JP2007036657A - Data identifying device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a data identifying device which can suppress incorrect data identification. <P>SOLUTION: The device reads an image formed on a paper based on predetermined image data, and discriminates the data of which a block takes charge. The image data are constituted by dividing an image region into a plurality of blocks, further dividing each block into a plurality of regions, and increasing or decreasing the pixel value of each region and each block with an increasing and decreasing value assigned by region in the block by a pattern, defined beforehand according to data in charge of the block. It is determined that a sum of decreasing values in a block is set to 0. The decreasing value of a region r is inferred, by acquiring sums of pixel values Q1, Q2, Q8 of 3 blocks B1, B2, B8 adjoining the region r1, obtaining the total Q1+Q2+Q8, and subtracting the value of the multiplied value of the total by a predetermined value A, from a total Rr of pixel values in the region r1. Executing the presumption for each region of the all blocks, data of each block are inferred from the decreasing value of each region. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a data identification device, and more particularly, to a data identification device that identifies predetermined data by reading an image formed on an image forming medium based on predetermined image data.

  In recent years, digital watermark technology that adds electronic information to image data in a format that cannot be recognized by the human eye at first glance has been extensively researched and developed, and has begun to be used in various forms ( For example, see Patent Document 1). In Patent Document 1, when electronic information is added to image data, for example, the entire image area is divided into a plurality of blocks each composed of a plurality of pixels, and each block is further divided in the vertical and horizontal directions. Dividing into a total of four areas, the pixel value of each pixel in each block is increased or decreased allocated in a predetermined pattern according to the data (0 or 1) assigned to the block for each area in the block Increases or decreases by value.

  Specifically, in the case of taking charge of 1, the pixel values in the upper left and lower right areas are decreased and the pixel values in the upper right and lower left areas are increased, and in the case of taking charge of 0, the upper left and lower right areas. The pixel value of the area is increased, and the pixel values of the upper right and lower left areas are decreased.

  Thus, an image is formed on a sheet based on the image data to which the electronic information is added.

  And the image formed in this way is read and the data which each block takes charge is identified.

By the way, in order to identify the data handled by each block as described above, it is necessary to clearly distinguish and recognize each area obtained by dividing the block. Therefore, in the invention described in Patent Document 1, the pattern includes “two vertical and horizontal edges passing through the center”, “the absolute value of the pixel value is the largest at the center, and the farther away from the center, the more the pixel value is. “Small” and “both are rectangular blocks of n × m pixels”. Thereby, the boundary of the center part is clear and each region can be clearly distinguished and recognized.
JP 2004-140764 A

  However, in the image based on the original image data, when the boundary between the high density area and the low density area is located along the diagonal line of the block, for example, the upper left, the upper right, and the lower right of the four areas. When the high density area corresponds to this area and only the lower left area corresponds to the low density area, it may be recognized as 0 although it should be recognized as 1 originally.

  The present invention has been made in view of the above facts, and an object of the present invention is to provide a data identification device capable of suppressing erroneous recognition of data.

  In order to achieve the above object, the invention described in claim 1 divides a predetermined image area into a plurality of blocks each composed of a plurality of pixels, further divides each block into a plurality of areas, An image forming medium based on image data configured by increasing / decreasing the pixel value of a pixel by an increase / decrease value assigned to each area in the block in a predetermined pattern according to data handled by the block A data identification device that reads the image formed in the block and identifies data handled by each block, and acquires image data of an identification block that is a target for identifying the data and an adjacent block adjacent to the identification block Based on the acquisition unit and the image data of the identification block and the adjacent block acquired by the acquisition unit, the data handled by the identification block is identified. Includes an identification unit that, the.

  That is, the present invention is a data identification device that reads an image formed on an image forming medium and identifies data.

  Here, the image that can be identified by the data identification device of the present invention is an image formed on the image forming medium based on the next image data. That is, a predetermined image area is divided into a plurality of blocks each composed of a plurality of pixels, each block is further divided into a plurality of areas, and the pixel value of each pixel of each block is determined according to the data handled by the block. The image data is configured to increase or decrease with an increase / decrease value assigned to each area in the block in a predetermined pattern.

  As described above, the image data obtained by reading the image formed on the image forming medium is obtained by converting the pixel value of each pixel of each block into data assigned to the block for each area in the block. Accordingly, since it corresponds to image data configured by increasing / decreasing with an increase / decrease value assigned in a predetermined pattern, in order to identify data handled by the block, from the image data, each pixel of each block is identified. It is necessary to divide the original pixel value and the increase / decrease value assigned in a predetermined pattern according to the data handled by the block.

  Here, there are various methods for identifying the data handled by the block from the image data obtained by reading the image formed on the image forming medium. Only image data is used. Therefore, conventionally, when there is a boundary between the high density portion and the low density portion in the image of the block, the image data of the identification block that is a target for identifying data from the relationship between the boundary and the contents of the pattern. In some cases, misidentification of data was sometimes caused by using only.

  Therefore, the acquisition unit of the present invention acquires an identification block that is a target for identifying the data and image data of an adjacent block adjacent to the identification block.

  The identification unit identifies data handled by the identification block based on the identification block acquired by the acquisition unit and the image data of the adjacent block.

  In this way, when identifying data for which a block is responsible, even if the image data of the identification block alone is erroneously identified, the image of the identification block to be identified and the adjacent block adjacent to the identification block Since data is acquired, the increase / decrease value in the image data of the identification block can be inferred using the image data of the adjacent block, and erroneous identification of the data handled by the block can be suppressed.

  According to a second aspect of the present invention, in the first aspect of the present invention, an increase / decrease value assigned to each area in the block is determined to be a predetermined value in the block, and the identification means is at least 1 The sum of pixel values based on the image data of the two adjacent blocks is calculated, and the sum of increase / decrease values of each area of the identification block is estimated using the calculated sum of the pixel values based on the image data of the adjacent blocks Then, the data is identified based on the estimated sum of the increase / decrease values of the respective areas of the identification block.

  Here, the increase / decrease value assigned to each area in the block is determined to be a predetermined value, for example, 0 when added in the block. Accordingly, when the sum of the pixel values based on the image data of the adjacent block is calculated, the increase / decrease value assigned to the block becomes a predetermined value (for example, 0), so that only the image information of the adjacent block is obtained. Since it is close to the image information of the adjacent block and the image information of the identification block, if the sum of the pixel values based on the image data of the adjacent block is used, the sum of the increase / decrease values of each area of the identification block can be estimated.

  As described above, since the increase / decrease value in the image data of the identification block is estimated using the image data of the adjacent block, it is possible to suppress erroneous identification of data handled by the block.

  According to a third aspect of the present invention, in the first or second aspect of the present invention, when the identification means uses image data of a plurality of adjacent blocks, the sum of increase / decrease values of each area of the identification block is estimated. In this case, the total sum of the pixel values based on the image data of a plurality of adjacent blocks is used by being divided by a value equal to or greater than the total number of regions of the used adjacent blocks.

  According to a fourth aspect of the present invention, the discriminating unit makes the discriminating impossible when the difference in the sum of the increase / decrease values of each area of the discrimination block is smaller than the minimum difference determined based on a predetermined pattern. . Thereby, it is possible to suppress erroneous identification of data.

  According to a fifth aspect of the present invention, each block takes charge of different first data or second data, and the pattern includes the first pattern and the second data determined according to the first data. And a second pattern determined in accordance with.

  The acquisition unit acquires image data of a plurality of adjacent blocks, and the identification unit determines that the first data corresponding to the first pattern is based on the acquired image data of the plurality of adjacent blocks. It is determined whether or not the second data corresponding to the second pattern is easily identified, and when it is determined that the first data is easily identified as the second data, the image data of the identification block is determined as the first data. The data is corrected so as to be easily identified, and the data is identified based on the corrected image data.

  In this way, each block is responsible for different first data or second data, and the pattern is a second pattern determined according to the first pattern and the second data determined according to the first data. When it is determined that the first data corresponding to the first pattern is easily discriminated from the second data corresponding to the second pattern at the time of data identification, the image of the identification block Since the data is corrected so that the first data can be easily identified, erroneous identification of the data can be suppressed.

  As described above, according to the present invention, when identifying data handled by a block, even if only the image data of the identification block is erroneously identified, the identification block to be identified is adjacent to the identification block. Since the image data of the adjacent block is acquired, the increase / decrease value in the image data of the identification block can be inferred using the image data of the adjacent block, and the erroneous identification of the data handled by the block can be suppressed. it can.

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

  FIG. 1 is a block diagram showing an example of the configuration on the side where additional information is embedded in the embodiment of the present invention. In the figure, 11 is a pattern size input unit, 12 is an embedding intensity input unit, 13 is a pattern attenuation rate input unit, 14 is a pattern creation unit, 15 is an image data input unit, 16 is an image data storage unit, and 17 is an additional information input. , 18 is an additional information encoding unit, 19 is an embedding position control unit, 20 is a pattern selection unit, 21 is a pattern superimposing unit, and 22 is a reference code referring to an image output unit.

  The pattern size input unit 11 is used to input and set a pattern size designated by the user through a personal computer or an operation panel (not shown). The embedding strength input unit 12 inputs and sets the embedding strength designated by the user through a personal computer or an operation panel (not shown). Further, the pattern attenuation factor input unit 13 is used to input and set a pattern attenuation factor designated by the user through a personal computer or an operation panel (not shown). These may be configured without providing each when using a preset fixed value.

  The pattern creating unit 14 creates two patterns based on the set pattern size, embedding strength, and pattern attenuation rate. Details of the pattern creating unit 14 will be described later.

  The image data input unit 15 receives input image data. For example, the image data can be acquired in various forms such as having a communication function, receiving image data from an external device, receiving it from software via an OS, or opening and reading a file. The input image data is multi-value data, and may be an arbitrary image created by a personal computer (not shown), a natural image or a CG (Computer Graphics) image input by a digital camera or a scanner. The image data storage unit 16 is used for recording input image data, temporarily holding work data being processed, and holding output data.

  The additional information input unit 17 receives input of additional information embedded in input image data from various sources such as a personal computer, an operation panel, software, and a file (not shown). The additional information may be various information such as a character string, a number, or image data.

  The additional information encoding unit 18 converts the additional information input by the additional information input unit 17 into a predetermined encoding format, and creates embedded information that is actually embedded in the image data. Details of the additional information encoding unit 18 will be described later. It is also possible to embed without encoding.

  The embedding position control unit 19 designates a position at which embedding information is embedded in the image data held in the image data storage unit 16 in accordance with a predetermined embedding format.

  The pattern selection unit 20 selects one of the two patterns created by the pattern creation unit 14 based on the embedded information created by the additional information encoding unit 18.

  The pattern superimposing unit 21 adds, for example, the pattern selected by the pattern selecting unit 20 to the image block existing at the address of the image data storage unit 16 specified by the embedding position control unit 19 and superimposes the pattern. Embed in. When the added value exceeds the maximum value (for example, 255), the value is set to the maximum value (255), and when the added value becomes a negative value, the value is set to 0.

  The image output unit 22 outputs an image in which additional information is embedded via an output device such as a printer, software, or a communication line.

  The outline of one configuration example on the side where the additional information is embedded in the embodiment of the present invention has been described above. Next, the main configuration among the above-described configurations will be described in detail.

First, the pattern creation unit 14 will be described in detail. The pattern creation unit 14 creates two patterns based on the values input and set by the pattern size input unit 11, the embedding strength input unit 12, and the pattern attenuation rate input unit 13 described above. The two patterns have the following characteristics.
When all pixel values (increase / decrease values) of corresponding pixels of the two patterns are added, all elements become zero.
When the pixel values (increase / decrease values) of all the pixels in each pattern are added, it becomes 0.
Each pattern has discontinuous pixel values called two or more edges that pass through the center and have different directions. The direction of the edge can be, for example, a direction along a vertical line and a horizontal line.
further,
The absolute value of the pixel value of each pattern is the largest at the center and decreases as the distance from the center increases.
It is good to have the characteristics. Also, regarding the shape, here
Both are composed of rectangular blocks of the same size with n × m pixels.
It shall have the characteristics.

  FIG. 2 is an explanatory diagram of an example of a pattern to be embedded. An example of the pattern having the above-described features is as shown in FIG. 2A is a basic pattern meaning additional information “1”, and FIG. 2B is a basic pattern meaning additional information “0”. All of these elements are shown in FIG. 2C. Expressions such as Expression (1) or Expression (2) are multiplied. Thereby, for example, patterns as shown in FIGS. 2D and 2E are generated. 2D and 2E, the difference in density is indicated by the difference in hatching for the sake of illustration. 2A corresponds to the first pattern, and one of the basic patterns meaning additional information “0” corresponds to the first pattern, and the other corresponds to the second pattern. Corresponds to the pattern.

  Here, the size of the basic pattern is set by the pattern size input unit 11. The example shown in FIG. 2 is an example in which the pattern size is 8 × 8. In equations (1) and (2), C is the embedding strength input by the embedding strength input unit 12, and α is the pattern attenuation rate set by the pattern attenuation rate input unit 13. x represents the horizontal axis, y represents the coordinate of the vertical axis, and the center of the pattern is the origin.

  The features of these patterns are for suppressing the influence on the image quality as much as possible and for facilitating the detection, and can be easily detected by the configuration on the information detection side of the present invention described later. Parameters such as the pattern size, pattern attenuation rate, and embedding strength are usually set for each output device in consideration of the image quality and detection rate. However, these parameters are set in advance on the information detection side described later. There is no need to know.

  The pattern used in the present invention is not limited to the example shown in FIG. 2. For example, a pattern such as a triangular wave is used instead of the expressions (1) and (2), and the expressions (1) and (2) are used. Any function may be used as an expression. It is also possible to omit the exponential function portion in these equations, or to use the patterns shown in FIGS. 2A and 2B as they are without using these equations. Further, in the example shown in FIG. 2, the edge direction is set to the vertical / horizontal direction. However, for example, it is optional if the side for extracting additional information, which will be described later, is combined with the edge extraction direction, such as edges at 45 degrees and 135 degrees. The edge direction may be sufficient.

  Next, the additional information encoding unit 18 will be described in detail. First, an example of a code format used in the present invention will be described. The code format includes a physical format that specifies the position and order of recording the embedded information, and a logical format that defines how the embedded information is encoded and decoded within the physical format. There is.

  FIG. 3 is an explanatory diagram of an example of a physical format in an example of a code format used in the present invention. In the figure, 31 is a pattern size, 32 is a macroblock size, and 33 is an input image. The pattern size 31 indicates the size of the pattern set by the pattern size input unit 11, and the macroblock size 32 is a collection of the pattern size 31 in a matrix of Y rows and X columns. The input image 33 is an image to be embedded with a pattern.

  The embedding position control unit 19 determines the macroblocks that can be arranged in the input image 33 from the size of the input image 33 to be embedded, the matrix size (Y, X) of the macroblock set in advance, and the pattern size. The number is calculated, and the macro blocks are arranged as close to the center of the input image 33 as possible without gaps. The macro block is accessed from the upper left to the lower right, that is, MB11, MB12, MB13, MB21,..., MB33 in the example shown in FIG. Address control in the order of

  By regularly arranging the blocks in which the pattern is embedded in this way, the block size and the block position can be easily detected on the side of extracting additional information to be described later.

  FIG. 4 is an explanatory diagram of an example of a logical format in an example of a code format used in the present invention. In the figure, 41 is a basic logical format, 42 is a cue header, 43 is encoding method information, 44 is a sequence number, 45 is effective code number information, and 46 is encoding information.

  The logical format is composed of one or more basic logical formats 41 as shown in FIG. The basic logical format 41 includes a cue header 42, encoding method information 43, sequence number 44, effective code number information 45, and encoding information 46. The size of the basic logical format 41 is equal to the size of the macroblock 32 and is X × Y bits. The cue header 42 is used to specify the position of the macro block 32, and the common header 42 is used for all the macro blocks 32. The encoding method information 43 indicates what error correction method the encoding information 46 to be described later is encoded in, and is also used in common for all the macroblocks 32. The sequence number 44 is used when the additional information received by the additional information input unit 17 is of a size that cannot be accommodated in one macroblock 32. After the additional information is encoded, it can be accommodated in the macroblock 32. This is divided into sizes, and sequence numbers are added to them in ascending order from “1”. When the encoded additional information has a length that can be accommodated in one macroblock 32, the sequence number is “1”. The effective code number information 45 indicates the effective code number of the encoded information accommodated in the last macroblock when the encoded additional information is divided, and the effective code number information other than the last block is All become "0". Note that the part to be error-corrected encoded includes not only the encoded information 46 but also the sequence number 44 and the effective code number information 45.

  FIG. 5 is a flowchart showing an example of the operation of the additional information encoding unit 18 on the side where the additional information is embedded in the first embodiment of the present invention. In S101, the additional information input from the additional information input unit 17 is replaced with binary information. For example, if a character string is received as additional information, it is converted into ASCII code or the like and converted into binary information. In S102, error correction coding is performed on the binary information obtained in S101. In S103, it is calculated from the code length of the information encoded in S102 whether or not it can be accommodated in one macroblock, and if not, it is divided. In S104, the information of a plurality of basic logical formats 41 is created by adding the cue header 42, the encoding method information 43, the sequence number 44, and the effective code number information 45 to the divided encoded information 46. In S105, information of a plurality of basic logical formats 41 created in S104 is embedded in order from the first macroblock, and embedding is repeatedly performed so that information is embedded in all macroblocks.

  Below, an example of operation | movement of the above-mentioned additional information encoding part 18 is further demonstrated using a specific example. Here, the physical format is 16 rows and 8 columns, that is, Y = 16 and X = 8. The cue header is 5 bits and the value is “11111”. Further, it is assumed that the encoding method is 3 bits, the value is “000”, and the code length is 15 bits and the check bit is 4 bits, indicating a Hamming code. Further, description will be made assuming that the sequence number is 7 bits and the number of effective codes is 4 bits.

  In the flowchart shown in FIG. 5, after the additional information is replaced with binary information in S101, error correction coding is performed on the binary information obtained in S101 in S102. In the case of a Hamming code with a code length of 15 bits and a check bit of 4 bits, binary information is taken 11 bits from the beginning, and a 4-bit check bit is added to obtain a 15-bit Hamming code. This is repeated until the end of the binary information. When the bit length of the binary information is not a multiple of 11 bits, all the missing bits are filled with the value “1” so as to be a multiple of 11 bits.

  In S103, it is calculated from the code length of the information encoded in S102 whether or not it can be accommodated in one macroblock, and if not, it is divided. Here, the macroblock size is 16 × 8 = 128, the cue header is 5 bits, the encoding method is 3 bits, the sequence number is 7 bits, the effective code number information is 4 bits, and the code length is 15 as the encoding method. A Hamming code of 4 bits and check bits is used. Therefore, the sequence number and valid code number information require 15 bits. As a result, the area for the encoded information 46 is 128− (5 + 3 + 15) = 105 bits. Therefore, if the encoded information exceeds 105 bits, multiple macroblocks are required to embed this information. When a plurality of macroblocks are required, it is divided every 105 bits. In S104, a plurality of basic logical formats 41 are created by adding the cue header 42, encoding method information 43, sequence number 44, and effective code number information 45 to the divided encoded information 46.

  In S105, information of a plurality of basic logical formats 41 created in S104 is embedded in order from the first macroblock, and embedding is repeatedly performed so that information is embedded in all macroblocks. For example, when the number of macroblocks 32 is nine and the maximum sequence number is 4 as in the example shown in FIG. 3, the basic logical format of sequence number 1 is stored in MB11, MB12, MB13, and MB21. The information, the basic logical format information of sequence number 2, the basic logical format information of sequence number 3, and the basic logical format information of sequence number 4 are embedded. Further, the basic logical format information of sequence number 1, the basic logical format information of sequence number 2, the basic logical format information of sequence number 3, and the basic logical format information of sequence number 4 are again stored in MB22, MB23, MB31, and MB32. And the basic logical format information of sequence number 1 is embedded in MB33.

  As will be described later, in order to decode this additional information, the decoding side only needs to know the macroblock size (Y, X) and the logical format, and the block size at the time of embedding, the output device and the input device Information such as resolution is not required. In addition, regarding the image quality, by using a pattern in which the amplitude is attenuated, the central portion of the pattern is particularly different from the original image. However, since this pattern is embedded almost regularly in the entire image at regular intervals, Even if you know that it is different from the image, you can suppress the sense of incongruity. In addition, even if the block size is made as small as possible within the range where the detection rate does not drop so much, or even if the block size cannot be reduced, the attenuation rate is set to an appropriate value so that almost no deterioration in image quality is felt compared to the original image. It can be suppressed to the extent.

  The configuration example on the side where the additional information is embedded has been described above in the first embodiment of the present invention. Next, in the first embodiment of the present invention, a configuration example on the side of extracting additional information from image data in which additional information is embedded in units of blocks will be described.

  FIG. 6 is a block diagram showing a configuration example of the side for extracting additional information in the first embodiment of the present invention. In the figure, 51 is an image data input unit, 52 is an image data storage unit, 53 is an input image inclination correction unit, 54 is a block size estimation unit, 55 is a block position detection unit, 56 is an additional information identification unit, and 57 is additional information. This is a reference code for referring to the decoding unit. The input image data is image data obtained from an image generated by the image processing apparatus or the image processing method shown in the configuration example on the side where the additional information is embedded as described above and printed out from the printing device.

  The image data input unit 51 has an interface with an image reading device such as a scanner or a digital camera, and inputs print image data in which additional information is embedded through this interface. The image data input unit 51 also has a function of converting print image data acquired by an image reading device such as a scanner or a digital camera into uncompressed data when the image data is compressed.

  The image data storage unit 52 stores the print image data obtained by the image data input unit 51, and stores intermediate results of the calculation.

  The input image inclination correction unit 53 has a function of detecting the inclination of the image in the input print image data and correcting the inclination. FIG. 7 is an explanatory diagram of an example of image inclination correction processing in the input image inclination correction unit. In the image inclination correction, for example, the input image is projected in the vertical direction and the horizontal direction as shown in FIG. 7 while being rotated. An angle that minimizes the range in which the height of the projected waveform is equal to or greater than a predetermined threshold value is estimated as an inclination angle, and rotation correction may be performed by the inclination angle.

  The block size estimation unit 54 estimates the block size in which the additional information is embedded from the print image data whose inclination has been corrected. The block size when the additional information is embedded may have been changed to a different block size through print output and input, and this block size estimation unit 54 estimates the changed block size in some cases. Yes. The block size can be estimated using the feature that the embedded pattern has edge components in a predetermined direction (for example, vertical and horizontal directions). For example, an edge is extracted from print image data that has undergone tilt correction to create an edge extracted image, and an image in which vertical and horizontal edge components are extracted from the edge extracted image is created, and the peak of the autocorrelation function of the image is created. The block size can be estimated from the position. Details of the processing in the block size estimation unit 54 will be described later.

  Based on the block size estimated by the block size estimation unit 54, the block position detection unit 55 detects the block position in which the pattern of the additional information is embedded from the print image data enlarged / reduced at an unknown magnification. The block position can be detected by utilizing the correlation between the mask image created by extracting only positive or negative polarity information from one of the patterns and the image in which the additional information is embedded. For example, a mask image corresponding to the block size obtained by the block size estimation unit 54 is created, a correlation operation is performed between the mask image and the print image data whose inclination is corrected, and the value is calculated from the image of the correlation operation result. Only the points where becomes the maximum or the minimum are extracted, are projected in the vertical direction and the horizontal direction, and the block position can be detected from the projection waveform and the block size obtained by the block size estimation unit 54. It is only necessary to create a mask image from either one of the patterns. The two patterns have opposite polarities. When a mask image is created from the other pattern, the maximum and minimum values are simply reversed. Because it is only. Details of processing in the block position detection unit 55 will also be described later.

  The additional information identification unit 56 is controlled by an additional information decoding unit 57 to be described later, and identifies additional information embedded in the block whose position and size are detected by the block position detection unit 55 and the block size estimation unit 54. The additional information identification process can be performed using the magnitude relation of the sum of the pixel values of the region divided into four by the edge in a predetermined direction. For example, the detected block is divided into four regions in the vertical and horizontal directions, the sum of all the pixels included in the four regions is obtained, and the additional information can be identified based on the magnitude relationship of the four sum values. . Details of the processing in the additional information identification unit 56 will be described later.

  The additional information decoding unit 57 reconstructs the original additional information embedded by assembling the individual information identified by the additional information identifying unit 56 according to a pre-defined format, and then decoding the information. . Details of the processing of the additional information decoding unit 57 will be described later.

  The outline of the configuration on the side for extracting additional information according to the first embodiment of the present invention has been described above. Next, the main part of the above-described configuration will be further described.

  FIG. 8 is a flowchart illustrating an example of the operation of the block size estimation unit 54. First, in S111, an edge extracted image is obtained by applying a differential filter such as a Sobel filter to the input image. Of course, the edge extraction method is arbitrary, such as using a Prewitt type or Kirsh type filter.

  Next, in S112, edge components in a predetermined direction, here in the horizontal and vertical directions, are further extracted from the edge extraction image obtained in S111. FIG. 9 is an explanatory diagram of an example of a mask image for extracting horizontal and vertical edges. As one method for extracting edge components in the horizontal and vertical directions from the edge extraction image, for example, cross-correlation may be calculated with a cross-shaped mask as shown in FIG. As a result, an edge image in which the edges of the horizontal and vertical components and their intersections are emphasized is created. If the additional information pattern is a pattern in which vertical and horizontal edges exist, the created edge image has a grid-like edge passing through the center of the rectangle (pattern size 31) shown in FIG. . Edges obtained from this pattern and vertical and horizontal edges existing in the original image exist in the edge image.

  In S113, an autocorrelation function is obtained from the edge image created in S112. With this autocorrelation function, only the edges obtained from the additional information pattern are extracted. As the autocorrelation offset range obtained at this time, it is usually sufficient to calculate about (2, 2) to (20, 20). In the edge image created in S112, vertical and horizontal line segments arranged at almost equal intervals are extracted. Therefore, if an offset at which the autocorrelation function is maximized is detected, it becomes the block size after scaling. You can think of it as a match. Therefore, in S114, an offset that maximizes the autocorrelation function may be estimated as the block size. The offsets (0, 0), (0, 1), (1, 0), (1, 1) are excluded from the calculation because of the autocorrelation properties of the images. This is because there is a property that the autocorrelation value is high, and it is usually considered that such a small value is not possible as the block size.

  As described above, the block size information for decoding the additional information can be obtained from the print image data without knowing the pattern size when the additional information is embedded, the output resolution, and the input resolution. However, it should be noted that the block size value obtained here is an integer value. In the combination of a printer and a scanner, the resolution used is usually a combination of 400 dpi, 600 dpi, 1200 dpi, etc., and the corresponding block size of the embedded image whose resolution is converted is often an integer, but it is input by a digital camera. In this case, since the input resolution depends on the distance between the digital camera and the print image that is the subject, the corresponding block size of the print image data subjected to resolution conversion is not always an integer. Therefore, it can be said that the block size calculated by the block size estimation unit 54 is an approximate value. However, since the block size obtained here is corrected by the block position detection unit 55 described below, there is no problem with the approximate value.

  Next, details of the block position detection unit 55 will be described. FIG. 10 is a flowchart illustrating an example of the operation of the block position detection unit 55. First, in S121, a mask image is created based on the block size estimated by the block size estimation unit 54. FIG. 11 is an explanatory diagram of an example of a mask image used in the block position detection unit 55. As the mask image created in S121, for example, a mask image as shown in FIG. 11A can be created when the block size is an even number, and as shown in FIG. 11B when the block size is an odd number. . The characteristics of these mask images are such that when the mask is divided into four regions on the vertical and horizontal axes, all the upper right and lower left regions are +1, and all the lower right and upper left regions are -1. This is equivalent to extracting only positive or negative polarity information from one of the two embedding patterns. However, when the block size is an odd number, the portion overlapping the vertical and horizontal axes is set to zero. This mask image corresponds to the pattern of the additional image shown in FIG.

  Next, in S122, a correlation image is created by calculating a cross-correlation between the mask image created in S121 and the print image data. Here, as can be seen by comparing the pattern of the additional image shown in FIG. 2 and the pattern of the mask image shown in FIG. 11, this correlation value is the same between the block in which the additional information “1” is embedded and the mask. When it overlaps, it tends to become maximum, and conversely, when the block in which the additional information “0” is embedded and the mask just overlap, it tends to become minimum. This tendency is particularly likely when the original image corresponding to the position of the block before embedding is flat. Conversely, when the original image corresponding to the position of the block before embedding has a local edge or the like, it does not necessarily become the maximum or the minimum when the embedded block and the mask just overlap. However, since this influence is reduced by the projection method described later, this is not a problem unless the image has extremely many edges.

  In S123, only points that are maximum or minimum are extracted from the correlation image created in S122. First, the maximum point extraction method will be described. In the extraction of local maximum points, a correlation value image is scanned in the order of raster scanning, and pixel values are compared within a 3 × 3 window, and all marks other than the pixel position showing the maximum value are marked as not being candidates for local maximum values. To do. Further, when the pixel position indicating the maximum value has already been marked, the pixel value cannot be the maximum value, and thus a mark is added. This operation is performed from the upper left to the lower right of the correlation image. As a result, the unmarked pixel position becomes the position showing the maximum value, and by setting all the pixel values at the position where the mark is added to 0, only the position where the maximum value exists and the maximum value are extracted. . In addition, the minimum value is extracted by first inverting the correlation value image and then performing the same operation as extracting the maximum value, and only the position where the minimum value exists and the minimum value are extracted. Then, an extreme value image can be obtained by adding the maximum value image and the minimum value image. FIG. 12 is an explanatory diagram of an example of an extreme value image created by the block position detection unit 55 and an example of a process for obtaining a block position from the extreme value image. The maximum value and the minimum value obtained as described above are indicated by white circles in FIG. The maximum value and the minimum value are detected at substantially the center position of each pattern.

  In S124, the extreme value image obtained in S123 is projected in the vertical and horizontal directions. By arranging each block vertically and horizontally, as shown in FIG. 12, a projection waveform having peaks at substantially constant intervals in the vertical and horizontal directions can be obtained.

  Next, in S125, an accurate block position is estimated from the peak positions of the projection waveforms in the vertical and horizontal directions obtained in S124. Specifically, the peak position at the extreme end is first obtained, and then the next peak position is sequentially searched in the range of the block size ± δ obtained by the block size estimation unit 54, thereby obtaining a vertical or horizontal direction. The peak position may be detected and a combination of vertical and horizontal peak positions may be used as each block position. In FIG. 12, the peak position is indicated by an arrow line, and these combinations become block positions. Here, the value of δ is preferably about 1 if the block size is 8 or less, and about 2 if the block size is larger than 8.

  As described in the description of S121, when there is a locally strong edge in the original image, the position of the maximum point or the minimum point obtained from the block including the edge is determined from the position interval of the extreme value obtained from the flat portion. There is a possibility of shifting. However, this variation is greatly reduced by searching for the above-mentioned projection method and peak positions at substantially constant intervals.

  Next, details of the operation in the additional information identifying unit 56 will be described. The additional information identifying unit 56 is controlled by the additional information decoding unit 57, and is added to the block based on the block position information obtained by the block position detecting unit 55 and the block size information obtained by the block size estimating unit 54. It identifies information.

  FIG. 13 is a flowchart illustrating an example of the operation of the additional information identification unit 56. First, in S131, a calculation window is set in which a block is divided into four areas in the vertical and horizontal directions based on the block size information. FIG. 14 is an explanatory diagram of an example of a calculation window used in the additional information identification unit 56. The size of this calculation window is equal to the block size estimated by the block size estimation unit 54, and is divided into four areas on the vertical and horizontal axes as shown in FIG. However, as shown in FIG. 14B, when the block size is an odd number, the portion overlapping the vertical and horizontal axes is not included in the region. Hereinafter, the upper right area is called R1, the upper left area is called R2, the lower left area is called R3, and the lower right area is called R4.

  Next, in S132, the calculation window created in S131 is applied to the block detected by the block position detection unit 55, and the sum of the pixel values included in each region is obtained. Hereinafter, the sum of the image values in the region R1 is also referred to as R1 unless it is confused with the region name. The same applies to R2, R3, and R4.

  In S133, it is determined whether the additional information embedded in the block is “1”, “0”, or indistinguishable from the magnitude relationship of the total values R1, R2, R3, R4. This determination is performed as follows.

  Next, the additional information identification processing in step 133 will be described in detail with reference to FIG. In step 202, the variable b for identifying each block is initialized to 0. In step 204, the variable b is incremented by 1. In step 206, the variable r for identifying the area in the block b identified by the variable b is set to 0. In step 208, the variable r is incremented by one.

  In step 210, the sum of pixel values of three blocks adjacent to the region r is obtained. That is, as shown in FIG. 16, for example, blocks B1, B2, and B8 exist as blocks adjacent to the region r1. The sum of the pixel values of the block B1 is Q1, the sum of the pixel values of the block B2 is Q2, and the sum of the block B8 is Q8. In step 210, the sum Qr1 + Qr2 + Qr3 (Q1, Q2, Q8 in the above example) of the pixel values of three blocks adjacent to the region r is acquired. In step 212, the subtraction value Ur of the region r is estimated by calculating the following equation using the total sum of the three obtained blocks and the sum Rr of the region r pixel values of the block b.

  Ur = Rr−A × (Qr1 + Qr2 + Qr3)

  Here, the subtraction value Ur of the region r is estimated from the sum of the pixel values of adjacent blocks and the sum of the pixel values of the region r for the following reason. That is, in this embodiment, the sum of subtraction values in a block is determined to be a predetermined value, and in this embodiment, it is determined to be 0, and the sum of the pixel values of each of the three blocks is the image of each block. This is a pixel value based on information (a value excluding a subtraction value).

  In this case, only one adjacent block may be used, but in consideration of errors, three blocks are used in the present embodiment. When a plurality of blocks, for example, three blocks are used in this way, the total number of pixels in the three blocks is the total number of pixels in one region r because each block is divided into four regions. 12 times. Therefore, the total value Qr1 + Qr2 + Qr3 of the pixel values of the three blocks is multiplied by a predetermined value A as in the above equation. For example, in the above example, 1/12 (= 0.083). From many experiments, good results were obtained even if the predetermined value A was not necessarily determined uniquely from the relationship between the total number of pixels in the region r and the total number of pixels in adjacent blocks to be used. In the above example, good results were obtained even at 0.005.

  As described above, the subtracted value of the region r is estimated by subtracting the value obtained by multiplying the sum of the pixel values of the adjacent blocks (value excluding the subtracted value) by the predetermined value from the sum Rr of the pixel values of the region r can do.

  In step 214, it is determined whether or not the variable r is 4. If it is determined that the variable r is not 4, subtraction values have not been obtained for all the areas in the block b. Returning, the above processing (step 208 to step 214) is executed.

  On the other hand, when the variable r is equal to 4, since the subtraction values of all the areas of the block b are obtained, in step 216, the sum Ul of the subtraction values of the upper right area r1 and the subtraction values of the lower left area r3 are calculated. The sum of the sum U3 is I1, and in step 208, the sum of the subtraction values U2 in the upper left region r2 and the sum U4 of the subtraction values in the lower right region r4 is I2, and in step 220, Assuming that the absolute value of I1-I2 is H, it is determined in step 222 whether H is greater than or equal to a predetermined value T.

  Here, T is the minimum value (theoretical value) of the difference between the sums of the subtraction values in the two areas. If H is equal to or greater than T, it can be determined that there is no error, and the assigned code can be identified. On the other hand, if it is determined that H is less than the minimum difference value T, it can be determined that the determination is impossible.

  Therefore, if it is determined in step 222 that H is equal to or greater than T, the code can be identified. In step 224, it is determined whether or not I1 is greater than I2. If I1 is greater than I2, as shown in FIG. 2A, it can be determined that 1 is allocated as additional information to block B. Therefore, in step 226, the additional information is stored as 1, If it is not determined in step 224 that I1 is greater than I2, as shown in FIG. 2B, it can be determined that 0 is assigned to the additional information. Therefore, in step 228, the additional information is 0. Remember as.

  If it is determined that H is less than T, it is stored in step 230 that determination is impossible.

  In the next step 232, it is determined whether or not the variable b is equal to the total number B of blocks. If it is determined that the variable b is not equal to the total number B, there is a block whose additional information has not yet been identified. Returning to step 204, the above processing (step 204 to step 232) is executed. On the other hand, when the variable b is equal to B, since the additional information is identified for all the blocks, this process is terminated.

  Next, details of the operation in the additional information decoding unit 57 will be described. FIG. 17 is a flowchart showing an example of the operation of the additional information decoding unit 57. First, a macroblock search is performed in S141. Specifically, the block position detected by the block position detection unit 55 is identified by controlling the additional information identification unit 56 from the upper left direction, and a location matching the cue header (for example, “11111”) is detected. Since the additional information decoding unit 57 knows that the size of the macroblock is Y rows and X columns (for example, 16 rows and 8 columns), further, the cue header is located 8 blocks in the right direction from that point. If (for example, “11111”) exists, or if a cue header (for example, “11111”) exists 16 blocks below, the position of the first macroblock is determined. If the first macroblock position can be determined, other macroblock positions can be determined using the fact that the macroblocks are regularly arranged. Of course, even if the cue header includes an error, the position of the macroblock can be specified as long as most cue headers are not in error.

  Next, an encoding method is detected in S142. This can be detected by reading the coding method information of all the macroblocks and taking majority decision decoding.

  In S143, the information of all macroblocks is decoded according to a known logical format. In S144, the majority block decoding is performed between the decoded macroblock information having the same sequence number. In S145, when there is a sequence number other than “1”, additional information is connected and assembled in the order of the sequence number.

  As described above, in the present embodiment, when identifying data handled by a block, even if the image data of the identification block alone is erroneously identified, the block to be identified and the identification block are identified. Since the image data of the adjacent block is acquired, the increase / decrease value in the image data of the block to be identified can be inferred using the image data of the adjacent block, and erroneous identification of data handled by the block can be performed. Can be suppressed.

  Next, a modification of the present embodiment will be described. In this modification, the additional information identification process shown in FIG. 18 is executed instead of the additional information identification process shown in FIG. That is, in step 252, a variable b for identifying a block is initialized to 0, and in step 254, the variable b is incremented by 1.

  In step 256, as shown in FIG. 19, the sum Cb1 (C1 in FIG. 19) of the pixel values of the upper left block bLU adjacent to the block b identified by the variable b and the sum Cb5 of the pixel values of the lower right block bLD. (C5 in FIG. 19) is added to L, and in step 258, the sum Cb3 (C3 in FIG. 19) of the pixel values in the upper right block bRU and the sum Cb7 (in FIG. 19) of the pixel values in the lower left block bLD. The sum of C7) is M.

  In step 260, it is determined whether L is greater than M. If it is determined that L is greater than M, as shown in FIG. 19, the edge that is the boundary between the high density portion and the low density portion of the image positions the block diagonally, Since it can be determined that the upper left and lower right regions are high density portions, even if 1 is assigned to the block b, it is easily identified as 0. Therefore, in step 262 and step 264, processing that is easily identified as 0 is performed. That is, first, in step 262, the sum R1 of the pixel values in the upper right region is added to the predetermined value d to obtain a sum R1 of pixel values. In step 264, the sum R3 of the pixel values in the lower left region is set to the predetermined value d. The sum is defined as a sum R3 of pixel values. The predetermined value d is determined in advance.

  On the other hand, if it is not determined in step 260 that L is greater than M, the edge of the image positions the block diagonally, and the upper left, lower left, and lower right regions of the block b are high density portions. Since it is easy to identify 1 even if 0 is allocated, in step 266, the predetermined value d is added to the sum R2 of the pixel values in the upper left area so that it can be easily identified as 0. Is set to R2, and in step 268, a sum of pixel values R4 in the lower right region and a predetermined value d is added to R4.

  In step 270, the sum of R1 and R3 is set as P. In step 274, the sum of R2 and R4 is set as Q. In step 274, it is determined whether P is larger than Q. If it is determined that P is greater than Q, the additional information is stored as 1, and if it is not determined in step 274 that P is greater than Q, the additional information is stored as 0. In step 280, it is determined whether or not the variable b is equal to the total number B of blocks, and if it is determined that the variable b is not equal to the total number B, there is a block for which additional information has not been identified. Returning to 254, the above processing (steps 254 to 280) is executed. On the other hand, if the variable b is equal to the total number B, since the identification of the additional information has been executed for all the blocks, this process is terminated.

 In this modification, L is obtained by adding the sum Cb1 of the pixel values of the upper left block bLU adjacent to the block b identified by the variable b and the sum Cb5 of the pixel values of the lower right block bLD. The sum of the pixel values Cb3 of the block bRU and the sum Cb7 of the pixel values of the lower left block bLD is M, and it is determined whether L is greater than M. It is determined whether or not it is easily misidentified. However, the present invention is not limited to this and may be determined as follows.

 That is, for example, L is obtained by adding the sum Cb1 of the pixel values of the upper left block bLU adjacent to the block b identified by the variable b and the sum Cb5 of the pixel values of the lower right block bLD. Whether or not the assigned additional information is easily misidentified as 0 by determining whether L is greater than M, where M is the sum of the pixel values of all the blocks to be performed (sum of Cb1 to Cb8). You may make it judge.

It is a block diagram which shows one structural example on the side which embeds the additional information in the 1st Embodiment of this invention. It is explanatory drawing of an example of the pattern to embed. It is explanatory drawing of an example of the physical format in an example of the code format used by this invention. It is explanatory drawing of an example of the logical format in an example of the code format used by this invention. It is a flowchart which shows an example of operation | movement of the additional information encoding part 18 by the side of embedding the additional information in the 1st Embodiment of this invention. It is a block diagram which shows one structural example on the side which extracts additional information in the 1st Embodiment of this invention. It is explanatory drawing of an example of the inclination correction process of the image in an input image inclination correction | amendment part. 5 is a flowchart illustrating an example of the operation of a block size estimation unit 54. It is explanatory drawing of an example of the mask image for extracting the edge of a horizontal / vertical direction. 5 is a flowchart illustrating an example of the operation of a block position detection unit 55. It is explanatory drawing of an example of the mask image used with the block position detection part 55. FIG. It is explanatory drawing of an example of the process which calculates | requires a block position from an example of the extreme value image produced in the block position detection part 55, and an extreme value image. 5 is a flowchart illustrating an example of an operation of an additional information identification unit 56. It is explanatory drawing of an example of the calculation window used by the additional information identification part 56. FIG. It is the flowchart which showed the detail of identification of the additional information of step 133 of FIG. It is explanatory drawing explaining the principle of identification of additional information. 12 is a flowchart illustrating an example of an operation of an additional information decoding unit 57. It is the flowchart which showed the detail of identification of the additional information of step 133 of FIG. 13 in a modification. It is explanatory drawing explaining the principle of identification of the additional information in a modification.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Pattern size input part 12 Embedding intensity input part 13 Pattern attenuation rate input part 14 Pattern creation part 15 Image data input part 16 Image data storage part 17 Additional information input part 18 Additional information encoding part 19 Embedding position control part 20 Pattern selection part 21 Pattern superposition unit 22 Image output unit 31 Pattern size 32 Macro block size 33 Input image 41 Basic logic format 42 Cue header 43 Encoding method information 44 Sequence number 45 Effective code number information 46 Encoding information 51 Image data input unit 52 Image Data storage unit 53 Input image inclination correction unit 54 Block size estimation unit 55 Block position detection unit 56 Additional information identification unit 57 Additional information decoding unit 61 Pattern attenuation rate calculation unit 71 Embedding strength control unit 72 Minimum embedding strength input unit 73 Maximum embedding strength Input unit 8 Image data analysis unit 82 the image data correction unit

Claims (5)

The image area is divided into a plurality of blocks each composed of a plurality of pixels, each block is further divided into a plurality of areas, and the pixel value of each pixel of each block is determined in advance according to data handled by the block. The image formed on the image forming medium is read based on the image data configured by increasing / decreasing with the increase / decrease value assigned to each area in the block with the assigned pattern, and the data handled by the block is identified. A data identification device,
An acquisition block for acquiring an identification block that is a target for identifying the data and image data of an adjacent block adjacent to the identification block;
Identification means for identifying data handled by the identification block based on the image data of the identification block and the adjacent block acquired by the acquisition means;
A data identification device comprising: The increase / decrease value assigned to each area in the block is determined to be a predetermined value in the block,
The identification means includes
Calculating a sum of pixel values based on image data of at least one of the adjacent blocks;
Using the calculated sum of the pixel values of the adjacent blocks, estimate the sum of the increase / decrease values of each area of the identification block
Identifying the data based on a sum of increase / decrease values of each area of the estimated identification block;
The data identification device according to claim 1.   In the case where the image data of a plurality of adjacent blocks is used, the identifying means, when estimating the sum of the increase / decrease values of each area of the identification block, calculates the total sum of the pixel values based on the image data of the plurality of adjacent blocks, 3. The data identification device according to claim 1, wherein the data identification device is divided by a value equal to or greater than the total number of adjacent blocks used.   2. The identification unit, wherein the difference between the sum of the increase and decrease values of each area of the identification block is made unidentifiable when the difference is smaller than a minimum value determined based on a predetermined pattern. The data identification device according to any one of claims 3 to 4. Each block is responsible for different first data or second data,
The pattern has a first pattern determined according to the first data and a second pattern determined according to the second data,
The acquisition means acquires image data of a plurality of adjacent blocks,
The identification means includes
Determining whether or not the first data corresponding to the first pattern is easily identified from the second data corresponding to the second pattern based on the acquired image data of the plurality of adjacent blocks;
When it is determined that the first data is easily identified as the second data, the image data of the identification block is corrected so that the first data is easily identified,
Identifying the data based on the corrected image data;
The data identification device according to claim 1.

Images