JP4910145B2 - Information embedding method and information reading method in pattern image, information embedding device and information reading device in pattern image - Google Patents

Description

  The present invention relates to a method and apparatus for embedding information in a pattern image, and a method and apparatus for reading embedded information data from pixel value data obtained by photographing the image with an imaging apparatus.

  A conventional technique for embedding information in an image (for example, see Non-Patent Document 1) is called “digital watermark” or “steganography” depending on the application. In these techniques, the basic approach is to embed information by converting the pixel value data of an image in a color or frequency domain.

  On the other hand, as a technique for embedding information for a printed image and having a resistance to analog noise and deformation like a two-dimensional barcode, a coding method using color information (Patent Document 1), for each image block There are proposed a method using the gradation difference (Non-Patent Document 2), a method of introducing a redundant code (Patent Document 2), and the like.

  When applying methods such as the existing digital watermark used for digital data to data that has undergone an analog process where noise and deformation are unavoidable, ensuring the tolerance of noise and deformation Requires more redundancy than required for tampering. Therefore, there is a problem that only a few bytes of information can be embedded in the conventional method in which the method for digital data is simply expanded.

Further, since it is difficult to add a design element to the two-dimensional barcode, there is a drawback that the appearance of the printed matter is impaired when used in large quantities.
JP 2001-195536 A JP 2000-287073 A “Digital Watermarking Technology, Digital Content Security” edited by the Institute of Image Electronics Engineers, Tokyo Denki University Press, 2004 Fujitsu Journal June 2005, “New Technology (1)”, http://jp.fujitsu.com/about/journal/281/newtechnology/

  The problem to be solved is that it is difficult to add design elements to images that have been secured with robustness against noise contamination and image deformation during printing and shooting, etc. It is a problem of the prior art that it is damaged.

Therefore, in the present invention, by using a group of images generated by deforming a design that can be intentionally designed by humans, more information is more robust than before with respect to noise and distortion mixed during printing and shooting. A method and apparatus are provided that allow embedding and reading.

  The present invention realizes a method for embedding and reading information that is not perceived by human vision in a group of images to be printed. The present invention also provides two problem solving means.

  As a first problem solving means, a method of embedding information in an image when there is a detection frame at the time of shooting will be described.

  A sample image is deformed to generate a plurality of modified symbols.

  N mask regions are extracted from each modified pattern.

  M-bit data is embedded in each extracted mask area. However, m-bit data corresponds to the proportion of colors distributed in each mask area. Thereby, n × m-bit data is embedded in each modified symbol.

Using the embedded n × m bit data, each modified symbol is classified into 2 ^nm classes.

Next, the information data to be embedded is divided for each n × m bits, and the corresponding modified symbol is selected from 2 ^nm classes.

  By regularly arranging the selected deformed symbols, a pattern image in which information data is embedded is generated.

  A frame for detecting a pattern image in which information data is embedded is installed, and an information embedded pattern image (hereinafter referred to as a framed pattern image) having a detection frame is generated.

  An apparatus for realizing an information embedding method according to the first problem solving means is shown in FIG. The information embedding device 11 includes a mask area extraction unit 13, a color ratio calculation unit 14 in a mask area, a class type determination unit 15, a character string division and deformation symbol selection unit 16, a modified symbol arrangement connection unit 17, and a detection It is composed of a frame installation unit 18.

  In the mask region extraction unit 13, n mask regions are extracted from each of the plurality of modified symbols 12 of FIG. 4.

  The color ratio calculation unit 14 in the mask area embeds m-bit data in each area from the ratio of colors distributed in each mask area. Thereby, n × m-bit data is embedded in each modified symbol.

The class type determination unit 15 classifies each modified symbol into 2 ^nm classes using the n × m-bit data embedded by the ratio calculation unit 14.

The character string division and modified symbol selection unit 16 divides the information data to be embedded into n × m bits and selects a corresponding modified symbol from 2 ^nm classes.

  The modified symbol arrangement connecting unit 17 generates a pattern image in which information data is embedded by regularly arranging the selected modified symbols.

  The detection frame setting unit 18 sets a frame for detecting a pattern image area in which information data is embedded at the time of shooting.

  By using this information embedding device 11, the framed pattern image 19 of FIG. 4 is generated.

  A means for reading information from a framed pattern image in which information data is embedded by the first problem solving means of the present invention will be described.

  Based on the geometric information of the frame detected from the target image, a modified design in which n × m-bit data is embedded is cut out.

  For each deformed pattern, the same n mask areas as in the embedding are extracted.

  The m-bit data embedded in each extracted mask area is decoded. Thereby, n × m-bit data is decoded from each modified symbol.

  Next, the decoded n × m-bit data is concatenated according to the arrangement order of the modified symbols, and the embedded information data is read out.

  FIG. 5 shows an apparatus for detecting embedded information from a framed pattern image in which information data is embedded by the first problem solving means of the present invention. The embedded information detection device 21 includes a frame recognition unit 23, an image calibration unit 24, an image division unit 25, a mask region extraction unit 13, a color ratio calculation unit 14 in the mask region, a data decoding unit 26, and a data connection unit 27. Consists of

  The frame recognition unit 23 and the image calibration unit 24 detect the frame from the detection target image 22 and calibrate the image based on the geometric information of the frame.

  The image dividing unit 25 cuts out a deformed design in which n × m-bit data is embedded from the image calibrated by the image calibrating unit 24.

  The mask area extraction unit 13 and the color ratio calculation unit 14 in the mask area extract the mask area from the deformed pattern and calculate the ratio of the colors distributed in each mask area.

  The data decoding unit 26 decodes n × m-bit data from the color ratio calculated by the color ratio calculation unit 14 in the mask area. Thereby, n × m-bit data is decoded from each modified symbol.

  The data concatenation unit 27 concatenates the n × m-bit data decoded by the data decoding unit 26 in the order in which the modified symbols are arranged.

  By using this embedded information detecting device 21, the embedded information data 28 of FIG. 5 is obtained.

  Next, as a second problem solving means, a method of embedding information in a pattern image when there is no detection frame at the time of shooting will be described.

  First, a sample image is deformed to generate a plurality of deformed symbols.

  In order to detect embedded information data, a specific position detection condition is determined.

  A symbol that matches the position detection condition is selected from each modified symbol and used as a position detection symbol.

  N mask regions are extracted from each modified pattern.

  M-bit data is embedded in each extracted mask area. However, m-bit data corresponds to the proportion of colors distributed in each mask area. Thereby, n × m-bit data is embedded in each modified symbol.

Using the embedded n × m bit data, each modified symbol is classified into 2 ^nm classes.

Next, the information data to be embedded is divided into n × m bits, and a corresponding modified pattern is selected from 2 ^nm classes.

  After the position detection symbols are arranged at the head, the selected modified symbols are regularly arranged to generate a pattern image in which information data is embedded.

  The pattern image in which the information data is embedded is further regularly arranged to generate an information embedded pattern image without a detection frame (hereinafter, frameless pattern image).

  FIG. 6 shows an apparatus for realizing an information embedding method according to means for solving the second problem. The information embedding device 31 includes a position detection symbol determination unit 33, a mask region extraction unit 13, a color ratio calculation unit 14 in the mask region, a class type determination unit 15, a character string division and modification symbol selection unit 16, and a modification. It is comprised from the symbol arrangement | positioning connection part 34 and the cyclic | annular arrangement | positioning connection part 35. FIG.

  The position detection symbol determination unit 33 selects a symbol that matches the position detection condition determined in the position detection condition determination 32 of FIG. 6 from the plurality of modified symbols 12 of FIG.

  In the mask region extraction unit 13, the color ratio calculation unit 14 in the mask region, the class type determination unit 15, and the character string division / deformation symbol selection unit 16, parts having the same reference numerals in the information embedding device 11 of FIG. The same processing as

  In the deformed symbol arrangement connecting unit 34, after the position detecting symbol selected by the position detecting symbol determining unit 33 is arranged at the head, the character string division and the deformed symbol selected by the deforming symbol selecting unit 16 are regularly arranged, A pattern image in which information data is embedded is generated.

  In the annular arrangement connecting unit 35, the pattern image in which the information data generated by the modified symbol arrangement connecting unit 34 is embedded is arranged more regularly.

  By using this information embedding device 31, the frameless pattern image 36 of FIG. 6 is generated.

  A procedure for reading out information from a frameless pattern image in which information data is embedded by the second problem solving means of the present invention will be described.

  First, a deformed pattern (position detection pattern) matching the position detection condition is detected from the target image to cut out an image area in which information data is embedded.

  A modified design in which n × m-bit data is embedded is cut out from the image area in which the information data is embedded.

  The same n number of mask areas as those at the time of embedding are extracted from each modified pattern.

  The m-bit data embedded in each extracted mask area is decoded. Thereby, n × m-bit data is decoded from each modified symbol.

  Next, the decoded n × m-bit data is concatenated according to the arrangement order of the modified symbols, and the embedded information data is read out.

  An apparatus for realizing detection of embedded information data from a frameless pattern image in which information data is embedded by the second problem solving means of the present invention is shown in FIG. The embedded information detection device 41 includes an image region detection unit 43, an information embedding region detection unit 44, an image calibration unit 24, an image division unit 25, a mask region extraction unit 13, a color ratio calculation unit 14 in the mask region, and data decoding. A unit 26 and a data connection unit 27 are included.

  The image region detection unit 43 detects an image region from the detection target image 42 in FIG.

  The information embedding area detection unit 44 recognizes an area where information data is embedded by detecting a deformed pattern (position detection pattern) that matches the position detection condition.

  In the image calibrating unit 24, the image dividing unit 25, the mask region extracting unit 13, the color ratio calculating unit 14, the data decoding unit 26, and the data linking unit 27 in FIG. The same processing as that of the part with

By using the embedded information detecting device 41 in FIG. 7, the embedded information data 28 in FIG. 7 is obtained.

In the first problem solving means of the present invention, since information is embedded in a specific mask area in the deformed pattern, it is possible to generate a deformed pattern holding the information only by manipulating a part of the image. As a result, an information-embedded image can be generated using a deformation rule that does not impair the design appearance. Further, the second problem solving means of the present invention makes it possible to embed information and detect information without impairing the appearance of the printed matter even if a pattern image in which a large amount of modified symbols in which information is embedded is arranged.

  As the best mode for carrying out the present invention, a first problem solving means and a second problem solving means for a black and white basic design are described below.

  The information embedding method of the first problem solving means will be described with reference to FIG. FIG. 4 shows an apparatus for realizing the information embedding method of the first problem solving means, and FIG. 5 shows an embedded information detection apparatus.

  A method for embedding information in a pattern image is described in a. 1, a. 2, a. 3, a. 4, a. 5, a. 6, a. It consists of 7 procedures. Also, the information embedding device 11 shown in FIG. 4 includes a mask area extraction unit 13, a color ratio calculation unit 14 in a mask area, a class type determination unit 15, a character string division and deformation symbol selection unit 16, and a modified symbol arrangement. The connecting portion 17 and the detection frame installing portion 18 are configured.

  In FIG. 1 (a.1), a plurality of modified symbols for embedding information are prepared. This corresponds to a plurality of modified symbols 12 in FIG.

  In FIG. 1 (a.2), n mask regions are extracted from each modified pattern. This process is performed by the extraction part 13 of the mask area.

In FIG. 1 (a.3), m-bit data is held from the ratio of black and white distributed in one mask area. The range of the black and white ratio in the mask area is divided by (2 ^m −1) threshold values and divided into 2 ^m classes to hold m-bit data. Thereby, n × m-bit data is embedded in each modified symbol. This step is processed by the color ratio calculation unit 14 in the mask area.

In FIG. 1 (a.4), each deformation symbol is classified into 2 ^nm classes using n × m-bit data embedded in each deformation symbol. For example, when 4 (= n) mask areas are used for one modified symbol and 1 (= m) bit data is allocated in each mask area, the number of classes is 16 (= 2 ^nm ). It becomes. This step is processed by the class type determination unit 15.

In FIG. 1 (a.5), the information to be embedded is divided into n × m bits, and the corresponding modified symbols are selected from 2 ^nm classes. This process is processed by the character string dividing and deformed symbol selection unit 16.

  In Fig. 1 (a.6), a pattern image in which information data is embedded is generated by regularly arranging selected deformation symbols. This step is processed by the modified symbol arrangement connecting unit 17.

  In FIG. 1 (a.7), a frame for detecting an area of a pattern image in which information data is embedded at the time of shooting is provided. This step is processed by the detection frame setting unit 18.

  By using the information embedding device 11 of FIG. 4, a framed pattern image 19 is generated.

  Next, regarding the first problem solving means, the data reading mechanism is shown in FIG. 1, b. 2, b. 3, b. 4, b. 5, b. It consists of 6 procedures. 5 includes a frame recognition unit 23, an image calibration unit 24, an image division unit 25, a mask region extraction unit 13, a color ratio calculation unit 14 in the mask region, a data decoding unit 26, The data connection unit 27 is included.

  1B, a frame for detecting a pattern image in which information data is embedded is recognized from a photographed image (detection target image 22 in FIG. 5). This step is processed by the frame recognition unit 23.

  In FIG. 1 (b.2), a calibration process is performed to return the captured image to the original rectangular area based on the frame information. This step is processed by the image calibration unit 24.

  In FIG. 1 (b.3), an image is cut out for each modified symbol unit in which each n × m-bit data is embedded. This step is processed by the image dividing unit 25.

  1 (b.4), the same n mask regions as in FIG. 1 (a.2) are extracted from the modified pattern. This process is performed by the extraction part 13 of the mask area.

In FIG. 1 (b.5), m-bit data is decoded from the ratio of black and white in each extracted mask area. Similarly to embedding, the range of the ratio of black and white distributed in one mask area is divided by (2 ^m −1) threshold values and divided into 2 ^m classes to decode m-bit data. As a result, n × m-bit data is decoded for each modified symbol. This step is processed by the color ratio calculation unit 14 and the data decoding unit 26 in the mask area.

  In FIG. 1 (b.6), n × m bits of data decoded in each modified symbol are concatenated according to the arrangement order of the modified symbols, and the embedded information data is read out. This step is processed by the data connection unit 27.

  By using the embedded information detecting device 21 of FIG. 5, embedded information data 28 is obtained.

  Next, the second problem solving means will be described with reference to FIG. FIG. 6 shows an apparatus for realizing the information embedding method of the second problem solving means, and FIG. 7 shows an embedded information detection apparatus.

  The mechanism for embedding information in the pattern image is shown in FIG. 1, a. 8, a. 9, a. 2, a. 3, a. 4, a. 5, a. 10, a. It consists of 11 procedures. Further, the information embedding device 31 shown in FIG. 6 includes a position detection symbol determination unit 33, a mask region extraction unit 13, a color ratio calculation unit 14 in the mask region, a class type determination unit 15, a character string division and a modified symbol It comprises a selection unit 16, a modified symbol arrangement connection unit 34, and an annular arrangement unit 35.

  In Fig. 1 (a.8), a position detection condition is determined in order to detect embedded information data. Here, the position where the center coordinates of the square image area surrounding the modified pattern (hereinafter referred to as the image center) and the barycentric coordinates of the area colored with the pattern (hereinafter referred to as the area center of gravity) are shifted. Detect condition. Therefore, the image center of the deformed pattern and the area centroid of the pattern are first matched for all the deformed symbols. This process corresponds to the determination 32 of the position detection condition in FIG.

  In Fig. 1 (a.9), a deformed symbol matching the position detection condition is determined. Here, a symbol obtained by moving the symbol center of gravity of the symbol from the image center of the deformed symbol is defined as a position detection symbol. This step is processed by the position detection symbol determination unit 33.

  In Fig. 1 (a.10), after the position detection symbols determined in Fig. 1 (a.9) are arranged at the head, the modified symbols selected in Fig. 1 (a.5) are regularly arranged, A pattern image in which information data is embedded is generated. This step is processed by the deformed symbol arrangement connecting unit 34.

  In FIG. 1 (a.11), pattern images in which information data is embedded are arranged more regularly. This step is processed by the annular arrangement connecting portion 35.

  By using the information embedding device 31 of FIG. 6, a frameless pattern image 36 is generated.

  Next, as for the second problem solving means, the data reading mechanism is shown in FIG. 7, b. 8, b. 2, b. 3, b. 4, b. 5, b. It consists of 6 procedures. 7 includes an image region detection unit 43, an information embedding region detection unit 44, an image calibration unit 24, an image division unit 25, a mask region extraction unit 13, and a color ratio calculation in the mask region. Section 14, data decoding section 26, and data linking section 27.

  In FIG.1 (b.7), the symbol in a deformation | transformation symbol is recognized from a detection target image. This step is processed by the image area detection unit 43.

  In FIG.1 (b.8), the symbol in the deformation | transformation symbol which corresponds to position detection conditions is detected. After calculating the area centroid of each symbol, the area centroid deviated from the arrangement of the area centroids is detected, and the area in which the information data is embedded is recognized. This step is processed by the information embedding area detection unit 44.

By using the embedded information detecting device 41 of FIG. 7, embedded information data 28 is obtained.

  FIG. 3 is a framed pattern image (450 × 450) in which 192-bit information is embedded by the first problem solving means of the present invention. For the 64 x 64 pixel modified pattern, four mask areas (8 x 8 pixels) are prepared at a location 12 pixels away from the center of the image, and 1-bit data is held in each mask area. It was. In this way, 4-bit data was embedded in one modified pattern, and 49 pieces (7 vertical and 7 horizontal) were arranged in a two-dimensional lattice pattern. The width of the information data detection frame is 1 pixel.

  The first problem solving of the present invention is performed on image data obtained by monochrome printing this framed pattern image with a printing device, for example, a printer, and photographing it with a camera of a mobile phone substantially parallel to the close-up lens. Table 1 shows the result of calculating the number of error bits by comparing the information data detected by the decoding calculation by the means and the original information data.

The above table shows the detection results when the information-embedded image of FIG. 2 is printed at various sizes and photographed 10 times at a fixed resolution (1200 × 1600 pixels). However, the average bit error rate is the number of bit errors / the number of embedded bits. From this experimental result, it can be seen that the detection accuracy varies depending on the deformation pattern to be used. Further, it is indicated that the embedded information data cannot be completely taken out as the print size is reduced. Therefore, a general error correction code is introduced to ensure detection accuracy.

  FIG. 3 is a frameless pattern image in which an image (616 × 616) in which 192-bit information is embedded is further arranged in a two-dimensional lattice pattern by the second problem solving means of the present invention. For a modified pattern of 88 x 88 pixels, four mask areas (12 x 12 pixels) are prepared at a location 12 pixels away from the center of the image, and each mask area holds 1-bit data. It was. 4-bit data was embedded in one modified pattern and arranged as 49 (7 vertical and 7 horizontal) image blocks in a two-dimensional grid. The position detection condition is that the area centroid of a certain deformed pattern is deviated from the area centroids of surrounding deformed patterns. Therefore, the position detection pattern is obtained by moving the center of gravity of the pattern area by 10 pixels from the center of the deformed pattern image at the city distance. In the deformed symbol in which other data is embedded, the image center of the deformed symbol is matched with the area centroid of the symbol.

  The image data obtained by monochrome printing the frameless pattern image to a size of 15 inches with a printing apparatus, for example, a printer, and using a mobile phone camera in parallel with the close-up lens, is used for the image data of the present invention. Table 2 shows the result of calculating the number of error bits by comparing the information data detected by the decoding calculation by the second problem solving means and the original information data.

The above table shows the detection results when the information-embedded image of FIG. 3 is captured 10 times at a constant resolution (1200 × 1600 pixels). This experimental result shows that the embedded data cannot be completely extracted. Such a case can be dealt with by introducing a general error correction code. In the first sample 1 and the third sample 3, the number of bit errors is large because the position detection symbol has failed to be detected.

Not only image media content through printed matter such as magazines and newspapers, but also large-scale wallpaper, printed matter on cloth, and even images projected on a screen by a projector can be handled as target content. With respect to Claims 1, 2, and 8, as software products for generating a pattern image in which information is embedded, as for Claims 3 and 4, software for providing information acquisition means using a camera image As a product, and with regard to claims 9 and 10, as a software product providing information acquisition means using a camera from wrapping paper, wallpaper, etc., claim 5, claim 6, claim 7, claim 11, claim With regard to the twelfth, thirteenth and thirteenth claims, industrial applications as products in which these software products are combined with hardware devices such as cameras, printers and mobile phones can be considered.

Explanatory diagram showing the process of embedding and reading information into pattern images Example of an image generated by the first problem solving means of the present invention Examples of images generated by the second problem solving means of the present invention The block diagram of the information embedding apparatus by the 1st subject solution means of this invention 1 is a block diagram of an embedded information detection apparatus according to a first problem solving means of the present invention. The block diagram of the information embedding apparatus by the 2nd problem-solving means of this invention The block diagram of the embedded information detection apparatus by the 2nd problem-solving means of this invention

Explanation of symbols

11 Information Embedding Device According to First Solution of the Present Invention 12 Modified Pattern for Embedding Information 13 Mask Area Extraction Section 14 Color Ratio Calculation Section in Mask Area 15 Class Type Determination Section 16 Character String Division and Modification Pattern Selection unit 17 modified symbol arrangement connecting unit 18 detection frame setting unit 19 pattern image with frame 21 embedded information detection device 22 according to the first solving means of the present invention 22 detection target image 23 frame recognition unit 24 image calibration unit 25 image dividing unit 26 Data Decoding Unit 27 Data Linking Unit 28 Embedded Information Data 31 Information Embedding Device According to Second Solution of the Present Invention 32 Determination of Position Detection Condition 33 Position Detection Symbol Determination Unit 34 Modified Symbol Arrangement Linking Unit 35 Ring Arrangement Linking Unit 36 Frame No pattern image 41 Embedded information detection apparatus 42 according to the second solving means of the present invention 42 Detection target Image 43 Image area detection unit 44 Information embedding area detection unit

Claims (14)

Symbol (hereinafter, deformation pattern) generated by transforming the pattern underlying the buried n × m bits of data to, and arranging a plurality of deformed image regularly embed the desired information in the pattern image A method for
For each of the modified symbols, n local regions (mask regions) at a predetermined position are specified, and the color ratio of the mask region is classified into m types according to a predetermined criterion. Based on n × m information determined when bit information is provided in a mask area and the mask areas are arranged in a predetermined order, each deformation image is classified into 2 ^{n × m} types of classes, The digital information to be embedded is divided into units of n × m bits, a modified pattern of a class that matches each divided n × m bit information is selected, and the modified patterns are arranged in a predetermined order. A method for embedding information in a pattern image. 2. The method of embedding information in a pattern image according to claim 1, wherein the color ratio of the mask area is a black and white ratio, and the m bits are 1 bit. The information embedding method according to claim 1 or 2, wherein a detection frame for detecting the deformed symbol is provided in a range in which the deformed symbol of the class matching the n × m bit information obtained by dividing the information to be embedded is arranged. A method for embedding information in a pattern image, characterized by comprising: An apparatus for embedding information in a pattern image by the information embedding method in a pattern image according to claim 1 or 2,
For each of the modified symbols, a mask region extraction part that identifies n local regions (mask regions) at predetermined positions;
A color ratio calculation unit for calculating a color ratio of the mask area, classifying into m types according to a predetermined criterion, and providing m-bit information in the mask area;
Based on n × m information determined when the mask regions are aligned in a predetermined order, each deformation image is converted into 2 ^{n × m} A class type determination unit for classifying into different types of classes,
A character string dividing / symbol selecting unit that divides digital information to be embedded into units of n × m bits and selects a modified symbol of a class that matches each divided n × m bit information;
A modified symbol arrangement connecting portion for arranging the modified symbols in a predetermined order;
An apparatus for embedding information in a pattern image, comprising: An apparatus for embedding information in a pattern image by the information embedding method in a pattern image according to claim 3,
For each of the modified symbols, a mask region extraction part that identifies n local regions (mask regions) at predetermined positions;
A color ratio calculation unit for calculating a color ratio of the mask area, classifying into m types according to a predetermined criterion, and providing m-bit information in the mask area;
Based on n × m information determined when the mask regions are aligned in a predetermined order, each deformation image is converted into 2 ^{n × m} A class type determination unit for classifying into different types of classes,
A character string dividing / symbol selecting unit that divides digital information to be embedded into units of n × m bits and selects a modified symbol of a class that matches each divided n × m bit information;
A modified symbol arrangement connecting part for arranging the modified symbols in a predetermined order;
A detection frame placement unit for providing a detection frame for detecting the modified symbol in a range where the modified symbol of the class matching the n × m bit information obtained by dividing the information to be embedded is arranged;
An apparatus for embedding information in a pattern image, comprising: A pattern image created by the information embedding method according to claim 1, wherein the image is displayed on a display device or output by a printing device. A method for extracting information from a pattern image in which information is embedded by the information embedding method according to claim 3,
A detection frame is detected from the pattern image in which the information is embedded, a plurality of modified symbols arranged in the range of the detection frame are identified, n mask areas are identified for each modified symbol, and each mask is identified. Decodes m-bit data from the region, decodes n × m-bit information embedded in each modified symbol, and connects each n × m-bit information of each modified symbol according to a predetermined arrangement order for the modified symbol. An information retrieval method characterized by: An apparatus for extracting embedded information from a pattern image in which information is embedded by the information extracting method according to claim 7,
A frame recognition unit for detecting a detection frame from the pattern image in which the information is embedded;
An image dividing unit for specifying a plurality of modified symbols arranged within a range of a detection frame detected by the frame recognition unit;
For each modified pattern specified by the image dividing unit, a mask region extracting unit that specifies n mask regions;
A color ratio calculation unit for calculating a ratio of colors in each of the mask regions;
A data decoding unit that decodes information of n × m bits incorporated in each modified pattern by decoding m-bit data from the color ratio calculated by the color ratio calculation unit;
A data linking unit for linking each n × m bit information of each modified symbol in accordance with a predetermined arrangement order for the modified symbol;
An embedded information extraction device comprising: 3. The information embedding method according to claim 1, wherein a position for detecting the head position at the head of a range in which a modified pattern of a class matching the n × m bit information obtained by dividing the information to be embedded is arranged. A method for embedding information in a pattern image, comprising providing a detection symbol. An apparatus for embedding information in a pattern image by the information embedding method in a pattern image according to claim 9,
A position detection design determining unit for determining a position detection design to be a position detection condition from the basic design or the modified design;
For each modified symbol excluding the position detection symbol, a mask region extraction part that identifies n local regions (mask regions) at a predetermined position;
A color ratio calculation unit for calculating a color ratio of the mask area, classifying into m types according to a predetermined criterion, and providing m-bit information in the mask area;
Based on n × m information determined when the mask regions are aligned in a predetermined order, each deformation image is converted into 2 ^{n × m} A class type determination unit for classifying into different types of classes,
A character string dividing / symbol selecting unit that divides digital information to be embedded into units of n × m bits and selects a modified symbol of a class that matches each divided n × m bit information;
A modified symbol arrangement connecting unit that arranges the other modified symbols in a predetermined order, with the position detection symbol at the top;
An apparatus for embedding information in a pattern image, comprising: A pattern image created by the information embedding method according to claim 9, which is displayed on a display device or output by a printing device. A method for extracting information from a pattern image in which information is embedded by the information embedding method according to claim 9,
A position detection symbol is detected from the pattern image in which the information is embedded, a plurality of deformation symbols arranged with a predetermined positional relationship with respect to the position detection symbol are specified, and n number of deformation symbols are determined for each deformation symbol. The mask area is specified, m-bit data is decoded from each mask area, n × m-bit information embedded in each modified symbol is decoded, and each n × m-bit information of each modified symbol is decoded. An information extraction method comprising connecting symbols according to a predetermined arrangement order. An apparatus for extracting embedded information from a pattern image in which information is embedded by the information extracting method according to claim 7,
An image area detection unit for detecting an image area from the detection target image;
An information embedding area detection unit in which a position detecting pattern is detected from the image area and a pattern in which information is embedded is arranged;
An image dividing unit for specifying a plurality of modified symbols arranged in the range of the information embedding region detected by the information embedding region detecting unit;
For each modified pattern specified by the image dividing unit, a mask region extracting unit that specifies n mask regions;
A color ratio calculation unit for calculating a ratio of colors in each of the mask regions;
A data decoding unit that decodes information of n × m bits incorporated in each modified pattern by decoding m-bit data from the color ratio calculated by the color ratio calculation unit;
A data linking unit for linking each n × m bit information of each modified symbol in accordance with a predetermined arrangement order for the modified symbol;
An embedded information extraction device comprising: An information embedding device according to claim 5 and an embedding information extraction device according to claim 8, or an information embedding device according to claim 10 and an embedding information extraction device according to claim 13. An image processing apparatus comprising the image processing apparatus.