JP2007174181A - Document processing apparatus, document processing method, and document processing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a document processing apparatus, a document processing method, and a document processing program capable of recording attached information with high invisibility, high indivisibility from contents, and high robustness against repetitive copying to a paper document. <P>SOLUTION: A parameter value sequence extract section 401 extracts a parameter value sequence associated with a document style from a character sequence on a document received from a first document input section 10 as an information embedding object, a code sequence group generating section 402 generates a reference code sequence and an information embedding code sequence, an information coding section 403 represents embedding information received from an embedding information input section 30 as a phase difference between the reference code sequence and the information embedding code sequence, and the sum of both the code sequences is used for coding information sequence. Then a correction parameter value sequence generating section 404 composes the coding information sequence with the parameter value sequence to produce a corrected parameter value sequence, and a document correction section 405 corrects document data by the corrected parameter value sequence to embed information to the document. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a document processing apparatus, a document processing method, and a document processing program, and particularly suitable for use in an image forming apparatus such as a copying machine, a printer, a facsimile machine, or a multifunction machine having these functions combined. The present invention relates to an apparatus, a document processing method, and a document processing program.

  In recent years, in offices, computerization of storing documents as electronic data has become commonplace, and image forming apparatuses such as copiers and printers that convert electronic data into paper documents have become widespread. Along with this, leakage of confidential information from inside the company has become a serious problem, with highly confidential information such as customer lists and information on products under development leaking out of the company. In particular, electronic documents (electronic data) can be easily duplicated exactly as the original, and can be taken out to the outside very easily using a network. It was.

  However, due to recent advances in encryption technology, authentication technology, access restriction technology, etc., it is relatively difficult to take out electronic documents if appropriate measures are taken. On the other hand, once an electronic document is printed on paper and converted into a paper document, there is no way to prevent copying or taking it out of the office. Therefore, the current situation is that the progress of countermeasures against the leakage of confidential information of paper documents is more noticeable than that of electronic documents.

  As one of countermeasures against confidential leakage of a paper document, there is a technique for adding tracking information to the paper document in advance. In this technology, information such as output person information, output device information, output date and time when an electronic document is printed out and converted into a paper document is recorded on the paper document. With this tracking information addition technology, even if a paper document on which confidential information is printed flows out of the company, the outflow source can be identified from the tracking information recorded on the paper document.

  Such tracking information is desired to be highly invisible and inseparable from the content. For example, when tracking information is recorded with a highly visible recording method such as a barcode, it is clear where the tracking information is recorded on the paper document, so there is a risk that the recorded tracking information may be decrypted or altered. Get higher. Further, since the barcode is separable from the content, it is possible to easily prevent the outflow source from being identified by cutting or painting the barcode portion.

  The tracking information is also desired to be highly resistant to repeated copying. This is because the leaked confidential information is not only distributed as it is, but it can also be distributed as a duplicate copy of the original, as a secondary copy or as a tertiary copy.

  Conventionally, as a technique with high invisibility in a method of recording additional information such as tracking information on a paper document, for example, a technique of embedding information by shifting the character position and changing the blank length before and after the character Is known (see, for example, Patent Document 1).

  In this prior art, when the space lengths before and after the character to be embedded with information bits are P and S, the character position is such that P> S if the information bit is “1”, and P <S if the information bit is “0”. Information is embedded by shifting. Additional information is embedded in the document by performing this operation on a number of characters corresponding to the number of bits of the embedded information. At the time of embedding, if the character positions before and after the punctuation marks are shifted, it is easy to see them.

  A technique for embedding information by changing the shape of a character pattern, such as the size of the character pattern and the font type, is also known (see, for example, Patent Document 2).

  In this conventional technique, characters such as “no” and “ha”, which frequently appear in a normal document, are set as information embedding target characters. Information is embedded in the target characters in order from the beginning of the document. Only when the information bit is “1”, the embedded information is expressed by replacing it with another font or increasing the character width by about 10%. When the information bit is “0”, nothing is done. When restoring the embedded information, it is compared with the original document before the information is embedded, and it is interpreted that 1 is embedded if the character pattern is different and 0 is embedded if they are the same.

JP2003-230001 (in particular, refer to pages 5-7) Japanese Patent Laid-Open No. 2001-251490 (see particularly, pages 4 to 6)

  However, with the information embedding methods described in Patent Documents 1 and 2, the information tracking function, which is an expected function, cannot be sufficiently exerted on paper documents that have been copied many times. This is because, in a paper document that has been copied many times, image quality deterioration such as character thickening or thinning, or noise or roughness on the character occurs.

  For example, in the method described in Patent Document 1 in which information is embedded in the shift amount of the character position, the space length between characters changes due to image quality deterioration due to repeated copying, and the embedded information may not be restored correctly. In order to avoid this, it is necessary to increase the shift amount of the character position so that the size relationship of the space length before and after the character is maintained even if the space length changes slightly. However, if the shift amount of the character position is increased, the change made to the document becomes visually apparent, and the invisibility of the embedded information is impaired.

  In addition, in a document in which the character spacing is adjusted according to the characters before and after the character string so that the character string looks natural, that is, a document subjected to so-called kerning, the character spacing is not constant. If an attempt is made to change the blank length before and after a certain character so that P <S, the character position must be shifted greatly, and the invisibility of the embedded information is also lost.

  Similarly, the method described in Patent Document 2 has low resistance to image quality degradation due to repeated copying. In other words, in order to determine the difference in shape such as the character pattern size and font type from a paper document whose image quality has deteriorated, it is necessary to change the character size greatly or use fonts with greatly different shapes. . However, if the shape of the character pattern is greatly changed, the change made to the document becomes visually apparent, and the invisibility of the embedded information is lost. In addition, there is a problem that information cannot be embedded in a document in which the frequency of appearance of characters to be embedded such as “no” and “ha” is small.

  The present invention has been made in view of the above problems, and the object of the present invention is to provide additional information that is highly invisible and inseparable from content with respect to a paper document and has high robustness to repeated copying. An object is to provide a document processing apparatus, a document processing method, and a document processing program that enable recording.

  In order to achieve the above object, in the present invention, when embedding additional information in document data consisting of a character string, a parameter value series relating to the document appearance in the character string is extracted from the input document data, Generating a code sequence and an information embedding code sequence, generating the input additional information as an encoded information sequence represented by a phase difference between the reference code sequence and the information embedding code sequence; A modified parameter value series is generated by synthesizing the encoded information series with a value series, and the document data is modified by the modified parameter value series.

  By embedding additional information in the document in this way, it is possible to embed additional information such as tracking information in an invisible state, not only in the digital data of the document but also in a text document printed on paper. At the same time, the additional information can be directly embedded in the parameter value of the document content itself relating to the document format obtained from the character string on the document. Further, since the additional information is embedded in the character image itself, even if the copy is repeatedly made, the embedded information is stored as long as the character itself does not disappear. In addition, since information is embedded as a phase difference between the reference code sequence and the information embedding code sequence, even if a character cutout error occurs when the character cutout process is performed when the embedded information is restored, the information is embedded. The phase difference between the reference code sequence and the information embedding code sequence, that is, the embedding information is unchanged.

  According to the present invention, the invisibility and the inseparability from the content are high with respect to the paper document, and even if the copy is repeatedly made, the embedded information is preserved as long as the characters themselves do not disappear. On the other hand, it is possible to record additional information with high robustness. In addition, when performing character segmentation processing, even if a character segmentation error occurs, the phase difference between the embedded reference code sequence and the information embedding code sequence remains unchanged. Even if it exists, the embedded information can be reliably restored.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram showing an outline of the configuration of a document processing apparatus according to an embodiment of the present invention. As shown in FIG. 1, the document processing apparatus 1 according to the present embodiment includes a first document input unit 10, a second document input unit 20, an embedded information input unit 30, a process control unit 40, a document output unit 50, and an information output. The configuration includes the portion 60.

  The first document input unit 10 is for inputting document data to be embedded with additional information to the processing control unit 40. Document data input from the first document input unit 10 to the processing control unit 40 includes page description language such as PostScript, electronic document data in a word processor output format, or raster data of a document photographed by a scanner or a digital camera. Is mentioned.

  The first document input unit 10 exchanges data via, for example, a hard disk drive storing document data, a mass storage device such as a DVD (Digital Video Disc) -RAM / ± RW / ± R drive, a network, or the like. It can be realized by a data transfer device to be performed, a scanner device and its control device, a memory reader device for taking out an image stored in a memory (card) of a digital camera, its control device, or the like.

  The second document input unit 20 is for inputting raster data of a document in which additional information is embedded to the processing control unit 40. Examples of document raster data include image data of a paper document photographed by a scanner or a digital camera.

  The embedded information input unit 30 is used to input additional information to be embedded in the document data input from the first document input unit 10 to the processing control unit 40. As the additional information input from the embedded information input unit 30 to the processing control unit 40, any digital data such as numbers, characters, URL (Uniform Resource Locator), multimedia data such as voice and images, and the like is used. be able to.

  The embedded information input unit 30 includes, for example, a keyboard for inputting characters and the like, a hard disk drive device (HDD) for inputting multimedia data, and a DVD (Digital Versatile Disk) -RAM / ± RW / ± R drive. It can be realized by a large-capacity storage device such as a data transfer device that exchanges information via a network or the like.

  The processing control unit 40 includes a CPU (Central Processing Unit) 41, an I / O circuit 42, a ROM 43, a RAM 44, and an HDD (Hard Disk Drive Device) 45, and these components are mutually connected via a bus line 46. It is the structure connected so that communication was possible.

  In the process control unit 40, the CPU 41 controls the entire process including the calculation process. The I / O circuit 42 manages input / output to / from peripheral devices such as the first document input unit 10, the second document input unit 20, the embedded information input unit 30, and the document output unit 50 and the information output unit 60. The ROM 43 stores processing programs for various processes executed under the control of the CPU 41. The RAM 44 is a primary storage device used when executing the various processes. The HDD 45 stores document data processed under the control of the CPU 41.

  The document output unit 50 outputs the document data processed by the processing control unit 40 in a predetermined format. From the document output unit 50, for example, a document printed on paper, a page description language such as PostScript, electronic document data in a word processor output format, or a file in a raster data format of a document is output as an output result. For this reason, the document output unit 50 is realized by, for example, a printer device and its control device, a read / write device such as a magnetic disk or a memory card and its control device, or a data transfer device that exchanges data via a network or the like. Is done.

  The information output unit 60 outputs additional information restored by the processing control unit 40. For example, a CRT (Cathode Ray Tube) and its control device, a printer device and its control device, a magnetic disk, a memory card, etc. The read / write device and its control device, or a data transfer device that exchanges data via a network or the like.

(Information embedding process)
Next, information embedding processing, which is a feature of the present invention and is executed in the document processing apparatus 1 having the above-described configuration, will be described. This information embedding processing is executed by the information embedding processing unit 400 built in the processing control unit 40.

  FIG. 2 is a block diagram illustrating an example of a functional configuration of the information embedding processing unit 400. As shown in FIG. 2, the information embedding processing unit 400 according to this example includes a parameter value sequence extraction unit 401, a code sequence group generation unit 402, an information encoding unit 403, a correction parameter value sequence generation unit 404, and a document correction unit 405. It is the composition provided with.

  The parameter value series extraction unit 401 analyzes the electronic document data input from the first document input unit 10 or the raster data of the document, and obtains a character interval, a character position, a word interval, a line interval obtained from a character string on the document, A series of parameter values related to the document appearance such as character height, character width, and character inclination is obtained.

  The code sequence group generation unit 402 generates encoded information by encoding additional information input from the embedded information input unit 30, and includes a single reference code sequence and one or more information embedding codes. A code sequence group of the sequence is generated.

  The information encoding unit 403 generates an encoded information sequence obtained by encoding the additional information input from the embedded information input unit 30 with the code sequence group generated by the code sequence group generating unit 402. In this information encoding unit 403, additional information is represented by a phase difference between the reference code sequence and the information embedding code sequence, and an information embedding code sequence is added to the reference code sequence by a phase difference corresponding to the additional information. The phase shift is performed, and the information embedding code sequence and the reference code sequence after the phase shift are added to obtain an encoded information sequence.

  The correction parameter value series generation unit 404 corrects the parameter value series related to the input document extracted by the parameter value series extraction unit 401. In this modified parameter value sequence generation unit 404, each symbol value of the parameter value sequence is modified according to the corresponding symbol value of the encoded information sequence obtained by the information encoding unit 403 (hereinafter, this modified parameter (The value series is described as “corrected parameter value series”.)

  The document correction unit 405 corrects the appearance of the document input from the first document input unit 10 in accordance with each symbol value of the correction parameter value series generated by the correction parameter value series generation unit 404. In the document correction unit 405, parameter values relating to the document appearance such as character spacing, character position, word spacing, line spacing, character height, character width, and character inclination obtained from the character string on the corrected document are corrected. Correction is performed so as to be equal to each symbol value of the parameter value series.

  Regarding each component of the information embedding processing unit 400 described above, that is, the parameter value sequence extraction unit 401, the code sequence group generation unit 402, the information encoding unit 403, the correction parameter value sequence generation unit 404, and the document correction unit 405, the PC It can be realized using a computer device that realizes an information storage processing function, an image processing function, an arithmetic processing function, and the like by executing a predetermined program, such as a (personal computer).

  In this case, a predetermined program necessary for realizing the parameter value sequence extraction unit 401, the code sequence group generation unit 402, the information encoding unit 403, the correction parameter value sequence generation unit 404, and the document correction unit 405 is stored in advance in the PC. It is possible to install it. However, it may be provided by being stored in a computer-readable storage medium instead of being installed in advance, or distributed via wired or wireless communication means. .

  That is, the document processing apparatus having the above-described configuration functions as a document processing apparatus using the computer connected to the first document input unit 10, the second document input unit 20, the embedded information input unit 30, the document output unit 50, and the information output unit 60. This can also be realized by a document processing program.

(Processing procedure of document processing method)
Next, a processing procedure of information embedding processing (including a case where it is executed by a document processing program) executed in the information embedding processing unit 400 having the above configuration will be described.

  FIG. 3 is a flowchart illustrating an example of the procedure of information embedding processing in the information embedding processing unit 400. This information embedding process is executed under the control of the CPU 41. In the following example, a case will be described in which the additional information to be embedded is n-bit binary data, and the character spacing is used as a parameter relating to the document format for embedding the additional information.

  First, when electronic document data to be embedded with additional information is input from the first document input unit 10 to the parameter value series extraction unit 401 (step S11), the parameter value series extraction unit 401 inputs the electronic document data. The electronic document data and the raster data of the document are analyzed to obtain an interval between characters constituting the character string on the document, and this is used as a series of parameter values relating to the document appearance (step S12).

Specifically, as shown in FIG. 4, a series in which the intervals between the characters constituting the character string on the document input from the first document input unit 10 are arranged is a parameter value series a (k ) (K = 0, 1, 2,..., L−1). As the character spacing, the character spacing at any position on the document may be used, but in this example, L character spacings from the beginning of the document (L ≧ 2 ⁿ ) are used. When the number of character intervals obtained from one line on the document is less than L, a character value parameter having a sequence length L is obtained by concatenating character interval parameters obtained from a plurality of subsequent lines.

  If the input document data is a page description language such as PostScript or electronic document data in a word processor output format, the character spacing can be directly obtained by analyzing the electronic document data. The method of analyzing the electronic document data is not limited to a specific method, and an appropriate method can be used according to the format of the electronic document data.

  In addition, when the input document data is raster data of a document taken by a scanner or digital camera, a character segmenting process generally performed as a character recognition pre-process is performed, Can be set as the character spacing. The character cut-out processing is not limited to a specific method, and any generally used appropriate method can be used.

  In this example, the character spacing is used as a parameter related to the document format for embedding information. However, the present invention is not limited to this, and the parameters related to the document format include character position, word spacing, line spacing, and character. It is also possible to use height, character width, character inclination, or some combination of these.

  Next, in the information encoding unit 403, n-bit binary data input from the embedded information input unit 30 is encoded by the code sequence group generated by the code sequence group generation unit 402 to obtain an encoded information sequence. (Step S13). Here, the n-bit binary data input from the embedded information input unit 30 is represented by b.

  Here, details of the information encoding process executed in step S13, that is, the process of the information encoding unit 403 for encoding the embedded information and obtaining the encoded information sequence will be described.

  As the code sequence generated by the code sequence group generation unit 402, a code sequence having a small average value and having randomness is used. Both the generated reference code sequence and the information embedding code sequence have sharp autocorrelation characteristics as shown in FIG. Further, a small cross-correlation between the reference code sequence and the information embedding code sequence is used.

  An example of a known code sequence having such properties is a Gold code sequence. However, the code sequence generated by the code sequence group generation unit 402 is not limited to the Gold code sequence, and may be any code sequence having the above properties.

Here, an example in which a binary sequence having a sequence length of L as one cycle is used as the code sequence generated by the code sequence group generation unit 402 will be described. However, the sequence length L is 2 ⁿ or more. Further, the reference code sequence is represented by ref (k), the information embedding code sequence is represented by pn (k), and the symbol value thereof takes any of ± 1.

The reference code sequence ref (k) and the information embedding code sequence pn (k) are expressed by the following equations (1) and (2), respectively.
ref (k) = {± 1} (1)
(K = 0, 1, 2,..., L−1 and L ≧ 2 ⁿ )
pn (k) = {± 1} (2)
(K = 0, 1, 2,..., L−1 and L ≧ 2 ⁿ )

Then, as shown in FIG. 6, n-bit embedded information b is encoded using the phase difference between these two code sequences ref (k) and pn (k). An encoded information sequence c (k) obtained by encoding n-bit embedded information b is expressed by the following equation (3).
c (k) = ref (k) + pn (k + b) (3)
(K = 0, 1, 2,..., L−1 and L ≧ 2 ⁿ )
However, in the above equation (3), the code sequence pn (k) is a periodic code sequence of period L (pn (k + L) = pn (k)).

The code of the sequence length L can express L states from the shift 0 to the shift L-1. Therefore, when the reference code sequence ref (k) having the sequence length L and the information embedding code sequence pn (k) are used, the reference code sequence ref (k) and the information embedding code sequence pn (k) Log ₂ L-bit information can be encoded using the phase difference.

  Returning to the flowchart of FIG. After the embedded information is encoded in step S13, the parameter value sequence of the document obtained in step S12, that is, the character spacing sequence a (k) is obtained in the modified parameter value sequence generation unit 404 by the encoded information obtained in step S13. Correction is made according to the series c (k) to obtain a correction parameter value series d (k) (step S14).

The corrected parameter value series d (k) is expressed by the following equation (4).
d (k) = a (k) + δ · c (k) (4)
In the above formula (4), δ is a constant indicating the embedding strength of information, and the invisibility of embedded information increases as the value becomes smaller.

  As will be described later, in the present embodiment, when a code sequence having a sequence length L of several hundred to several thousand is used, the information embedding strength δ in the above equation (4) is the symbol of the character interval sequence a (k). The value can be much smaller than the value. That is, if a code sequence having a long sequence length is used, the invisibility can be increased to such an extent that the embedded information can hardly be visually discriminated.

  Next, the document correction unit 405 corrects the appearance of the document data input from the first document input unit 10 using the correction parameter value series d (k) obtained in step S14 (step S15). In this document correction step, as shown in FIG. 7, processing for correcting the character spacing obtained from the character string on the corrected document to be equal to each symbol value of the correction parameter value series d (k) is performed. Is called.

  Specifically, when the original document data is a page description language such as PostScript or electronic document data in a word processor output format, the character spacing is made equal to each symbol value of the correction parameter value series d (k). Next, the description of the electronic document data is corrected.

  On the other hand, when the original document data is raster data of a document photographed by a scanner or a digital camera, it is cut out using the result of the character cutout process performed in the parameter value series extraction step (step S12). The character images are rearranged so that the distance between the character rectangles is equal to the symbol values of the correction parameter value series d (k).

  Finally, the document data in which the character spacing is corrected using the correction parameter value series d (k) is output from the document output unit 50 in the form of a paper document or electronic document data (step S16).

  By executing the above series of processing, a document in which additional information is embedded is obtained. According to the information embedding method according to the present embodiment, the additional information is embedded in an inseparable form from the document content as the character spacing of the character string, unlike the case where the additional information is written in the document with a barcode or the like. Therefore, if embedded information such as tracking information of confidential documents is embedded in the document as an additional information, the document content itself will be lost if an attempt is made to remove the embedded information in order to hinder the identification of the leaked source, thus providing high security. .

  In addition, the code sequence generated by the code sequence group generation unit 402 has a small average value of symbols and has randomness. Looking at equation (4) on the premise of this, the spacing a (k) between characters is randomly expanded or narrowed, and the amount of correction is ascertained by looking at a certain range (for example, one line). It turns out that the sum of becomes a small value. That is, since the length of each corrected line is almost the same as the line length of the original document, the invisibility of the information embedding process can be increased.

  As described above, in the information embedding processing unit 400 or the information embedding processing method according to the present embodiment, a series of parameter values related to document appearance is obtained from a character string on a document input as an information embedding target, and a reference code series and information An embedding code sequence is generated, embedding information is expressed as a phase difference between the two code sequences, and a reference code sequence and an information embedding code sequence are added to form an encoded information sequence. The encoded information series and the parameter value series are combined to obtain a corrected parameter value series, and the document format of the electronic document data or document image is changed in accordance with the corrected parameter value series, so that information is stored in the document. Since it is embedded, the following effects can be obtained.

  That is, additional information such as tracking information can be embedded not only in the digital data of the document but also in the text document printed on paper. In addition, since additional information is directly embedded in the parameter values of the document content itself relating to the document format such as character spacing, character position, word spacing, line spacing, character height, character width, etc. obtained from the character string on the document, the document content It is inseparable, making it difficult to tamper and remove information.

  In addition, since the method of embedding information in the character image itself is used, even if repeated copies are made, the embedded information (additional information) is saved as long as the characters themselves do not disappear. In addition, it has a feature that it is robust against repeated copying, that is, it is highly robust against repeated copying.

  In particular, since the information is embedded as the phase difference between the reference code sequence and the information embedding code sequence, even when a character segmentation error occurs during the character segmentation process performed when embedding information is restored, the information is embedded. Since the phase difference between the reference code sequence and the information embedding code sequence, that is, the embedded information is unchanged, the embedded information can be restored even when a character cut-out error occurs.

(Embedded information restoration process)
Next, an embedded information restoration process, which is a feature of the present invention and is executed in the document processing apparatus 1 according to this embodiment described above, will be described. This embedded information restoration processing is executed by an embedded information restoration processing unit 450 built in the processing control unit 40.

  FIG. 8 is a block diagram illustrating an example of a functional configuration of the embedded information restoration processing unit 450. As illustrated in FIG. 8, the embedded information restoration processing unit 450 according to the present example includes a parameter value sequence extraction unit 451, a code sequence group generation unit 452, and an embedded information restoration unit 453. The embedded information restoration processing unit 450 receives raster data of a document in which additional information is embedded from the second document input unit 20.

  The parameter value series extraction unit 451 has the same configuration as the parameter value series extraction unit 401 of the information embedding processing unit 400, for example, and is input from the second document input unit 20 and includes raster data and electronic data of a document embedded with additional information The document data is analyzed to obtain a series of parameter values relating to the document appearance such as character spacing, character position, word spacing, line spacing, character height, character width, and character inclination obtained from the character string on the document.

  The code sequence group generation unit 452 has the same configuration as the code sequence group generation unit 402 of the information embedding processing unit 400, for example, and the same reference code sequence as the reference code sequence generated by the information embedding processing unit 400 and the code sequence of the code sequence A shift version, one or more information embedding code sequences that are the same as the one or more information embedding code sequences generated by the information embedding processing unit 400, and a shift version of the information embedding code sequence Generate.

  The embedded information restoration unit 453 restores the embedded information from the parameter value series related to the document format of the document embedded with the additional information extracted by the parameter value series extraction unit 451. In the embedded information restoration unit 453, the parameter value sequence extracted by the parameter value sequence extraction unit 451 is subjected to decoding processing with each of the code sequence groups generated by the code sequence group generation unit 452. , Restore the embedded additional information.

  Here, the embedding information restoration processing unit 450 according to the present example mainly aims to restore the embedding information from the image of the document embedded in the additional information and output to the paper. It is assumed that embedded information (additional information) can be restored from page description language such as PostScript or electronic document data in a word processor output format. Therefore, the document data input from the second document input unit 20 is not limited to raster data, and may be electronic document data in a page description language or a word processor output format.

  Next, an information embedding process executed in the embedded information restoration processing unit 450 configured as described above will be described.

  FIG. 8 is a flowchart illustrating an example of the procedure of the embedded information restoration process in the embedded information restoration processing unit 450. This embedded information restoration process is executed under the control of the CPU 41 (see FIG. 1). In the following, a case will be described as an example where the n-bit embedded information b embedded in a paper document is restored by the above-described information embedding processing procedure. Assume that the input paper document includes image quality degradation due to repeated copying.

  First, when document data of a paper document in which information is embedded is input as raster data from the second document input unit 20 such as a scanner or a digital camera (step S21), the parameter value series extraction unit 22 inputs the second document. The raster data of the document input from the unit 20 is analyzed, the interval between the characters constituting the character string on the document is obtained, and this is used as a series of parameter values relating to the document format (step S22).

  Here, as an example, a case will be described in which embedded information is restored from an image of a document output on a paper in which additional information is embedded, but a page description language such as PostScript before output on paper or a word processor output format Electronic document data may be used as input data. In either case, the method for extracting the character interval series is the same as that for information embedding, and the description thereof is omitted.

The character interval series a ′ (k) obtained by the parameter value series extraction unit 22 can be expressed by the following expression (5) using the expressions (3) and (4) described above.
a ′ (k) = d (k) + e (k)
= A (k) + δ · c (k) + e (k)
= A (k) + δ · ref (k) + δ · pn (k + b) + e (k) (5)
In the above equation (5), e (k) represents an error component due to character thickening or thinning due to repeated copying, or noise or roughness on the character, and a detection error component in the character interval detection processing.

  Next, in the code sequence group generation unit 23, the same reference code sequence ref (k) as the reference code sequence ref (k) used at the time of information embedding, and a shifted version of the same reference code sequence ref (k), Similarly, an information embedding code sequence pn (k) identical to the information embedding code sequence pn (k) used at the time of information embedding and a shifted version of the same information embedding code sequence pn (k) are generated (steps). S23). Since the code sequence group generated here is the same as that shown in the equations (1) and (2) described above, description thereof is omitted.

  Next, the embedding information restoration unit 24 restores the n-bit embedding information b embedded from the character interval series a ′ (k) (step S <b> 24). Details of the processing here will be described with reference to FIG.

  FIG. 10 shows a case where the h-th character segmentation fails and is segmented and segmented with the previous character. Since there is a character segmentation error, the character interval series a '(k) has a phase shift after the h-th element.

  As described above, in the information embedding process according to the present embodiment, as described above, even when a phase shift occurs in the character interval sequence a ′ (k) due to the extraction error, as described above, the reference code sequence ref (k). Since the additional information is embedded using the phase difference between the information embedding code sequence pn (k), the embedded information (additional information) can be correctly restored. This is because a phase shift due to a cut-off error occurs in the same amount in both the reference code sequence ref (k) and the information embedding code sequence pn (k), and the phase difference is unchanged. This principle will be described with reference to FIG.

FIG. 11 shows the state of information restoration when the h-th character segmentation fails and is segmented and segmented with the previous character. A character interval series a ′ (k) is obtained from the result of character segmentation. Since the character up to the (h-1) th character is correctly cut out, the character interval series a ′ (k) is obtained from the equation (5):
a ′ (k) = a (k) + ref (k) + pn (k + b) + e (k) (6)
It becomes. However, k <h, and the embedding strength δ of information is set to δ = 1 to simplify the equation.

On the other hand, since the h-th character is erroneously combined with the next character and cut out, for k ≧ h, the character interval series a ′ (k) is expressed by the following equation (7).
a ′ (k) = a (k + 1) + ref (k + 1)
+ Pn (k + 1 + b) + e (k + 1) (7)

  For such a character interval series a '(k), the embedded information restoration unit 453 restores information in the following procedure.

<Step 1>
The partial correlation between the character spacing sequence a ′ (k), the reference code sequence ref (k) and the shifted version of the reference code sequence ref (k) is calculated. In FIG. 11, the calculation range of the partial correlation is indicated by a dotted rectangular window, and the window width is 3. Actually, the window width needs to be several tens to several hundreds, but here it is 3 for simplicity.

  When the partial correlation between the character interval sequence a ′ (k), the reference code sequence ref (k), and the shifted version of the reference code sequence ref (k) is calculated at window position 0 in FIG. 11, the reference code sequence ref (k ), That is, the correlation with {ref (0), ref (1), ref (2)} is maximized. Since the reference code sequence has a sharp autocorrelation characteristic as shown in FIG. 5, the correlation with the reference code sequence ref (k) is larger than the correlation with other shift versions.

  According to equation (6), the character interval series a ′ (k) includes other signal components a (k), pn (k + b), and e (k). These signal components a (k) and pn ( The correlation value between k + b), e (k) and the reference code sequence ref (k) is so small that it can be ignored. This is because the reference code sequence ref (k) is a sequence having randomness, and the sequence having randomness has a low correlation with any code sequence other than itself.

<Step 2>
Next, the partial correlation between the character interval sequence a ′ (k), the information embedding code sequence pn (k) and the shifted version of the information embedding code sequence pn (k) is calculated. For the same reason as described in Step 1, the partial correlation at window position 0 is the largest value for pn (k + b).

<Step 3>
The phase difference between the codes for which the maximum partial correlation is obtained in step 1 and step 2 is obtained, and this is restored as embedded information. Speaking of the window position 0, it is additional information in which the phase difference between the reference code sequence ref (k) and the information embedding code sequence pn (k + b), that is, b-0 = b is embedded.

<Step 4>
The process from step 1 is repeated by shifting the window position. For example, for window position 1, since the correlation between the reference code sequence ref (k + 1) and the information embedding code sequence pn (k + 1 + b) is maximized, the phase difference, that is, (1 + b) -1 = b is embedded. It becomes additional information.

  By repeating the procedure from Step 1 to Step 4 described above, the phase difference at each window position is obtained. Of the phase differences at each window position, the phase difference that appears most frequently is restored as embedded information. That is, the embedded information is restored based on the majority rule.

  A value with a large partial correlation between the other signal components a (k) and e (k) included in the character spacing sequence a ′ (k) and the reference code sequence ref (k) and the information embedding code sequence pn (k). Although the probability is low, correct information is restored by the majority rule.

  The above is the information restoration procedure in the embedded information restoration processing unit 20, but as shown in FIG. 11, it will be shown that information can be correctly restored even when character segmentation fails. In FIG. 11, the restoration of information after the position where there is a clipping error is as follows.

  The character interval series a ′ (k) at the window position h in FIG. 11 is expressed by Expression (7). Accordingly, when the partial correlation between the character interval sequence a ′ (k), the reference code sequence ref (k), and the shifted version of the reference code sequence ref (k) is calculated at the window position h, the reference code sequence ref (k + h + 1) is calculated. The correlation with is the largest.

  On the other hand, the partial correlation between the information embedding code sequence pn (k) and the shifted version of the information embedding code sequence pn (k) is the largest value for pn (k + h + b + 1). Therefore, the phase difference is (h + b + 1) − (h + 1) = b, and it can be seen that the correct embedded information b can still be restored even if there is a clipping error.

  In this manner, the embedded information (additional information) restored by the embedded information restoring unit 24 is output from the information output unit 60 (step S25).

  As described above, in the embedded information restoration processing unit 450 or the embedded information restoration processing method according to the present embodiment, the document format is determined from the character string on the document in which the information is embedded by the information embedding processing unit 400 or the information embedding processing method. And a partial correlation between the parameter value series and the reference code series, the information embedding code series, and the shifted version of the information embedding code series.

  Next, the phase at which the maximum partial correlation is obtained is obtained for each of the reference code sequence and the information embedding code sequence, and the phase difference is set as an embedded information restoration value at the partial correlation position. Then, since such embedded information restoration values are obtained for a plurality of partial correlation positions, and the embedded information restoration value that appears with the highest frequency is finally restored as the embedded information, the following operation is performed. An effect can be obtained.

  That is, in the input document image or electronic document data, additional information is embedded as a phase difference between the reference code sequence and the information embedding code sequence, so that a clipping error occurs when performing character clipping processing. Even in such a case, it is possible to restore the embedded information. FIG. 12 shows an example of character cutout errors. In FIG. 12, (A) shows an example in which one character is divided, and (B) shows an example in which a plurality of characters are connected.

  As described above, even when a character cut-out error, specifically, a character split (A) or a combination of multiple characters (B) occurs during character cut-out processing necessary for restoration of embedded information, the character is cut out. Since the phase difference between the reference code sequence and the information embedding code sequence is unchanged, the embedded information can be reliably restored based on the phase difference.

(Modification 1)
In the above embodiment, a single information embedding code sequence pn (k) is used. However, a plurality of information embedding code sequences pn (m, k), (m = 0, 1,..., M -1) can be used to multiplex and embed M pieces of information {b0, b1,..., BM-1}. In this manner, by multiplexing and embedding M pieces of information using a plurality of information embedding code sequences, the amount of information that can be embedded in a certain character string can be increased.

  M pieces of information {b0, b1,..., BM-1} are embedded as a phase difference between the reference code sequence ref (k) and the information embedding code sequence pn (m, k), as in the above embodiment. That is, the information bm is expressed by an information embedding code sequence pn (m, k + bm). Each of the information embedding code sequences pn (m, k), (m = 0, 1,..., M−1) has a sharp autocorrelation characteristic as shown in FIG. A small cross-correlation with the information embedding code sequence is used.

(Modification 2)
In the above embodiment, by monitoring the partial correlation value at each window position (partial correlation position), it detects (estimates) the approximate position with the clipping error and the type (concatenation / division) of the clipping error. For example, it is possible to issue a warning and prompt the user to correct the cutout result.

  In step 1 in the above-described embedded information restoration procedure, the character spacing sequence a ′ (k), the reference code sequence ref (k), and the shifted version of the reference code sequence ref (k) at window position 0 in FIG. It was shown that when the correlation was calculated, the correlation with the reference code sequence ref (k) was the largest. Further, in step 4, it is shown that the correlation between the window position 1 and the reference code sequence ref (k + 1) is maximized.

  As described above, the partial correlation value between the character interval sequence a ′ (k), the reference code sequence ref (k), and the shifted version of the reference code sequence ref (k) is the same as the window position w if there is no clipping error. Thus, the correlation with the reference code sequence ref (k + w) is maximized.

  However, as shown in FIG. 11, when there is a clipping error (concatenation), the correlation with the reference code sequence ref (k + w + 1) with respect to the window position w is maximized after the clipping error position. Similarly, when there is a cutout error (division), the correlation with the reference code sequence ref (k + w-1) is maximized with respect to the window position w after the cutout error position.

  Therefore, by monitoring the phase of the reference code sequence ref (k) that maximizes the partial correlation at each window position, it is possible to detect the approximate position where the cut-out error occurs and the type of error (concatenation / division). . FIG. 13 shows a partial correlation between the character interval series a ′ (k) and the reference code series ref (k + w) and ref (k + w ± 1) at each window position.

  In FIG. 13, (A) represents a case where there is a clipping error (connection) at the position h. Until the clipping error position h is reached, the partial correlation with the reference code sequence ref (k + w) is large, and the partial correlation with the shifted version is small. However, the partial correlation with the reference code sequence ref (k + w) starts to decrease from the vicinity of the clipping error position h, and finally becomes almost zero.

  On the other hand, the partial correlation with the reference code sequence ref (k + w + 1) starts to increase in the vicinity of the extraction error position h, and finally the reference code sequence ref (k + w + 1) has the maximum correlation. When such an increase / decrease pattern of partial correlation is observed, it can be seen that there is a cutout error (connection) in the vicinity of the cutout error position h, so that the user can be prompted to correct the cutout result.

  In FIG. 13, (B) represents a case where there is a clipping error (division) at the position h. Until the clipping error position h is reached, the partial correlation with the reference code sequence ref (k + w) is large, and the partial correlation with the shifted version is small. However, the partial correlation with the reference code sequence ref (k + w) starts to decrease from the vicinity of the clipping error position h, and finally becomes almost zero.

  On the other hand, the partial correlation with the reference code sequence ref (k + w-1) starts to increase in the vicinity of the extraction error position h, and finally the reference code sequence ref (k + w-1) has the maximum correlation. When such a partial correlation increase / decrease pattern is observed, it can be seen that there is a clipping error (division) in the vicinity of the clipping error position h, so that the user can be prompted to correct the clipping result.

  In this way, by monitoring the partial correlation between the character interval sequence, the reference code sequence, and the shifted version of the reference code sequence, the approximate position where the clipping error occurs and the type of error (concatenation / division) are detected (estimated). The user can be prompted to correct the cutout result. As a result, the user can reliably correct, for example, a clipping error that is often overlooked visually on the display. Therefore, by executing, for example, a restoration process again on the corrected document by the user, the embedding information can be reduced. There is an advantage that it can be restored reliably.

  In the above-described embodiment and its modifications 1 and 2, the case where the present invention is applied to a document processing apparatus including both the information embedding processing unit 400 and the embedded information restoration processing unit 450 has been described as an example. The present invention is not limited to the application example, and can be similarly applied to a document processing apparatus including any one of the information embedding processing unit 400 and the embedded information restoring processing unit 450.

It is a block diagram which shows the outline of a structure of the document processing apparatus which concerns on one Embodiment of this invention. It is a block diagram which shows an example of a function structure of an information embedding process part. It is a flowchart which shows an example of the procedure of the information embedding process in an information embedding process part. It is explanatory drawing of encoding of embedded information. It is explanatory drawing about the autocorrelation characteristic of a reference code sequence and a code sequence for information embedding. It is explanatory drawing of the process which encodes the embedding information b using the phase difference of two code series ref (k) and pn (k). It is explanatory drawing of the process which corrects the appearance of document data. It is a block diagram which shows an example of a function structure of an embedded information restoration process part. It is a flowchart which shows the process sequence of the embedded information restoration process in an embedded information restoration process part. It is a figure which shows the case where h-th character cutout fails and it cuts out by connecting with the character before one. It is a figure which shows the mode of the information restoration | restoration in case the hth character cutout fails and it cuts out by connecting with the 1st previous character. It is a figure which shows the example of the character cut-out error, A) has shown the example in which one character is divided | segmented, (B) has each shown the example in which several characters are connected. It is a figure which shows the partial correlation of the character space | interval series a '(k) in each window position, the reference code series ref (k + w), and ref (k + w +/- 1).

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Document processing apparatus, 10 ... 1st document input part, 20 ... 2nd document input part, 30 ... Embedded information input part, 40 ... Processing control part, 41 ... CPU, 42 ... I / O circuit, 43 ... ROM, 44 ... RAM, 45 ... HDD (hard disk drive), 50 ... document output unit, 60 ... information output unit, 400 ... information embedding processing unit, 401 ... parameter value series extraction unit, 402 ... code sequence group generation unit, 403 ... Information encoding unit, 404... Modified parameter value sequence generation unit, 405... Document correction unit, 450.

Claims (10)

A document processing apparatus capable of embedding additional information in document data consisting of character strings,
First parameter series extraction means for extracting a parameter value series relating to the document appearance in the character string from the input document data;
First code sequence group generation means for generating a reference code sequence and an information embedding code sequence;
Information encoding for generating the input additional information as an encoded information sequence represented by a phase difference between the reference code sequence generated by the first code sequence group generation means and the information embedding code sequence Means,
Modified parameter value sequence generating means for synthesizing the encoded information sequence generated by the information encoding means with the parameter value sequence extracted by the first parameter series extracting means to generate a corrected parameter value sequence When,
A document processing apparatus comprising: a document correction unit that corrects the document data by using the correction parameter value series generated by the correction parameter value series generation unit. The information encoding means phase-shifts the information embedding code sequence by a phase difference corresponding to the additional information with respect to the reference code sequence, and the information embedding code sequence after the phase shift and the reference code sequence The document processing apparatus according to claim 1, wherein the encoded information sequence is added. The first code sequence group generation means generates a plurality of the information embedding code sequences,
The information encoding unit generates a plurality of encoded information sequences represented by a phase difference between the reference code sequence generated by the code sequence group generating unit and each of the plurality of information embedding code sequences. The document processing apparatus according to claim 1 or 2, wherein A document processing apparatus capable of restoring the additional information from the document data in which the additional information is embedded by the document processing apparatus according to claim 1,
The same reference code sequence as the reference code sequence generated by the first code sequence group generation means, the shifted version of the same reference code sequence, the same information embedding code sequence as the information embedding code sequence, and the same Second code sequence group generation means for generating a shifted version of the information embedding code sequence;
Second parameter series extraction means for extracting the parameter value series from the document data in which the additional information is embedded;
The parameter value sequence extracted by the second parameter sequence extraction means, the same reference code sequence generated by the second code sequence group generation means, and a shift version of the same reference code sequence and the same An information restoring means for restoring the additional information from the document data in which the additional information is embedded based on a partial correlation between the information embedding code sequence and a shifted version of the same information embedding code sequence. Feature document processing device. The information restoring means includes the parameter value sequence, the same reference code sequence and a shifted version of the same reference code sequence, the same information embedding code sequence and a shifted version of the same information embedding code sequence, And obtaining a phase difference that obtains the maximum partial correlation value for each of the reference code sequence and the information embedding code sequence, and using the phase difference as an information restoration value at the partial correlation value position. 5. The document processing apparatus according to claim 4, wherein a restoration value is obtained for a plurality of partial correlation positions, and an information restoration value that appears most frequently is used as a final restoration value. The information restoring means is configured to detect a character segmentation error from a transition pattern of partial correlation values between the parameter value series at the plurality of partial correlation positions and the same reference code series and a shifted version of the same reference code series. 6. The document processing apparatus according to claim 5, wherein the rough position and the type of clipping error are detected. The document processing apparatus according to claim 6, wherein the information restoration unit prompts the user to correct a character cutout result when detecting the cutout error. A document processing method capable of embedding additional information in document data consisting of character strings,
Extracting a parameter value series related to the document appearance in the character string from the input document data,
Generating a reference code sequence and an information embedding code sequence;
Generating the input additional information as an encoded information sequence represented by a phase difference between the reference code sequence and the information embedding code sequence;
A modified parameter value sequence is generated by combining the encoded information sequence with the parameter value sequence,
The document processing method, wherein the document data is corrected by the correction parameter value series. A document processing method capable of restoring the additional information from the document data in which the additional information is embedded by the document processing method according to claim 8,
Generates the same reference code sequence as the reference code sequence, the shifted version of the same reference code sequence, the same information embedding code sequence as the information embedding code sequence, and the shifted version of the same information embedding code sequence And
The parameter value sequence extracted from the document data in which the additional information is embedded, the same reference code sequence, a shifted version of the same reference code sequence, the same information embedding code sequence, and the same information A document processing method comprising: restoring the additional information from the document data in which the additional information is embedded based on a partial correlation with a shifted version of an embedded code sequence. A computer connected to a document input device for inputting document data consisting of character strings.
First parameter series extraction means for extracting a parameter value series relating to a document format in the character string from the document data input from the document input device;
First code sequence group generation means for generating a reference code sequence and an information embedding code sequence;
Information encoding for generating the input additional information as an encoded information sequence represented by a phase difference between the reference code sequence generated by the first code sequence group generation means and the information embedding code sequence Means,
Modified parameter value sequence generating means for synthesizing the encoded information sequence generated by the information encoding means with the parameter value sequence extracted by the first parameter series extracting means to generate a corrected parameter value sequence When,
A document processing program that functions as document correction means for correcting the document data by the correction parameter value series generated by the correction parameter value series generation means.

Images