is a method of adding specific data “2, 3” to a symbol as shown in the comparative example embedded data string of FIG. , 1, 2, 3 quaternary modulation embedding example). This specific data is data that is not subject to rotation and does not appear in the symbol.

JP 2001-313822 A Japanese Patent No. 3720748 JP 2001-292302 A

  However, the former method is effective only for a finite type of image having a predetermined feature point, and thus has a problem of lack of versatility.

  In the latter method, another image processing is added to the image processing for information embedding, so that there is a problem that the image quality deteriorates.

  The present invention has been made in view of the above, and while maintaining the versatility applicable to various container signals, it suppresses the deterioration of the quality of the container signal when embedding information, and sets the synchronization state of embedded data. An object is to provide an information embedding device, an information extraction device, an information embedding method, an information extraction method, an information embedding program, and an information extraction program that can be correctly determined.

  In order to solve the above-described problems and achieve the object, the present invention is based on embedded data in which content data is divided into intervals of a predetermined interval and a predetermined plurality of intervals connected to each other is defined as one cycle. An information embedding device that modulates the content data, wherein the embedded data includes at least one non-modulation section for synchronization in which the content data is unmodulated at a fixed position within one period in addition to the embedded information. A degree-of-similarity measuring means for measuring the degree of similarity between the embedded data and the content data for each section; and whether to embed the embedded data according to the degree of similarity for each section measured by the similarity measuring means Embedded determination means for determining the embedded data and signal modulation means for performing signal modulation when the embedded data is embedded, In the embedded information section, using the similarity measurement means, the embedding determination means and the signal modulation means, the embedded data and the content data are processed for each section to embed embedded data, and in the unmodulated section, The content data is not subjected to signal modulation processing.

  Further, according to a preferred aspect of the present invention, the embedded data includes, in addition to embedded information, at least three non-modulating sections for synchronization in which the content data is unmodulated at a fixed position within one cycle. It is desirable.

  Further, according to a preferred aspect of the present invention, the embedded data includes a plurality of non-modulating sections for synchronization in which the content data is not modulated in addition to the embedded information at fixed positions within one cycle, and is asynchronous It is desirable to disperse the positions of the non-modulation intervals so that the number of overlaps of the non-modulation intervals in the state is equal to or less than a predetermined number.

In order to solve the above-described problems and achieve the object, the present invention is an information extraction device that extracts embedded data embedded in content data using any one of the information embedding devices, Similarity measuring means for measuring the degree of similarity between the embedded data and the content data for each section; information extracting means for extracting embedded information according to the similarity for each section measured by the similarity measuring means; Statistics extraction means for extracting the embedded data by taking statistics of the number of extractions in the same interval while synchronizing in the non-modulation interval using the extraction information extracted by the information extraction means And

  In order to solve the above-described problems and achieve the object, the present invention is based on embedded data in which content data is divided into predetermined intervals, and a plurality of concatenated intervals are defined as one cycle and correspond to the one cycle. An information embedding method for modulating the content data, wherein the embedded data includes, in addition to the embedded information, a non-modulation section for synchronization in which the content data is unmodulated in one section at a fixed position within one cycle. The similarity measurement step for measuring the similarity between the embedded data and the content data for each section, and whether to embed the embedded data according to the similarity for each section measured by the similarity measurement step. An embedding determination step for determining whether or not, and a signal modulation step for performing signal modulation when embedding the embedded data, In the embedded data section of embedded data, the embedded data is embedded by processing the embedded data and the content data for each section by the similarity measurement step, the embedded determination step, and the signal modulation step, and the unmodulated section Then, signal modulation processing of the content data is not performed.

  In order to solve the above-described problems and achieve the object, the present invention provides an information extraction method for extracting embedded data embedded in content data using the information embedding method, wherein the embedded data and the embedded data A similarity measurement step for measuring similarity between content data for each section, an information extraction step for extracting embedded information according to the similarity for each section measured by the similarity measurement step, and extraction by the information extraction step And a statistical extraction step of extracting the embedded data by taking statistics of the number of extractions in the same interval while synchronizing in the non-modulation interval using the extracted information.

  In order to solve the above-described problems and achieve the object, the present invention is characterized in that the above-described information embedding method is executed by a computer.

  In order to solve the above-described problems and achieve the object, the present invention is characterized in that the above-described information extraction method is executed by a computer.

  According to the present invention, the embedded data includes, in addition to the embedded information, at least one non-modulation section for synchronization in which the content data is unmodulated at a fixed position within one period. The degree of similarity between the embedded data and the content data is measured for each section, and whether to embed the embedded data is determined by the embedding determination unit according to the similarity, and when the embedded data is embedded, the signal modulation unit performs signal modulation processing. Is done. In the embedded information section of the embedded data, signal modulation processing for embedding the embedded data in the content data is performed using the similarity measuring unit, the embedded determining unit, and the signal modulating unit, and in the non-modulated section of the embedded data, the signal modulation of the content data is performed. Do not process. As described above, since the embedded data of the present invention has one or more non-modulating sections for synchronization in which the embedded information and the content data are unmodulated at a fixed position within one period, the embedded data is provided with the content data. Compared to the conventional method of extracting information based on the characteristics that are present, there is an effect of versatility. In addition, in the non-modulation period, if the statistics of the number of embedded data are taken, it is always “0”, so that it can be used as a reliable reference for synchronizing embedded data. This means that even if a change (attack) is applied to the content data and data error or data cutout occurs, synchronization can be established based on the non-modulation period, and correct data can be extracted. To do. Further, since the non-modulation section is a section that is not processed with respect to the content data, there is an effect that image quality deterioration can be suppressed accordingly.

  Further, according to the present invention, the similarity measurement unit measures the similarity between the embedded data and the content data for each section, and the information extraction unit extracts the embedded information according to the similarity, and the extracted information is The embedded data is extracted by taking statistics of the number of samples extracted in the same section while synchronizing in the non-modulation section using the statistics extracting means. As described above, in the embedded data of the present invention, since one or more non-modulating sections for synchronization are provided at fixed positions within one period, the same section can be used while reliably synchronizing using the non-modulating sections. By taking statistics of the number of extractions for each, it is possible to extract the correct embedded data while maintaining the versatility of the content data, reducing quality degradation.

  Hereinafter, embodiments of an information embedding device, an information extracting device, an information embedding method, an information extracting method, an information embedding program, and an information extracting program according to the present invention will be described in detail with reference to the drawings. Note that the present invention is not limited to the embodiments.

(First embodiment)
1 is a system configuration diagram including an information embedding device and an information extraction device according to the present invention, FIG. 2 is a block diagram illustrating the configuration of the information embedding device and the information extraction device in FIG. 1, and FIG. 1 is a functional configuration block diagram of a CPU or a signal processing unit in the information embedding device of FIG. 1, FIG. 4 is a functional configuration block diagram of a CPU or a signal processing unit in the information extraction device of FIG. 1, and FIG. 6 is a main flowchart of the information embedding process according to this embodiment, FIG. 6 is a main flowchart of the information extraction process according to the present embodiment, FIG. 7 is a waveform diagram of embedded data, and FIG. FIG. 8-2 is a waveform diagram of a container signal after embedding, FIG. 9-1 is a cross-correlation waveform diagram before embedding, and FIG. Padding Fig. 10 is a cross-correlation waveform diagram at the time of post-extraction, Fig. 10 is a diagram showing various embedded data strings, and Fig. 11 is an example of a data string from the embedded data string to the extracted data string according to the present embodiment. FIG. 12 is a statistical diagram in which statistics of the number of embeddings are taken for each period, and FIG. 13 is a statistical chart in which statistics of the number of extractions are taken for each period.

  In the system including the information embedding device and the information extracting device shown in FIG. 1, when content data such as an image signal is input to the information embedding device 10, another information (container signal) is displayed on the content data body (container signal). For example, processing for embedding user ID information) is performed, and the output device 20 converts it into a form suitable for use (for example, printing on paper 50 by a printer).

  On the other hand, the input device 30 converts the image printed on the paper 50 into a signal from which information can be extracted (for example, the image data is read and digitized by a scanner), and the information extraction device 40 extracts information embedded therein. Is done.

  In this way, copyright data for preventing unauthorized use or accessor information for preventing unauthorized leakage of confidential data is embedded in the content data as embedded data. However, if various changes (attack) are applied from the outside while the container signal is propagated from the above-described embedding process to extraction process, the embedded data may be lost.

  Therefore, in the first embodiment of the present invention, as shown in FIG. 2, the information embedding device 10 and the information extracting device 40 in FIG. 1 both have a CPU 100, an ASIC (Application Specific Integrated Circuit) and a DSP (DSP). A signal processing unit 101 that processes data at high speed with a circuit such as a digital signal processor), a ROM (or HDD) 102 that stores programs and data of the CPU 100, and a RAM 103 as a working memory that develops processing data and execution programs. , Operation / display unit 104 including an operation unit (switch, button, etc.) and a display unit (LED, LCD, etc.), and I / F (bus, Ethernet, etc.) with external devices and storage media (IC card, CD, DVD, etc.) (Registered trademark), telephone line, wireless, etc.) and is configured by an external I / F unit 105 that inputs and outputs container signals. It is.

  The information embedding process and the information extracting process according to the present invention can be implemented by hardware processing by the ASIC or the like of the signal processing unit 101 in FIG. Therefore, the configuration and operation of the information embedding device 10 of the present invention will be described with reference to FIGS. 3 and 5, and the configuration and operation of the information extraction device 40 of the present invention will be described with reference to FIGS. 4 and 6. To do. In the following, a case where a one-dimensional signal is mainly used will be described. However, a two-dimensional signal such as image data may be used, and the principle is the same.

[Information embedding device]
As shown in FIG. 3, the signal processing unit 101 of the information embedding device 10 (which is a functional block diagram based on software in the CPU 100) includes a similarity measuring unit 200, an embedding determination unit 201, a signal modulation unit 202, and the like. .

  First, as shown in FIG. 8A, the container signal waveform before the embedding process is divided by a section of a constant interval (Ts), and a predetermined plurality of sections (here, two sections) are connected in one cycle (T). Are treated as The embedded data string corresponding to one period is also called a symbol, and is data of a plurality of bits (here, 2 bits, value is 2 in decimal notation), and which bit corresponds to which section in one period. This is fixed. Here, 1 is assigned to the most significant bit (MSB) in the left section, and 0 is assigned to the least significant bit (LSB) in the right section. In other words, in the example illustrated in FIG. 8A, “1, 0, 1, 0” is to be embedded in four sections in order from the left.

  Further, the wavelength (Te) of the embedded waveform shown in FIG. 7 has a relationship of Te <Ts (section), and a part (length Te) of one section (length Ts) of the container signal waveform shown in FIG. ) Is brought close to this embedded waveform, information is embedded. For example, it may be replaced with the embedded waveform as it is in the real space, may be replaced with a waveform obtained by linearly interpolating both signals in the amplitude direction, or the coefficient may be changed in the frequency space. This is signal modulation processing.

  Specifically, the embedded waveform shown in FIG. 7 is set to “1”, and a waveform obtained by inverting this waveform in the amplitude direction is set to “0” to express embedded data 1 bit. As an example in which the container signal waveform is changed according to the embedded data, “1 and 0” are inserted in the two left sections as shown in FIG. However, as shown in the right end section of FIG. 8A, when a flat container signal waveform is brought closer to the embedded waveform, a large change is added, resulting in a deterioration in quality.

  Therefore, in the first embodiment of the present invention, a certain standard is provided, and the embedding of information in a section that does not satisfy this standard is stopped. Here, since the same embedded data is repeatedly embedded in preparation for missing information in the propagation process of the container signal, there is no problem even if the embedding is stopped in some sections. For example, as shown in FIG. 8B, embedding is stopped in the right two sections, but no information is lost because the same data is embedded in the two left sections.

The above-mentioned constant reference is the similarity between the waveform of the container signal and the waveform of the embedded data. In the similarity measurement unit 200 of the information embedding device 10 shown in FIG. 3, the local waveform of the same wavelength Te as that of the embedded data is extracted from the waveform of the container signal while being gradually shifted, and the similarity to the waveform of the embedded data is measured. (Step S400 in FIG. 5). This similarity includes a cross-correlation and a difference. Expressing the cross-correlation with an equation,
ΣX (n) * Y (n)
It becomes. Moreover, when the difference is expressed by an expression,
Σ | X (n) −Y (n) | or Σ {X (n) −Y (n)} * {X (n) −Y (n)}
It becomes. In the above equation, X (n) is the waveform of the container signal, and Y (n) is the waveform of the embedded data. The latter difference is simpler to calculate, but in the case of the difference, processing time is required in that it is necessary to calculate the waveform of Y (n) embedded data and its inverted waveform.

  The result of obtaining the cross-correlation for the container signal in FIG. 8-1 is the cross-correlation waveform shown in FIG. The embedding determination unit 201 of the information embedding device 10 in FIG. 3 determines whether or not to stop signal modulation by comparing the cross-correlation value R in the corresponding section with predetermined threshold values Re and -Re (FIG. 3). 5 step S401). For example, in FIG. 9A, when the embedded data satisfies the cross-correlation value R ≦ Re in the section “1”, or when the embedded data satisfies the cross-correlation value R ≧ −Re in the section “0”, When the signal modulation is stopped and the above condition is not satisfied, it is determined that the signal modulation is executed based on the embedded data. As a result, since the above condition is not satisfied in the left two sections of FIG. 9A, signal modulation is executed with “1 and 0”, and since the above two conditions are met, the signal modulation is stopped. (See FIG. 8-2).

  In the signal modulation unit 202 in FIG. 3, information is embedded by bringing a part of one section of the container signal waveform close to the waveform of the embedded data based on the determination result of the embedding determination unit 201. As described above, this signal modulation processing (step S402 in FIG. 5) replaces the embedded waveform as it is in the real space, replaces both signals with a waveform obtained by linear interpolation in the amplitude direction, or converts the coefficient in the frequency space. Change it.

  A feature of the present invention resides in that an embedded data string as shown in FIG. 11 is adopted in addition to the configuration of the information embedding apparatus 10 in FIG. That is, the embedded data string as the comparative example is configured by adding specific embedded data “2, 3” to the symbol as shown in the comparative example embedded data string of FIG. However, the embedded data string of the present invention does not add specific embedded data, but provides a section where data is not embedded at a position indicated by “X” in FIG. In the example of FIG. 11, among 1 period = 9 sections, the left section is not embedded (synchronization non-modulation section), and the remaining eight sections are used for symbol embedding.

  That is, in the information embedding process flow of FIG. 5, all the processes (similarity measurement process to signal modulation process) are passed only in this non-modulation section (one section), and in the remaining eight sections, “similarity “Measurement” and “embedding determination” processes are performed. Of course, in the 8 sections for symbol embedding, as shown in the actually embedded data string in FIG. 11, a non-modulation section (marked with X) occurs. However, as shown in the embedded number statistics in FIG. If statistics of the number of embeddings are taken, the non-modulation section for synchronization is 0, and the other sections are not 0. This is because the embedding process is repeated, and the third interval from the left of the actually embedded data string in FIG. 11 is normally the non-modulation interval “X”, but the next period, that is, 12 from the left. In the second section, “1” is embedded. As described above, since there is a place where a long repetition is necessarily filled, the frequency of the eight sections for symbol embedding does not become “0”. On the other hand, the non-modulation section for synchronization is always “0”. The present invention uses this principle for symbol synchronization.

  In the extracted data string in FIG. 11, the first 3 bits are cut off in the interval p, and the embedded data changes in the intervals q, r, s, and t. In this regard, FIG. 13 shows statistics of the number of extracted “0” and “1” every one period = 9 intervals. The statistical value closer to “0” than in FIG. 12 is the result of a change such as q resulting in a decrease in the number of extractions. Moreover, by changing like r, s, and t, some erroneous data may be extracted. However, these effects are small as a whole, and if either “0” or “1” is a predetermined number (k) or more, it can be estimated as appropriate embedded data, and both are predetermined. If it is less than the number (seventh section from the left in FIG. 13), it can be estimated as a symbol delimiter.

  In the information embedding device 10 of FIG. 1, when the embedded data is embedded in the container signal by the above configuration unit, it is printed on the paper 50 by a printer or the like as the output device 20.

[Information Extraction Device]
Subsequently, when the embedded data embedded in the container signal is extracted, the image information is read from the printed paper 50 by a scanner or the like as the input device 30 before the extraction processing is performed by the information extraction device 40 in FIG. Done.

  The image information read by the input device 30 is output to the information extraction device 40 as a container signal in which the information is embedded. The container signal input to the information extraction device 40 in this way is highly likely to have a waveform that is changed (attacked) in the propagation process. Therefore, in the first embodiment, extraction processing of information embedded in the container signal is performed by hardware processing of the signal processing unit 101 or software processing of the CPU 100 in the information extraction device 40 of FIG. As shown in FIG. 4, the functional configuration of the information extraction apparatus 40 includes a similarity measurement unit 300, an information extraction unit 301, a statistical extraction unit 302, and the like.

  In the information extraction processing according to the first embodiment, as shown in FIG. 6, the similarity (here, cross-correlation) between the container signal and the embedded waveform is measured (step S500), and based on the correlation. In this process, the embedded information is extracted (step S501).

  The similarity measurement is performed by the similarity measurement unit 300 of the information extraction device 40, but is substantially the same as the processing in the similarity measurement unit 200 of the information embedding device 10 described above. First, as illustrated in FIG. 9B, the similarity measurement unit 300 of the information extraction device 40 performs a cross-correlation between a waveform in which a change (attack) is added to a container signal in which information is embedded and an embedded waveform. Is measured (step S500).

  Based on the measurement result, the information extraction unit 301 of the information extraction device 40 extracts information embedded in the container signal. For example, as shown in FIG. 9-2, data is extracted as 1 if the cross-correlation value is equal to or greater than a predetermined value Rd (point A), and is 0 if it is equal to or less than a predetermined value −Rd (point B).

  The extracted data may be embedded data, but immediately after embedding, even if it exceeds the predetermined value Rd, if the cross-correlation value amplitude decreases due to a change (attack) in the propagation process, it exceeds the predetermined value Rd. There may be no local missing data section. Therefore, by using the fact that the same data is repeatedly embedded as shown in the embedded data string of the present invention in FIG. Even if there is a local data missing section, it is possible to compensate for this, and correct embedded data can be extracted (step S501).

[Information Embedded Program Information Extraction Program]
The information embedding program executed by the information embedding device 10 of the first embodiment and the information extracting program executed by the information extracting device 40 are an installable format or an executable format file and are a CD-ROM. , Recorded on a computer-readable recording medium such as a flexible disk (FD), a CD-R, and a DVD (Digital Versatile Disk).

  The information embedding program executed by the information embedding device 10 of the first embodiment and the information extraction program executed by the information extracting device 40 are stored on a computer connected to a network such as the Internet. Alternatively, the program may be provided by being downloaded via a network. Further, the information embedding program executed by the information embedding device of the first embodiment and the information extraction program executed by the information extracting device 40 are configured to be provided or distributed via a network such as the Internet. Also good.

  In addition, the information embedding program and the information extraction program of the first embodiment may be provided by being incorporated in advance in a ROM or the like.

  The information embedding program executed by the information embedding device of the first embodiment and the information extraction program executed by the information extracting device 40 are the above-described units (similarity measuring unit 200, embedding determining unit 201). , A signal modulation unit 202, a similarity measurement unit 300, an information extraction unit 301, and a statistical extraction unit 302). As actual hardware, a CPU (processor) 100 stores information from the recording medium. By reading and executing the embedded program, the above-described units are loaded on the main storage device, and the similarity measuring unit 200, the embedding determining unit 201, the signal modulating unit 202, the similarity measuring unit 300, the information extracting unit 301, the statistics An extraction unit 302 is generated on the main storage device.

  As described above, according to the first embodiment of the present invention, even if data is embedded in a container signal and a data string extracted from the container signal after propagation has a data error or data clipping. Thus, it becomes possible to accurately synchronize symbols and to extract correct data. Further, in the comparative example embedded data string shown in FIG. 10, specific data (2, 3, etc.) is embedded for synchronization, but in the first embodiment, as shown in FIG. Since synchronization is achieved using a non-modulation section (expressed by X) in which the signal is not modulated, versatility applicable to various container signals can be maintained, and image quality degradation can be further reduced.

(Second Embodiment)
In the first embodiment, the embedded data string of FIG. 11 used for embedding cannot cope with the rotation attack. This is because the measured non-modulation interval (X) for synchronization is a target of rotation, and it is impossible to determine whether the embedded data string has been rotated or not just by looking at the non-modulation interval for synchronization. Therefore, in the second embodiment, the configuration of the information embedding device 10 and the information extracting device 40 of the first embodiment is almost the same, but the synchronization data of the embedded data string used is the same. It is characterized in that the configuration is not rotated.

  FIG. 14 is a diagram illustrating an example of a data string in which at least three non-modulation sections for synchronization are provided. In the second embodiment of the present invention, as shown in the one-dimensional data string of FIG. 14, in addition to the eight embedding intervals for embedding information, there are at least three synchronization intervals (non-modulation interval X). Is provided. Thereby, when extracting data from the container signal in which information is embedded, it is possible to determine whether the data is rotating by looking at the extracted non-modulation section (X). That is, it can be made non-rotating.

  This is the same even when the embedding target is a two-dimensional signal as shown in the two-dimensional data string of FIG. In the two-dimensional data string of FIG. 14, one symbol is formed by three synchronization intervals (non-modulation interval X) and 13 embedding intervals.

  As described above, according to the second embodiment of the present invention, in addition to the embedding interval for embedding information, embedded data in which three or more synchronization non-modulation intervals are provided at fixed positions within one cycle. By using a column, rotation can be made non-target, so that even if there is a rotation attack in the propagation process of the container signal, the correct synchronization state can be determined.

(Third embodiment)
The embedded data string according to the third embodiment is an embedded data string corresponding to a rotation attack as in the second embodiment. However, by using a different configuration, the data can be cut or rotated. The feature is that even if there is an attack, it is possible to synchronize with high accuracy and extract correct information.

  FIG. 15 is a diagram showing all patterns of an embedded data string composed of four synchronization non-modulation sections and eight embedding sections. FIG. 16 shows the positions of the synchronization non-modulation sections in FIG. It is a figure which shows all the patterns of the embedding data sequence which made it the dispersion | distribution and made the number of overlap of a non-modulation area below a predetermined number. In each of the data strings shown in FIGS. 15 and 16, one symbol is associated with a total of 12 sections including 4 synchronization non-modulation sections and 8 embedding sections.

  When extracting the embedded data embedded in the container signal, a total of 24 types of patterns can be considered by combining 12 types of clipping and 2 types of inversion in 12 sections of embedded data. In FIG. 15 and FIG. 16, all 24 patterns are shown.

  The correct phase and direction (synchronized state) can be obtained by examining which of the 24 patterns matches the observed pattern of the non-modulation section for synchronization. 15 and 16, the upper left pattern and each 24 pattern are compared, and the number of synchronization non-modulation sections (X) overlapping at the same position is shown on the right side of each pattern. The number of overlaps between the upper left patterns is of course the maximum of 4 because they are the same pattern. A non-modulation section for synchronization is set so that none of the remaining 23 patterns have a match number of four. The process of obtaining the synchronization state using this characteristic can be said to be a process of examining a pattern in which the number of overlaps in the non-modulation period for synchronization is maximum.

  Therefore, in FIG. 13, the number of extracted “0” and “1” is less than the predetermined number (k) due to data cutting, rotation attack, etc., even though it is a symbol embedding section, Consider the case where it is estimated as a non-modulation interval. In other words, this is a case where one more unmodulated section for synchronization is extracted than expected.

  As shown in the pattern on the right side of FIG. 15, when the falsely estimated false non-modulation section for synchronization comes in any section indicated by the black triangle mark ▼, there are two overlapping numbers of four patterns. It becomes impossible to determine which is the correct synchronization state.

  On the other hand, in the case of FIG. 16, the erroneous determination as shown in FIG. 15 does not occur. This is because, in the pattern of FIG. 16, the positions of the non-modulation sections are dispersed so that the number of overlaps of the non-modulation sections in the asynchronous state is equal to or less than a predetermined number (here, 2 which is two less than the maximum number 4). Because. As a result, even if the non-modulation section for synchronization is estimated incorrectly, the maximum number is not yet reached, so that the correct synchronization state can be determined.

  Thus, according to the third embodiment of the present invention, a plurality of non-modulating sections for synchronization are provided at fixed positions within one period, and the number of overlapping non-modulating sections in the asynchronous state is greater than the maximum number. By distributing the positions of the non-modulation sections so as to decrease by two or more, the correct synchronization state can be determined even if there is a data cut or rotation attack in the container signal propagation process.

  In each of the above-described embodiments, as illustrated in FIG. 1, the information embedding device 10, the output device 20, the input device 30, and the information extraction device 40 are described as examples. The embedding device 10 and the information extraction device 40 may be configured as one device.

  Further, in each of the above-described embodiments, the information embedding device 10 and the output device 20 or the information extraction device 40 and the input device 30 shown in FIG. 1 are configured separately, but each may be configured as one device. good.

  An information embedding device, an information extracting device, an information embedding method, an information extracting method, an information embedding program, and an information extracting program according to the present invention include copyright holder information, accessor information, etc. in content data of audio data and image data. This is useful for digital watermark technology that embeds. Information that can embed information while reducing the impact of content data quality degradation and reliably extract the embedded information, especially in situations where various changes (attacks) are applied to the content data from the outside. It is suitable for an embedding device, an information extracting device, an information embedding method, an information extracting method, an information embedding program, and an information extracting program.

1 is a system configuration diagram including an information embedding device and an information extraction device of the present invention. FIG. 2 is a configuration block diagram of the information embedding device and the information extracting device in FIG. 1. It is a functional block diagram in CPU or a signal processing part in the information embedding device of FIG. It is a functional block diagram in CPU or a signal processing part in the information extraction device of FIG. It is a main flowchart of the information embedding process concerning this Embodiment. It is a main flowchart of the information extraction process concerning this Embodiment. It is a wave form diagram of embedding data. It is a wave form diagram of a container signal before embedding. It is a wave form diagram of the container signal after embedding. It is a cross-correlation waveform diagram before embedding. It is a cross-correlation waveform diagram at the time of extraction after embedding. It is a figure which shows various embedding data strings. It is a figure which shows the example of a data sequence from the embedded data sequence to an extraction data sequence concerning this Embodiment. It is a statistics figure which took statistics of the number of embeddings for every period. It is a statistics figure which took statistics of the number of extraction for every period. It is a figure which shows the example of a data row | line | column provided with at least three non-modulation area for a synchronization. It is a figure which shows all the patterns of the embedding data sequence which consists of four non-modulation areas for a synchronization and 8 embedding areas. FIG. 16 is a diagram illustrating all patterns of an embedded data string in which the positions of the synchronization non-modulation sections in FIG. 15 are dispersed so that the number of overlaps of the non-modulation sections is a predetermined number or less.

Explanation of symbols

10 Information embedding device 100 CPU
101 signal processing unit 102 ROM (HDD)
103 RAM
104 Operation / Display Unit 105 External I / F Unit 200 Similarity Measurement Unit 201 Embedding Determination Unit 202 Signal Modulation Unit 300 Similarity Measurement Unit 301 Information Extraction Unit 302 Statistical Extraction Unit

Claims (8)

An information embedding device that divides content data into predetermined intervals, sets a predetermined plurality of connected segments as one cycle, and modulates the content data based on embedded data corresponding to the cycle,
In the embedded data, in addition to the embedded information, a non-modulation section for synchronization in which the content data is unmodulated is provided in one or more sections at a fixed position within one cycle,
Similarity measuring means for measuring the degree of similarity between the embedded data and the content data for each section;
Embedding determination means for determining whether to embed the embedded data according to the similarity for each section measured by the similarity measuring means;
Signal embedding means for performing signal modulation when embedding the embedding data, and in the embedding information section of the embedding data, using the similarity measurement means, the embedding determination means, and the signal modulation means, An information embedding apparatus, wherein the content data is processed for each section to embed embedded data, and signal modulation processing of the content data is not performed in the non-modulation section.   2. The embedded data is provided with three or more non-modulating sections for synchronization in which the content data is not modulated in addition to the embedded information at fixed positions within one period. Information embedding device. In the embedded data, in addition to the embedded information, a non-modulation section for synchronization in which the content data is unmodulated is provided in a plurality of sections at fixed positions within one cycle,
2. The information embedding device according to claim 1, wherein the positions of the non-modulation sections are distributed so that the number of overlaps of the non-modulation sections in an asynchronous state is equal to or less than a predetermined number. An information extraction device that extracts embedded data embedded in content data using the information embedding device according to any one of claims 1 to 3,
Similarity measuring means for measuring the degree of similarity between the embedded data and the content data for each section;
Information extracting means for extracting embedded information according to the similarity for each section measured by the similarity measuring means;
Statistical extraction means for extracting the embedded data by taking statistics of the number of extractions in the same section while synchronizing in the non-modulation section using the extraction information extracted by the information extraction means. Feature information extraction device. An information embedding method that divides content data into predetermined intervals, sets a predetermined plurality of connected segments as one cycle, and modulates the content data based on embedded data corresponding to the cycle,
In the embedded data, in addition to the embedded information, a non-modulation section for synchronization in which the content data is unmodulated is provided in one or more sections at a fixed position within one cycle,
A similarity measurement step of measuring the similarity between the embedded data and the content data for each section;
An embedding determination step for determining whether to embed the embedded data according to the similarity for each section measured by the similarity measuring step;
A signal modulation step for performing signal modulation when embedding the embedded data, and in the embedded information section of the embedded data, the embedded data and the content are obtained by the similarity measurement step, the embedded determination step, and the signal modulation step. An information embedding method, wherein embedded data is embedded by processing data for each section, and signal modulation processing of the content data is not performed in the non-modulation section. An information extraction method for extracting embedded data embedded in content data using the information embedding method according to claim 5,
A similarity measurement step of measuring the similarity between the embedded data and the content data for each section;
An information extraction step of extracting embedded information according to the similarity for each section measured by the similarity measurement step;
A statistical extraction step of extracting the embedded data by taking statistics of the number of extractions in the same interval while synchronizing in the non-modulation interval using the extraction information extracted in the information extraction step. Information extraction method.   An information embedding program for causing a computer to execute the information embedding method according to claim 5.   An information extraction program for causing a computer to execute the information extraction method according to claim 6.

Images