JP2002232679A - Method and device for image processing, computer program, and storage medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a natural output by using the distance of blank parts between characters to embed information and to embed more information in comparison with use of the distance between words. SOLUTION: Circumscribed rectangles 104 of characters or elements (left-hand radicals, right-hand radicals, or the like) constituting the characters in an inputted document picture 101 are obtained (processing 102). A ratio of (P-S) to (P+S) where P is the distance between a target rectangle of the information embedding object and the preceding rectangle and D is the distance between the target rectangle and the following rectangle is quantized, and the extent of shift by which the position of the target rectangle should be shifted right or left is so determined that the quantization result may be equal to bits of objective watermark information 106. Then the image of the target rectangle is shifted in accordance with this extent of shift. This operation is repeated to generate an output image 105.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus and method for embedding digital watermark information in digital data of a document image or extracting digital watermark information therefrom, and a storage medium storing this method. is there.

[0002]

2. Description of the Related Art In recent years, digital watermark technology has attracted attention as a means for protecting the copyright of digital data such as images and sounds on the Internet. The digital watermark technology is a technology for embedding digital watermark information by manipulating digital data such as images and sounds to such an extent that humans cannot perceive it.

[0003] As a digital watermarking technique for a multi-valued image, there are generally known various methods utilizing the redundancy of pixel density. As a digital watermarking technique for a document image which is a binary image, for example, Japanese Patent Application Laid-Open No. 9-186603 (US Pat. No. 5,861,161) embeds digital watermark information by changing a blank length between words in English (European).
9) is known.

[0004]

However, in the above-mentioned prior art, a simple method of adjusting a distance between words by using a word composed of a plurality of characters as a processing unit is as follows.
It is difficult to get natural output results.

Therefore, the present invention embeds information by utilizing the distance of a blank portion between characters, enables a natural output to be obtained, and embeds more information than that between words. It is an object of the present invention to provide an image processing apparatus and method, a program, and a storage medium that make it possible.

Another object of the present invention is to provide an image processing apparatus and method, a program, and a storage medium that provide a more natural output result even when white space between words is used.

[0007]

In order to solve such a problem, for example, an image processing apparatus according to the present invention has the following arrangement. That is.

An image processing apparatus for embedding digital watermark information in a document image, comprising: extracting means for extracting a rectangle circumscribing a character image or an element image constituting a character in the document image; and information to be embedded. Based on the logical value of the bit, the shift amount calculating means for determining the shift amount for shifting the position of the target rectangle to any of the adjacent rectangles, and the shift amount calculated by the shift amount calculating means. Shifting means for shifting the character image or element image of the target rectangle.

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

<First Embodiment> In the present embodiment, a document image analysis technique consisting of area division and extraction of a character element circumscribed rectangle is used for Japanese document data (document data using kanji), A digital watermark in which information is embedded using a blank length between element circumscribed rectangles will be described. More specifically, for example, Japanese Patent Application Laid-Open No. 9-186603 (US Pat.
In the present embodiment, in consideration of the problem that an unnatural document image is created, for example, when the method applied between words in Japanese Patent No. 861619) is adopted as it is, the kanji characters “Hen” and “Tsukuri” are separated. In U.S. Pat. No. 5, it is possible to use the blank length between characters or elements (characters, etc.) constituting the character, not between words, and to minimize the unnaturalness, and to reliably extract watermark information digitally. Propose embedding.

The present embodiment is roughly divided into a digital watermark embedding method and a digital watermark extraction method. FIG. 1 shows an outline of a digital watermark embedding method, and FIG. 2 shows an outline of a digital watermark extraction method. Hereinafter, description will be made in order.

<Embedding Method> First, a digital watermark embedding method will be described. In FIG. 1, an input document image 101
(For example, read by an image scanner or the like) is first divided into a text area and a graphic area such as a graph by a document image analysis unit 102 described later, and a circumscribed rectangle is extracted for each character element in the text area. You. The result is indicated by reference numeral 104. Here, the character element is
A rectangular area extracted using the projection technology known as character recognition technology (a technology that creates a histogram of dots in the horizontal and vertical directions to determine the space between lines and between character elements). Circumscribed rectangle), which is a single character and the components of the character
It may be. In addition, area division (separation) into text area (character string area) and non-text area in document image
Uses a known technique. Also, reference numeral 104 shown in the figure.
Is shown in an easy-to-understand manner. Actually, for example, the origin is set at the upper left corner of the document image, and the coordinates of the upper left corner and lower right corner of each rectangle from the position, or the coordinates of the upper left corner of the rectangle and the height of the rectangle And data such as width and the like are extracted and stored in a memory in the order in which the characters are arranged, and a rectangle is not drawn as shown in the figure.

Further, a blank length between circumscribed rectangles is calculated from information (coordinate data) of the circumscribed rectangle of each character (or character element) in the extracted text area.

Although details will be described later, the number of rectangles in a certain row is three.
If there are 0 characters (not necessarily 30 characters),
Excluding the rectangles at both ends, the second rectangle, the fourth rectangle, ...
And the even-numbered rectangle. Now, a blank length P between a certain rectangle of interest and a rectangle located in front of it (hereinafter, a previous rectangle),
The blank length S between the target rectangle and the succeeding rectangle (the latter rectangle) is calculated, and the target rectangle is shifted left and right by a small distance to P,
By controlling the relationship of S, one bit (0 or 1) of information is embedded in the target rectangle. If one row is 30 rectangles, 30 / 2-1 = 14 bits will be embedded.
Here, an example is shown in which information is embedded in even-numbered rectangles. However, information may be embedded in rectangles in which neighboring rectangles exist, and as long as this condition is satisfied, information is embedded in odd-numbered rectangles. It may be embedded. In addition, every other rectangle to be embedded (one rectangle between
The reason for this is that if information is embedded sequentially in continuous rectangles, one blank distance will be shared by the preceding and following rectangles, and the distance that can be controlled will always be the distance to the rear rectangle. Errors gradually accumulate,
This is because the layout changes significantly.

As described above, by shifting the area within a certain circumscribed rectangle to the left and right, the document image 105 in which the digital watermark information 106 is embedded is subjected to the watermark embedding process (10
It will be generated in 3).

<Document Image Analysis Technology> The document image analysis technology is originally one of the elemental technologies of the character recognition technology, and divides an input document image into text regions and graphic regions such as graphs. For the text area, the projection is cut out in character units using projection. As an example, JP-A-6-68301 can be mentioned.

<Extraction Technique> Next, a digital watermark extraction technique will be described. In the digital watermark extraction method, similar to the digital watermark embedding method, first, a document image analysis unit 202 used in the character recognition technology converts a watermark embedded image 201 (read by an image scanner or the like) into a character by area division and projection. The circumscribed rectangle 203 is extracted. Next, a blank length between adjacent circumscribed rectangles is calculated using the extracted information on the circumscribed rectangle group.

Attention is paid to the even-numbered rectangles, and one bit of information embedded in the rectangle of interest is extracted from the relationship between the distance P between the preceding rectangles and the distance S between the following rectangles. This is similarly performed for the other rectangles (even-numbered rectangles) in the row, and the embedded information is extracted.

<Embedding Rule> Next, the embedding rule will be described. First, the blank lengths before and after the rectangle in which 1-bit information is embedded are determined as P and S as described above (see FIG. 3).
The rectangles for embedding this one bit are determined every other except for the rectangles at both ends in one row as described above. Next, for the blank lengths P and S determined here, (P
−S) / (P + S) (however, since this value is 0 or a small value, the denominator is multiplied by a small value or the numerator is multiplied by a large value), and an appropriate quantization step is used. Quantize. The quantization representative value is alternately set to 0.
Or 1 is assigned.

At the time of extracting the watermark information, for example, the embedded value is extracted according to Equation 1 in FIG. 4 (“mod” is a function that returns the remainder when divided by 2, and is binarized). That means you can). Here, α indicates a quantization step.

At the time of embedding the watermark information, while shaking the area within the circumscribed rectangle (of characters or character components) for embedding one bit left and right by one pixel,
Calculate Equation 1 in Figure 4 and the result is the value to be embedded
The direction (left or right) and width (number of pixels) of the parallel movement are searched until (0 or 1) is reached.

FIGS. 5 and 6 show flowcharts of this search. The details will be described below. In order to simplify the description, an example in which the document image is horizontally written will be described.

First, the meaning of a variable is as follows.

The variable i indicates a translation width candidate. Variable Fl
ag1 indicates whether or not the rectangle to be embedded for embedding information is in contact with the right rectangle when it is moved to the right by the distance i. Further, a variable Flag2 indicates whether or not a contact is made when the rectangle is moved to the left by the distance i. Like Flag1,
Take 1 if touching.

Next, each processing of this flowchart will be described along the processing flow. This flow is step S
It starts from 501.

When this processing is started, first, in step S
At 502, an initial value of a variable group is set. That is, i =
It is assumed that flag1 = flag2 = 0.

Next, in step S503, when a target rectangle (partial image of a target character or character element, hereinafter abbreviated) to be moved for embedding is translated rightward by a distance i, a character or character element adjacent to the right is moved. It is determined whether or not a rectangle (front rectangle) is touched. If they do, the flow advances to step S504 to set the variable flag1 to "1".

In step S505, when the target rectangle is translated leftward by a distance i, it is determined whether or not the rectangle touches the rectangle on the left. If contact is made, step S50
In step 6, "1" is set in flag2.

In step S507, the parallel movement distance i
In the case of the value indicated by, it is determined whether or not both sides are touched. If both sides are touched, the parallel movement width is set to 0 in step S508, and the calculation of the parallel movement width is terminated. In this case, embedding is impossible.

On the other hand, if it is determined that there is no contact on both the left and right sides, the process proceeds to step S601 in FIG. This step S
In 601, it is determined whether or not a bit to be embedded is obtained by Equation 1 in FIG. 4 when the target rectangle is translated rightward by a distance i. If it is obtained, the calculation of the translation width is ended in step S602 with the translation width being i. Here, the translation width means that when the value is positive, it translates to the right, and when the value is negative,
It is defined as meaning to translate to the left.

In step S603, the target rectangle is set at a distance i.
It is determined whether or not a bit to be embedded is obtained by Equation 1 in FIG. If it is obtained, the parallel movement width is set to -i in step S604, and the calculation of the parallel movement width ends.

In step S605, the value of i is incremented, and the process returns to step S503.

After the search result is determined as the direction and distance for the parallel movement, the area within the circumscribed rectangle of the character for actually embedding one bit is translated. By performing the above processing on the entire document image, the digital watermark information can be embedded in the document image.

The shift amount for all characters or all character elements to be embedded is determined (the determined shift amount of each rectangle is temporarily stored in a memory in association with information specifying the rectangle). The image in the corresponding rectangle in the document image is shifted according to the determined shift amount, and the embedded document image is generated.

The embedding and extraction methods described above are:
This can be realized by using the signal processing device shown in FIG.

In the figure, a host computer 701
Can be realized by, for example, a commonly used personal computer. The scanner 714 is for reading an original of a document in which watermark information is to be embedded. The computer 701 can edit and store the read image. Furthermore, the image obtained here can be printed from the printer 715. Also,
Various manual instructions and the like from the user
2. Performed by input from the keyboard 713.

Inside the host computer 701, each block described later is connected by a bus 716, and various kinds of data can be exchanged.

In the figure, reference numeral 703 denotes a CPU capable of controlling the operation of each internal block or executing a program stored therein. A ROM 704 stores a specific image that is not permitted to be printed, or stores a necessary image processing program and the like in advance. A RAM 705 temporarily stores a program and image data to be processed for processing by the CPU.
It is. 706 is a RAM for the OS and applications
And a hard disk (HD) capable of storing programs and image data to be transferred to the PC or the like in advance and storing processed image data. A scanner interface (I / I / 707) 707 is connected to a scanner that reads a document or a film by a CCD and generates image data, and can input image data obtained by the scanner.
F). 708 is a CD which is one of the external storage media
This is a CD drive that can read or write data stored in (CD-R). 709 is 70
8 is an FD drive capable of reading from the FD and writing to the FD. 710 is the same as 708
DV that can read from VD and write to DVD
D drive. If a program for image editing or a printer driver is stored in a CD, FD, DVD, or the like, these programs are installed on the HD 706 and transferred to the RAM 705 as necessary. . Reference numeral 711 denotes an interface (I / F) connected to a mouse 712 or a keyboard 713 to receive an input instruction.

In such a configuration, an original for which watermark information is to be embedded is read by the scanner 714 using an application (in some cases, the HD7 is used).
06 or a document image stored in a removable storage medium). Next, watermark information (for example, information such as the author name)
Is input from the keyboard 713 or the like, the information is converted into a bit string, and a program for executing the above-described processing is executed to embed information. The embedded and generated document image may be stored on a network, for example, so that a third party can view the document image, or may be printed by the printer 715.

In the above embodiment, the equation of FIG. 4 is used as the information of the bits to be embedded. Depending on whether the value obtained by subtracting S from P is greater than or equal to 0 or less than 0, one bit of “0” or “1” may be embedded, but in this case, the character size of the document image is not considered at all. That is, it is not possible to adaptively cope with a document having a small character or a document having a large character. In this regard, the configuration shown in FIG. 4 can be applied to the case where the distance between characters (rectangles) becomes relatively large when the character size is large, and a more natural output result can be obtained. In addition, α multiplied by the denominator in FIG. 4 takes a small value, but this value may be adaptively changed (for example, it changes according to the average distance between rectangles).

In the above-described embodiment, an example has been described in which the variable i indicating the shift amount is "1", that is, the shift is performed in units of one pixel in the read image. 1 in this case
Although the pixel depends on the reading resolution of the image scanner, a third party does not always read at the same resolution. Therefore, the minimum unit to be shifted may be determined depending on the reading resolution of the document. The simplest approach is based on the size of one pixel when reading at 200 dpi, since most existing image scanners will have a resolution of at least 200 dpi. Therefore, when reading is performed by a scanner having a resolution of 600 dpi, the variable i is increased in units of "3".

Further, in the above-described embodiment, the processing is performed by focusing on the characters or the elements constituting the characters. However, even if the formula in FIG. 4 is applied to a blank portion between words in a sentence in a language such as English. I do not care. Generally, English words are composed of several characters.Therefore, the number of blanks between words in one line is smaller than in the kanji area such as Japanese, and the amount of information to be embedded is smaller. There is an effect that it becomes possible to maintain in a natural state. The white space between words and the white space between characters may be distinguished by performing processing such as determining that a predetermined distance or more is between words. Of course, shifting will shift the entire word. The extraction of the embedded information is based on the case where the object in the first embodiment is changed from a character to a word, and the description thereof is omitted.

<Second Embodiment> The first embodiment has been described by taking an existing document image as an example, but the present invention is not limited to this. For example, when the OS running on the personal computer is Windows by Microsoft Corporation in the United States, the above function may be included in a printer driver installed in the OS.

Hereinafter, an operation process in the case of being realized by the printer driver will be described with reference to the flowchart of FIG.

When a print instruction is received from an application such as a word processor, the printer driver starts.

First, in step S81, a message for inquiring whether or not to embed a document to be printed is displayed, and the operator is made to make a selection. Note that this selection may be made by using a dialog box displayed during the printing process and displayed when the option is selected.

In any case, it is determined in step S82 whether embedding has been instructed. If not, step S82 is performed in order to perform the same operation as a normal printer driver.
Go to step 83 (for example, a process of converting to a PDL description language and outputting).

If it is determined in step S82 that embedding has been instructed, the process proceeds to step S84,
The operator inputs information to be embedded. The timing for inputting the embedding information does not need to be this timing. For example, the timing for inputting the embedding information may be input at the timing of embedding in step S81, or may be registered as a file in which copyright information is described in advance or registered in a registry. In advance, the contents may be used.

In step S85, the data to be printed received from the application via the OS is analyzed. If the data is related to characters, a character pattern for printing is generated, and its circumscribed rectangle (one character 2 for
And information embedding is performed on the target rectangle based on the distance from the preceding and following rectangles. Since it is necessary that a rectangle to be embedded has a succeeding rectangle, it is actually performed every time three rectangles are generated. In any case, analyzing the data passed from the application makes it possible to reliably determine whether it is related to characters or non-character information such as images and graphics, so image area separation is simple. It becomes possible.

In this manner, the embedded image of the minimum unit sufficient for printing out (for example, an image for one page)
Is generated, it is output to the OS in step S86, and the OS performs a process of outputting the data to the printer.

As described above, in the case of embedding watermark information in the second embodiment, document data including characters must be output to a printer in the form of image data. Can be embedded as part of normal application work. That is, it can be realized without special knowledge about embedding.

<Other Embodiments> As described above, the present invention provides:
It can be realized by a computer (CPU or MPU) of a general-purpose information processing device (such as a personal computer). Therefore, there is also a case where the computer is supplied with software program codes for realizing the above-described embodiment, and the computer of the system or apparatus operates the various devices according to the program code to realize the above-described embodiment. It is included in the category of the present invention. In this case, the program code of the software implements the functions of the above-described embodiment, and the program code itself and a storage medium for supplying the program code to a computer, specifically, the program Magnetic disk (floppy (registered trademark) disk or hard disk) storing codes, optical disk (CDROM, CD-R, etc.), memory card,
Tapes (paper tapes, magnetic tapes, etc.) and ROM elements can be used. In addition to the case where the computer controls the various devices in accordance with only the supplied program code to realize the functions of the above-described embodiment, the computer may execute the OS (Operating System) operating on the computer. Such a program code is also included in the scope of the present invention when the above-described embodiment is realized in cooperation with a (system) or other application software.

Further, after the supplied program code is stored in a memory provided in a function expansion board of the computer or a function expansion unit connected to the computer, the function expansion board or the function is stored based on the instruction of the program code. The case where the CPU or the like provided in the storage unit performs part or all of the actual processing, and the above-described embodiment is realized by the processing is also included in the scope of the present invention.

As described above, according to the present embodiment,
"Hen" or "Tsukuri" for each character or for each character
It is possible to embed digital watermark information suitable for a document image in which a blank length exists in each partial configuration such as. In addition, regardless of the blank length of each adjacent word, character, partial configuration, etc., the amount of change in this blank length for embedding digital watermark information should be kept as small as possible, and visual image quality degradation should be suppressed. Can be.

Regardless of the blank length of each adjacent word, character, partial configuration, etc., the amount of change in the blank length for embedding digital watermark information is kept as small as possible, and visual image quality degradation is suppressed. It becomes possible to suppress.

[0055]

As described above, according to the present invention, information can be embedded by utilizing the distance of a blank portion between characters, and a natural output can be obtained. , It is possible to embed more information.

According to another aspect of the present invention, a more natural output result can be obtained even when a space between words is used.

[Brief description of the drawings]

FIG. 1 is a functional block diagram for embedding a digital watermark in an embodiment.

FIG. 2 is a functional block diagram for extracting a digital watermark in the embodiment.

FIG. 3 is a diagram for describing an embedding process in the embodiment.

FIG. 4 is a diagram illustrating an operation expression for quantization in an embedding / extraction process according to the embodiment.

FIG. 5 is a flowchart of an embedding process in the embodiment.

FIG. 6 is a flowchart of an embedding process in the embodiment.

FIG. 7 is a block diagram when applied to the information processing apparatus in the embodiment.

FIG. 8 is a flowchart illustrating an operation processing procedure of a printer driver according to the second embodiment.

 ──────────────────────────────────────────────────の Continued on the front page F term (reference) 5B057 CE08 CE09 CE12 CH01 CH11 DC01 5C063 CA29 CA36 DA07 DA13 DA20 5C076 AA14 BA06 CA12 5J104 AA05 AA14

Claims (19)

[Claims] 1. An image processing apparatus for embedding digital watermark information in a document image, comprising: extracting means for extracting a rectangle circumscribing a character image or an element image constituting a character in the document image; A shift amount calculating unit that determines a shift amount for shifting the target rectangle to any of the adjacent rectangles based on a logical value of a bit of the information, and a shift amount calculated by the shift amount calculating unit. An image processing apparatus comprising: a shift unit that shifts a character image or an element image of a rectangle of interest based on the information. 2. The image processing apparatus according to claim 1, wherein the document image is input from an image scanner. 3. The shift amount calculating means calculates a ratio between P−S and P + S, where P is a distance from another rectangle located before the rectangle of interest and S is a distance from the rectangle located after the rectangle. 3. The method according to claim 1, wherein the determination of 0 or 1 is performed by quantization, and the shift amount of the position of the rectangle of interest is calculated so that the determination becomes the same as the value of the bit to be embedded. An image processing apparatus according to the item. 4. The image processing apparatus according to claim 3, wherein the quantization is performed by multiplying a ratio of P−S and P + S by a quantization step and performing binarization. 5. The rectangle according to claim 1, wherein a rectangle to be embedded with information is every other rectangle in a rectangle group in one line. Image processing device. 6. An image processing method for embedding digital watermark information in a document image, comprising: an extraction step of extracting a rectangle circumscribing a character image or an element image constituting a character in the document image; A shift amount calculating step of determining a shift amount for shifting to a neighboring rectangle based on a logical value of a bit of the information bit, and a shift amount calculated in the shift amount calculating step. A shifting step of shifting a character image or an element image of the rectangle of interest based on the image processing method. 7. A computer program for embedding digital watermark information in a document image, comprising: a program code for an extraction step of extracting a rectangle circumscribing a character image or an element image constituting a character in the document image; A program code for a shift amount calculating step of determining a shift amount for shifting the position of the target rectangle to any of the adjacent rectangles based on a logical value of a bit of information of interest; A computer program, comprising: a program code for a shift step of shifting a character image or an element image of a target rectangle based on the calculated shift amount. 8. The computer program according to claim 7, wherein the computer program is a printer driver program, and displays a selection user interface for selecting whether to embed digital watermark information when a print instruction is issued. When a program code and an instruction to embed are given, a character pattern is generated from information on characters in the data passed from the host process.
A computer program comprising a program code of a step of passing a generated character pattern or an element image pattern constituting the character pattern to the extraction step. 9. A storage medium for storing the computer program according to claim 7. Description: 10. An image processing apparatus for inputting a document image and extracting watermark information embedded in the document image, wherein the image processing device circumscribes a character image in the document image or an element image which is an element constituting the character. Circumscribed rectangle determining means for determining a rectangle; calculating means for calculating a distance to another rectangle adjacent to the target rectangle; information embedded in the target rectangle based on the distance calculated by the calculating means An image processing apparatus comprising: embedded information extracting means. 11. The embedding information extracting means sets a distance between a target rectangle and a rectangle located before the target rectangle to P, and sets a distance between the target rectangle and a rectangle positioned after the target rectangle to S.
11. The image processing apparatus according to claim 10, wherein a ratio of PS and P + S is multiplied by a quantization step, and a bit of the embedded information is extracted based on the binarized result. . 12. An image processing method for inputting a document image and extracting watermark information embedded in the document image, wherein the method circumscribes a character image in the document image or an element image which is an element constituting the character. A circumscribed rectangle discriminating step of extracting a rectangle; a calculating step of calculating a distance to another rectangle adjacent to the target rectangle; and information embedded in the target rectangle based on the distance calculated in the calculating step. An image processing method comprising an embedded information extracting step. 13. The embedding information extraction step includes the steps of: defining a distance between a target rectangle and a rectangle located before the target rectangle as P, and defining a distance between the target rectangle and a rectangle positioned after the target rectangle as S.
13. The image processing method according to claim 12, wherein a ratio of PS and P + S is multiplied by a quantization step, and a bit of the embedded information is extracted based on a result of the binarization. . 14. An image processing apparatus for embedding digital watermark information in a document image composed of a plurality of words, comprising: changing a distance between words adjacent to each word in the document image, so that the document image Digital watermark embedding means for embedding watermark information, wherein the digital watermark embedding means sets P-S when distances between a predetermined word and adjacent words before and after the predetermined word are P and S, respectively.
An image processing apparatus characterized in that digital watermark information is represented based on a quantization value of a ratio of P + S and P + S. 15. An image processing method for embedding digital watermark information in a document image composed of a plurality of words, the method comprising: changing a distance between words adjacent to each word in the document image so that the document image is electronically embedded. And a digital watermark embedding step of embedding watermark information. In the digital watermark embedding step, when each distance between a predetermined word and adjacent words before and after the predetermined word is P and S, the ratio of P−S to P + S An image processing method characterized in that electronic watermark information is represented based on a quantization value. 16. A storage medium storing an image processing program for embedding digital watermark information in a document image composed of a plurality of words in a computer-readable state,
The program includes a digital watermark embedding step of embedding digital watermark information in the document image by changing a distance between words adjacent to each word in the document image. A storage medium characterized in that when a distance between a word and adjacent words before and after the word is P and S, digital watermark information is represented based on a quantized value of a ratio between PS and P + S. . 17. An image processing apparatus for extracting digital watermark information from a document image including a plurality of words, wherein the digital watermark is added to the document image based on a distance between words adjacent to each word in the document image. The digital watermark information detected by the digital watermark detecting means is P-S, where P and S are distances between a predetermined word and adjacent words before and after the predetermined word. An image processing apparatus characterized by being represented based on a quantization value of a ratio of S and P + S. 18. An image processing method for extracting digital watermark information from a document image including a plurality of words, wherein the digital watermark is added to the document image based on a distance between words adjacent to each word in the document image. A digital watermark detecting step of detecting information; wherein the digital watermark information detected in the digital watermark detecting step is P-S, where P and S are distances between a predetermined word and adjacent words before and after the predetermined word. An image processing method characterized by being represented based on a quantization value of a ratio of S and P + S. 19. A storage medium storing an image processing program for extracting digital watermark information from a document image composed of a plurality of words in a computer-readable state, the program comprising: A digital watermark detection step of detecting digital watermark information in the document image based on a distance between adjacent words, wherein the digital watermark information detected in the digital watermark detection step is adjacent to a predetermined word before and after the predetermined word. A storage medium characterized in that, when each distance to a word to be performed is represented by P and S, the distance is represented based on a quantization value of a ratio between P−S and P + S.