JP5293357B2 - Image processing apparatus, authenticity determination apparatus, program, and system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To achieve: improvement in truth/false determination accuracy of a solid; and reduction in size of reference data for identifying a real solid. <P>SOLUTION: A truth/false determining device includes: an image capturing unit 31 for capturing an image of a printed matter 15 of a truth/false determination target; a code data restoring unit 35 for acquiring reference data from a QR code in the read image; a comparative data generating unit 33 for converting a comparative image extracted from the read image into pixel sequences, permuting the sequences at random thereafter, grouping the permuted sequences for each preset compression degree N, and generating comparative data for each group by calculating a total sum of pixel values included in the group; and a discrimination unit 36 for obtaining a correlation value between the reference data and the comparative data and determining the truth/false of the printed matter 15 from the correlation value. <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to an image processing device, an authenticity determination device, an image processing method, an authenticity determination method, and a solid.

  Since the paper is entangled with plant fibers having a thickness of 20 to 30 microns to create a random pattern, the surface of the paper is different from one sheet to another like a fingerprint. If this random pattern that differs for each paper is referred to as “paper fingerprint”, the paper fingerprint unique to each paper can be used for authenticating whether the paper such as a certificate is genuine. For example, image data of a paper fingerprint of a specific area on a real paper is encoded into a QR code and recorded on the paper. In this way, when it is determined whether or not a certain sheet of paper is genuine, a paper fingerprint of a specific area of the paper that has been brought in is photographed, and the data obtained from the captured data and the What is necessary is just to compare QR code of these. For example, a QR code of a real paper is recorded on a copy made by copying a real paper, but the paper fingerprint of the copy is different from that of the real paper. Thereby, it can be easily determined that the copy is not genuine.

  In this way, the authenticity of the paper can be determined using the paper fingerprint. For example, when the pixel value of one pixel of the paper is expressed by 1 byte length, the specific area on the paper from which the code data is obtained is displayed. If the size is 32 × 32 pixels, data of 32 × 32 = 1024 bytes can be obtained. However, since this data is additional data, the size is reduced when recording on paper. It is more convenient to do. For example, if JPEG compression technology is used, it is possible to compress it to about a quarter.

JP 2005-38389 A JP 2007-36317 A

  An object of the present invention is to improve the accuracy of determining the authenticity of a solid and to reduce the size of reference data for identifying a real solid.

The image processing apparatus according to the present invention is a whole or a part of solid image data having unique characteristics that can be read with randomness, and is two-dimensional with W × H pixels (W and H are both positive integers). An acquisition unit configured to acquire image data to be expressed; and a generation unit configured to generate reference data having identification of the solid from the acquired image data. The generation unit compresses the acquired image data. A first data compression unit , wherein the first data compression unit converts the acquired image data into a sequence of one-dimensional pixels having a length of W × H, and the converted one-dimensional pixel the sequence, rearranged randomly using a pseudo-random number, divided into groups of preset each compression degree N a sequence for changing parallel beauty, obtains the sum of the pixel values included in the group for each said group Turkey And features.
In addition, the generation unit includes a second data compression unit that further compresses the image data compressed by the first data compression unit, and the second data compression unit includes the first data compression unit. The reference data is generated by compressing the data values of the respective groups obtained by summing them according to a predetermined rule.

Authenticity determination apparatus according to the present invention, W × H pixels an image data of all or part of the solid body that have a readable unique features with random properties (W, H are both positive integers) a comparison image obtaining means for obtaining image data expressed as a comparison image authentication judgment in 2-dimensional, converts the comparison image acquired sequence of one-dimensional pixel length of W × H, is converted a sequence of one-dimensional pixel, rearranged randomly by using the same pseudo-random number as reference data serving as a reference for authenticity determination of the solid were used when it is created, a number sequences for changing parallel beauty, the reference data are divided into groups for each same compression degree N that was used when created, comparison data generating means for generating comparison data by obtaining the sum of the pixel values included in the group for each said group If the comparison with the reference data And having a correlation calculating means for calculating a correlation between the data, and the authenticity determination means for performing authenticity determination on the calculated phase function thus before Symbol solid body.

  The comparison image acquisition means acquires the comparison image while shifting the comparison image to be smaller than one pixel when sequentially acquiring the comparison image from the image data of the entire solid surface to be determined.

Further, when the authenticity determination unit performs the authenticity determination by comparing the correlation value calculated by the correlation calculation unit with a preset threshold value, the threshold value is obtained from the true solid image data. It is determined by performing a quantitative evaluation based on a correlation value obtained by comparing the number sequence and the number sequence obtained from solid image data to be determined.

Also, the second data compression unit, data values for each group, 0 if Kere smaller than a predetermined threshold, in accordance with a predetermined rule that is converted to 1 if the threshold value or more, the data values of each group Each is converted into 1-bit data.

  In addition, when the quartile of the data value of each group is Q1, Q2, Q3, the second data compression unit is 0 if the data value of each group is smaller than Q1, Q1 or more (Q1 + Q3). ) / 2, 1 if (Q1 + Q3) / 2 or more and 2 if less than Q3, and 3 if Q3 or more, the data value of each group is converted to 2-bit data. It is characterized by converting.

Program according to the present invention, computer, two-dimensional an image data of the whole or part of the solid with a readable unique features with randomness W × H pixels (W, H are both positive integers) in acquiring means for acquiring image data to be expressed, generating means for generating a reference data from the acquired image data with the identification of the solid, to function as the generating means, the compressed image data obtained 1 data compression unit , wherein the first data compression unit converts the acquired image data into a sequence of one-dimensional pixels having a length of W × H, and the sequence of converted one-dimensional pixels a rearranged randomly using a pseudo-random number, divided into groups of preset each compression degree N a sequence for changing parallel beauty, and Turkey obtains the sum of the pixel values included in the group for each said group It is characterized by.

Program according to the present invention, computer, solid body of all or part of the image data is a by W × H pixels having readable unique features with random properties (W, H are both positive integers) of Comparison image acquisition means for acquiring image data expressed in two dimensions as a comparison image for authenticity determination , converting the acquired comparison image into a sequence of one-dimensional pixels having a length of W × H , and converting the one-dimensional image sequence a, e replacement sequence randomly by using the same pseudo-random number as reference data serving as a reference for authenticity determination of the solid were used when it is created, the sequence rearranged, the reference data of the pixel There is divided into groups for each same compression degree N that was used when created, comparison data generating means for generating comparison data by obtaining the sum of the pixel values included in the group for each said group, said reference data Correlation calculating means for calculating the correlation between the comparison data, authenticity judging means for performing authenticity determination on the calculated phase function thus before Symbol solid body is intended to function as a.
The system according to the present invention includes an image processing device and an authenticity determination device, and the image processing device is a whole or a part of solid image data having a unique characteristic that can be read with randomness. Acquisition means for acquiring image data expressed in two dimensions of W × H pixels (W and H are both positive integers), and generation means for generating reference data having identification of the solid from the acquired image data; The generation means includes a first data compression unit that compresses the acquired image data, and the first data compression unit converts the acquired image data to a length of W × H. Converting into a sequence of one-dimensional pixels, rearranging the converted sequence of one-dimensional pixels randomly using a pseudo-random number, dividing the rearranged sequence into groups for each preset compression degree N, Pixels included in each group The sum of the values is obtained, and the authenticity determination device is a whole or a part of image data of a solid having a unique characteristic that can be read with randomness, and is W × H pixels (W and H are both positive integers). Comparison image acquisition means for acquiring image data expressed in two dimensions as a comparison image for authenticity determination, and the acquired comparison image is converted into a sequence of one-dimensional pixels having a length of W × H. A sequence of dimensional pixels is rearranged randomly by using the same pseudo-random numbers used when the reference data is created, and the rearranged number sequence is used when the reference data is created. A comparison data generating means for generating comparison data by dividing the same into groups for each compression degree N and calculating the sum of pixel values included in the group for each group; and the reference data and the comparison data Calculate correlation It comprises correlation calculating means for outputting, authenticity determining means for determining authenticity of the solid according to the calculated correlation, and output means for outputting the result of authenticity determination by the authenticity determining means.
The system according to the present invention includes an image processing device and an authenticity determination device, and the image processing device is a whole or a part of solid image data having a unique characteristic that can be read with randomness. Acquisition means for acquiring image data expressed in two dimensions of W × H pixels (W and H are both positive integers), and generation means for generating reference data having identification of the solid from the acquired image data; Means for putting the generated reference data into a data code or a magnetic stripe, and the generating means has a first data compression unit for compressing the acquired image data, and the first data The compression unit converts the acquired image data into a sequence of one-dimensional pixels having a length of W × H, rearranges the converted sequence of one-dimensional pixels randomly using pseudorandom numbers, and rearranges Preset sequence The group is divided into groups for each degree of compression N, and the sum of the pixel values included in the group is obtained for each group. Comparison image acquisition means for acquiring image data of a portion of the image and expressed in two dimensions of W × H pixels (W and H are both positive integers) as a comparison image for authenticity determination, and the reference data, Reference data acquisition means that acquires the data code or by decoding the magnetic stripe, and converts the acquired comparison image into a sequence of one-dimensional pixels having a length of W × H. The number sequence is randomly rearranged by using the same pseudo-random number used when the reference data was created, and the rearranged number sequence is the same as that used when the reference data was created A comparison data generating means for generating comparison data by dividing the data into groups for each degree of compression N and obtaining the sum of pixel values included in the group for each group, and calculating a correlation between the reference data and the comparison data It has correlation calculation means, and authenticity determination means for determining authenticity of the solid according to the calculated correlation.
The system according to the present invention includes an image processing device, an authenticity determination device, and a database unit, and the image processing device is a whole or a part of a solid having a unique readable characteristic having randomness. An acquisition means for acquiring image data that is expressed in two dimensions of W × H pixels (W and H are both positive integers), and generating reference data having the distinguishability of the solid from the acquired image data And generating means for associating the generated reference data with the fixed identification information and storing the reference data in the database means, wherein the generating means compresses the acquired image data. A data compression unit, wherein the first data compression unit converts the acquired image data into a sequence of one-dimensional pixels having a length of W × H, and converts the converted sequence of one-dimensional pixels, Using pseudo-random numbers Random rearrangement, the rearranged number sequence is divided into groups for each preset compression degree N, and the sum of the pixel values included in the group is obtained for each group, and the authenticity determination device reads random data Acquire, as a comparative image, image data expressed in two dimensions of W × H pixels (W and H are both positive integers), which are all or a part of the image data of a solid subject to authenticity determination having possible unique features. A comparison image acquisition means, a reference data acquisition means for acquiring the reference data by reading the reference data from the database means, and converting the acquired comparison image into a sequence of one-dimensional pixels having a length of W × H. The number sequence of the one-dimensional pixels is randomly rearranged by using the same pseudo-random number used when the reference data is created, and the reordered number sequence is A comparison data generating means for generating comparison data by dividing the groups into the same groups for each degree of compression N used at the time of creation, and calculating the sum of pixel values included in the groups for each group; It has correlation calculation means for calculating the correlation between the data and the comparison data, and authenticity determination means for determining the authenticity of the solid according to the calculated correlation.

  According to the present invention, the size of reference data for identifying a genuine solid can be further reduced.

  Further, even when the size of the reference data for identifying the real solid is further reduced, the accuracy of determining the authenticity of the solid can be improved.

  Further, by cutting out the comparison image from the image data on the surface of the solid while shifting it by less than one pixel, the accuracy of determining the authenticity of the solid can be further improved.

  Further, by determining the threshold value used for authenticity determination using Fisher's Z transformation and the cumulative distribution function of the standard normal distribution, the accuracy of determining the authenticity of a solid can be further improved.

1 is a block configuration diagram showing an embodiment of an image processing apparatus that implements an image processing method according to the present invention. FIG. 2 is a hardware configuration diagram of a computer forming an image processing apparatus according to the present embodiment. It is the block block diagram which showed one Embodiment of the authenticity determination apparatus which enforces the authenticity determination method which concerns on this invention. It is a hardware block diagram of the image processing apparatus which forms the authenticity determination apparatus in this Embodiment. It is the flowchart which showed the production | generation process of the reference data used for authenticity determination in this Embodiment. It is a figure used in order to demonstrate the procedure which produces | generates reference | standard data from a reference | standard image in this Embodiment. It is the flowchart which showed the authenticity determination method in this Embodiment. It is the figure which showed the graph of the result of having performed the determination with respect to a real thing, and the authenticity determination with respect to a fake 50 times each. It is the figure which showed the graph of the result of having performed the determination with respect to a real thing, and the determination with respect to a fake 50 times each about the case where a non-integer pixel shift is not used on a certain condition. In this Embodiment, it is the figure which showed the experiment result which discriminate | determined 50 times each with respect to a real thing and a fake using Fisher's z conversion and the cumulative distribution function of a standard normal distribution in a tabular form.

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

  FIG. 1 is a block diagram showing an embodiment of an image processing apparatus for carrying out an image processing method according to the present invention.

  The image processing apparatus 20 according to the present embodiment includes an image acquisition unit 21, a reference image extraction unit 22, a reference data generation unit 23, a coding unit 24, and a print processing unit 25. The image acquisition unit 21 acquires imaging data of the real printed material 13. The reference image extraction unit 22 extracts image data (hereinafter referred to as “reference image”) of a partial area that proves that the printed matter 13 is genuine from the imaged data of the printed matter 13. The reference data generation unit 23 generates reference data having identification of the printed matter 13 from the extracted partial image. The encoding unit 24 generates a data code such as a QR code by encoding the reference data. The print processing unit 25 prints the reference data encoded on the printed material 13.

  FIG. 2 is a hardware configuration diagram of a computer forming the image processing apparatus 20 according to the present embodiment. The computer forming the image processing apparatus 20 in the present embodiment can be realized with a general-purpose hardware configuration that has existed in the past. That is, as shown in FIG. 2, the computer is provided with a CPU 1, a ROM 2, a RAM 3, an HDD controller 5 connected to a hard disk drive (HDD) 4, a mouse 6 and a keyboard 7 provided as input means, and a display device. An input / output controller 9 for connecting a display 8 or a peripheral device (not shown) such as a printer and a network controller 10 provided as communication means are connected to an internal bus 11.

  Each component 21 to 25 in the image processing apparatus 20 is realized by a cooperative operation of a computer that forms the image processing apparatus 20 and a program that operates on the CPU 1 mounted on the computer.

  FIG. 3 is a block configuration diagram showing an embodiment of an authenticity determination device for executing an authenticity determination method according to the present invention. The authenticity determination device 30 according to the present exemplary embodiment includes an image acquisition unit 31, a comparative image extraction unit 32, a comparison data generation unit 33, a code image extraction unit 34, a code data restoration unit 35, a determination unit 36, and a result output unit 37. doing. The image acquisition unit 31 acquires a read image of the printed matter 15 that is a target for authenticity determination. The comparison image extraction unit 32 extracts image data (hereinafter referred to as “comparison image”) as original data to be compared with the reference data from the read image. The comparison data generation unit 33 generates comparison data to be compared with the reference data by performing a compression process described later on the comparison image. The code image extraction unit 34 extracts a code image printed on the printed matter 15 from the read image of the printed matter 15. The code data restoration unit 35 obtains reference data by restoring a code image. The discriminator 36 is provided as a correlation calculation means for obtaining a correlation between the reference data and the comparison data. The result output unit 37 outputs the determination result by the determination unit 36, that is, the determination result of whether the printed matter 15 to be authenticated is genuine or fake.

  FIG. 4 is a hardware configuration diagram of an image forming apparatus that implements the authenticity determination device 30 according to the present embodiment. The image forming apparatus is a multi-function machine having various functions such as a scanner function and a printing function, and is an apparatus having a built-in computer. In FIG. 4, the CPU 41 controls the operation of various mechanisms installed in the apparatus such as the scanner 44 and the printer engine 46 according to a program stored in the ROM 49. The address data bus 42 is connected to various mechanisms to be controlled by the CPU 41 and performs data communication. The operation panel 43 receives instructions from the user and displays information. The scanner 44 reads a printed material set by a user and accumulates it in an HDD (Hard Disk Drive) 45 or the like as electronic data. The HDD 45 stores an electronic document read using the scanner 44. The printer engine 46 prints an image on output paper in accordance with an instruction from a control program executed by the CPU 41. A network interface (I / F) 47 is connected to the network 17 for transmitting electronic data generated by the apparatus, receiving an e-mail transmitted to the apparatus, and accessing the apparatus via a browser. Used. The RAM 48 is used as a work memory during program execution and a communication buffer during electronic data transmission / reception. The ROM 49 stores various programs relating to control of the apparatus, encryption of electronic data, and transmission / reception of electronic data. By executing various programs, each component described later exhibits a predetermined processing function. An external media interface (I / F) 40 is an interface with an external memory device such as a USB memory or a flash memory.

  The components 31 to 37 in the authenticity determination device 30 are realized by a cooperative operation between a computer built in the image forming apparatus that forms the authenticity determination device 30 and a program that operates on the CPU 31 mounted on the computer. The authenticity determination device 30 may be configured by connecting a peripheral device such as a scanner to the computer shown in FIG. 2, or the image processing device 20 may be realized by the image forming device shown in FIG. Good.

  Further, the program used in this embodiment can be provided not only by communication means but also by storing it in a computer-readable recording medium such as a CD-ROM or DVD-ROM. The program provided from the communication means or the recording medium is installed in the computer, and various processes are realized by the CPU of the computer sequentially executing the installation program.

  Next, the operation in this embodiment will be described. First, reference data generation processing used for authenticity determination performed by the image processing apparatus 20 will be described with reference to the flowchart shown in FIG.

  The image acquisition unit 21 acquires imaging data of the printed matter 13 to be proved to be authentic (step 110). This imaging data is data obtained by optically reading a random pattern of paper fibers with a camera or the like, and may be data transmitted via a network (not shown) or connected to the image processing device 20. Data read by a scanner (not shown) may be used. Subsequently, the reference image extraction unit 22 extracts a partial image used for proving that the printed matter is genuine from the acquired imaging data (step 120). This partial image corresponds to a partial image (reference image) obtained by photographing a specific area of the sheet for use in authenticity determination. It is assumed that the size of the area for specifying the reference image and the position on the printed material 13 are determined in advance. Alternatively, the image acquisition unit 21 may be configured to acquire only a partial image corresponding to the reference image. In this case, the reference image extraction unit 22 is unnecessary, or the image acquisition unit 21 and the reference image extraction unit 22 are integrally formed. Since at least the size of the reference image is also used in the authenticity determination process described later, it is necessary to set this information in advance on the system so that the comparison image extraction unit 32 can use this information.

  Subsequently, the reference data generation unit 23 generates reference data as a reference for authenticity determination as follows based on the reference image extracted by the reference image extraction unit 22 (step 130).

  Assuming that the size of the reference image is W × H and the pixel value (luminance value) at the coordinates x and y of the reference image M is M [x, y], the reference data generation unit 23 first uses the function F based on the pseudorandom number. Create (x, y) = i. Here, if the degree of data compression is N (N is an integer), i is an integer and 0 ≦ i <W × H ÷ N, and F (x, y) = i regardless of the value of i. The function F is set so that the number of pairs of x and y that satisfy the above is constant. In the present embodiment, the degree of compression N is assumed to be about 2 to 32.

Subsequently, the reference data generation unit 23 calculates Ma [i] = Σ {M [based on the function F (x, y).
x, y] | F (x, y) = i} is obtained. In short, Ma [i] is obtained by shuffling the W × H pixels constituting the reference image M and then dividing them into N groups and taking the sum of the respective groups. As a result, the W × H reference image is compressed into W × H ÷ N data Ma [i]. In this embodiment, Ma [i] obtained by this calculation is further reduced to generate reference data Mb [i] that is expressed by 1 or 2 bits per element. When Mb [i] is expressed by 1 bit, 0 is set when the value of Ma [i] is smaller than the median value, and 1 is set otherwise. That is, Mb [i] takes a value of 0 or 1. When Mb [i] is expressed by 2 bits and Q1, Q2 and Q3 are quartiles, it is 0 if Ma [i] is smaller than Q1, 1 if it is greater than Q1 and smaller than (Q2 + Q3) / 2. , (Q2 + Q3) / 2 or more and 2 if smaller than Q3, and 3 otherwise. Thereby, Mb [i] takes any value of 0-3. As a result, the data Ma [i] having a total size of W × H bytes is further compressed into data Mb [i] having a total size of W × H ÷ N bits or a total size of 2 × W × H ÷ N bits. . In the present embodiment, the above-described median or quartile is used as the predetermined threshold. However, in addition to this, for example, the correlation coefficient of Ma [i] and Mb [i] is maximized. A method such as searching for such division points may be used. In the present embodiment, the threshold value is calculated and obtained at the time of execution, but a preset threshold value may be used.

  Here, the reference data generation processing performed by the reference data generation unit 23 will be specifically described with reference to FIG. Here, W = H = 32 and the compression degree N = 4.

  The 32 × 32 pixels constituting the reference image M are arranged in a line, and are rearranged at random, and then divided into groups each having a set value of the compression degree N. Since the pixel value is expressed in 1-byte length, each group is 4 bytes (= 32 bits) in length when divided into four. Then, Ma [i] (i = 0 to 255) is obtained by calculating the sum of each group, but Ma [i] can take a value of 0 to 1020 (= 255 × 4). [I] can be expressed in 10-bit length. That is, since the sum of the pixel values is obtained for each group, each of the 256 groups can be compressed from 32 bits to 10 bits at this point. In other words, the standard image 8192 bits (= 32 × 32 × 1 bytes) is reduced to 2560 (= 256 × 10 bits). In the present embodiment, the above-described data compression up to Ma [i] is referred to as “compression by addition”.

  Then, when reducing to 1 bit per element Ma [i], first, the median value of Ma [i] (i = 0 to 255) is obtained. Since the number of elements of Ma [i] is 256, the average of the 127th value and the 128th value from the largest value is the median value. This value is defined as MED. 0 is set in Mb [i] corresponding to Ma [i] smaller than MED, and 1 is set in other Mb [i]. Thus, in the present embodiment, Ma [i] is further reduced to Mb [i] expressed by 1 bit. In the present embodiment, data compression up to Mb [i] for reducing Ma [i] to 1 bit or 2 bits is referred to as “compression by level reduction”.

  As described above, when reducing to 1 bit per element Ma [i], the reference image having a size of 8192 bits is finally reduced to 1/32 × 256 bits = 256 bits. Further, when the reduction is made to 2 bits per element Ma [i], the reference image having a size of 8192 bits is finally reduced to 1/16 of 2 bits × 256 = 512 bits.

  When the reference data reduced as described above is calculated, the encoding unit 24 forms a data code by encoding the reference data (step 140). FIG. 1 shows an example in which a QR code is formed as an example of a data code. Then, the print processing unit 25 prints the QR code 14 on the printed matter 13 (step 150). For this purpose, the printed material 13 is set in advance in a printer (not shown), and a print request is transmitted so that the QR code 14 is printed on the printed material 13.

  Subsequently, the authenticity determination method in the present embodiment will be described with reference to the flowchart shown in FIG.

  When an image of the printed matter 15 is read by setting the brought-in printed matter 15 on the scanner 44 and performing a predetermined reading operation, the image acquisition unit 31 acquires a read image of the printed matter 15 (step 210). This printed matter 15 is a printed matter that is an object of authenticity determination because it is unknown whether it is genuine or fake. Of course, the image acquisition unit 31 may acquire a read image generated by being read using another device such as a network scanner instead of the scanner 44 via a network.

  Subsequently, the code image extraction unit 34 extracts the QR code 16 from the read image of the printed matter 15 (step 220). Since this extraction process can use an existing technique, a detailed description thereof will be omitted. The code data restoration unit 35 decodes the QR code 16 to obtain code data, that is, reference data Mb [i] (step 230).

  On the other hand, the comparison image extraction unit 32 extracts a partial image having the same area size as the reference image extracted in the reference image generation process from the read image (step 240). This partial image will be referred to as a “comparison image” in the following description. Although it may be necessary to perform image processing such as alignment of the read image of the printed matter 15 by the scanner 44, this processing is not the gist of the present embodiment, and thus description thereof is omitted.

  Subsequently, the comparison data generation unit 33 generates comparison data serving as a reference for authenticity determination based on the comparison image (step 250). Since the processing content to be performed by the comparison data generation unit 33 is basically the same as the reference data generation processing (step 130) in the reference data generation unit 23 that generates reference data from the reference image, the comparison data generation unit For the details of the comparison data generation process performed by 33, “reference image” in the description of the reference data generation process described above may be read as “comparison image”. Therefore, the comparison data generation process will be described while being omitted as appropriate.

  The size of the comparison image is of course the same W × H as that of the reference image. Here, the pixel value (luminance value) at the coordinates x and y of the comparison image S is S [x, y], and the upper left point cut out from the comparison image is (L, T). The initial values are 0 for both L and T.

The comparison data generation unit 33 first creates the same function F as that of the comparison data generation unit 23. That is, since the comparison data generation unit 33 shuffles using the same pseudorandom number used in the reference data generation process, the arrangement order of the pixels after the shuffle is the same as in the reference data generation process. Become. Then, the comparison data generation unit 33 obtains Sa [i] = Σ {S [x, y] | F (x, y) = i} based on the function F (x, y). In short, Sa
[I] is obtained by the same process as when Ma [i] is obtained in the reference data generation process. Sa [i] obtained by this processing is used as comparison data. In the authenticity determination process, only Sa [i] is obtained.

  Then, the determination unit 36 calculates a correlation value between the reference data Mb [i] and the comparison data Sa [i] obtained by the above processing (step 260), and stores the calculation result in the RAM 48 (step 270).

  When the correlation value is obtained, the comparison image extracting unit 32 shifts the comparison image extraction position by 0.25 pixels smaller than one pixel until the read image can be extracted from the entire read image (N in Step 270). Extracted from the read image (steps 280 and 240). Then, the generation of the comparison data, the calculation of the correlation value between the reference data Mb [i] and the comparison data Sa [i], and the storage are repeated. Note that when calculating an image shifted smaller than one pixel, an interpolation process is required. In this embodiment, a known interpolation process called bilinear interpolation is used.

  According to the correlation value obtained as described above, the determination unit 36 performs a quantitative evaluation of the correlation value (step 290). As a result, if the correlation value is a value equal to or greater than a predetermined threshold value (Y in step 300). ), It is determined that the reference image matches the comparison image. That is, it is possible to determine that the printed matter 15 to be authenticated is genuine. As a result, the result output unit 37 outputs the determination result on the operation panel 33 (step 310). On the other hand, when there is no correlation between the reference data Mb [i] and the comparison data Sa [i] (N in Step 300), the printed matter 15 to be determined as authenticity can be determined to be fake. As a result, the result output unit 37 outputs the determination result on the operation panel 33 (step 320).

  Here, experimental results are shown. Under certain conditions, prior art (5 × 5 pixels, 8 bits / pixel), compression by level reduction only (compression to 10 × 10 pixels, 8 bits / pixel, 2 bits / pixel) and compression by addition and compression by level reduction In the three conditions (compression of 30 × 30 pixels, 8 bits / pixel, compression degree 9, 2 bits / pixel), the determination for the real thing and the authenticity determination for the fake were performed 50 times each (total 300 times). The resulting graph is shown in FIG. All three conditions are for a standard image size of 200 bits. As shown in FIG. 8, it can be seen that when the compression by addition and the compression by level reduction are used together, the real article and the fake are well separated. For this reason, highly accurate authentication can be realized. In addition, the result of calculating the correlation is compared with a threshold value to determine whether it is genuine or fake. However, when compression by addition and compression by level reduction are used together, it is easy to set an optimal threshold value without error. .

  In this embodiment, the comparison image is extracted from the read image while being shifted by 0.25 pixels. Conventionally, the comparison image to be compared with the reference image is cut out from the printed matter 15 while shifting one pixel at a time. For this reason, there has been a problem that the correlation value does not become so high when the reference image and the comparison image are shifted by less than one pixel. This is due to the fact that no correlation value can be obtained for the wavelength components around one pixel.

  In order to solve this problem, in the present embodiment, when the comparison image is cut out, the upper left position to be cut out is shifted by 0.25 pixel as an example.

  By the way, when the pixel shift is a non-integer, interpolation is required. Although bilinear interpolation is used in the present embodiment, bicubic interpolation, Lanczos, or the like may be used. The value of 0.25 is not limited to this. As a result, a high correlation value can be obtained even when the reference image and the comparative image are shifted by 0.5 pixels or a value close thereto.

  When the non-integer pixel shift is used with less than 1 pixel under certain conditions, the determination for the real and the determination for the fake are performed 50 times each (total 200 times), as shown in the graph of FIG. Became. It can be seen that the non-integer pixel shift can be better separated from real and fake. As is clear from this result, although the processing time increases, a higher correlation value may be obtained by calculating the correlation by shifting the pixels by non-integer numbers such as 0.5 pixels or 0.25 pixels. As a result, the accuracy of authenticity determination can be improved. Also, as shown in FIG. 9, since the correlation value between the real product and the fake product is well separated, it is easy to determine which value should be set as the threshold value used for authenticity determination.

  By the way, as described in step 260 of FIG. 7, in order to determine whether or not the printed matter 15 is genuine, the correlation value itself or a value called a correlation score standard score (normalized score) is calculated. The calculated value is compared with a preset threshold value. Experiments were required to set this threshold value to an appropriate value. The threshold value depends on the camera, paper type, reference image size, and the like. Therefore, it is necessary to conduct an experiment for determining the threshold every time the camera, paper type, reference image size, and the like are changed, which is troublesome.

  Therefore, in the present embodiment, by performing quantitative evaluation of the correlation value (step 290), it is possible to present an appropriate threshold value or a standard value for finding it without performing an experiment although it is an approximation. I made it. Specifically, it is possible to calculate the probability P that the fake comparative image brings such an observation result to the correlation value (evaluation result) calculated in step 260 of FIG. As a result, fine adjustment for increasing the accuracy of authenticity determination is facilitated. A method for calculating the probability P is as follows.

First, a correlation value obtained by comparing a reference image with a comparative image cut out from an image of the printed matter 15 is {Ci}, and the length of {Ci} is N. When the reference image is 32 × 32 pixels and the pixels are shifted by N = 0.25, the number of non-integer pixels is included by 3 between the integer pixels, so 32+ (32−1) × 3 = 125. That is, i is 0 to 125. First,
Xi = log ((1 + Ci) / (1-Ci))
Calculate This calculation is called “Fischer's Z conversion”. By performing this calculation, the correlation coefficient between random data can be converted into a normal distribution. Subsequently, an average E and a standard deviation s of {Xi} are obtained. And
Zmax = (max {Xi} −E) / s
Ask for. Standardization is performed by this calculation. And
P = 1− (Φ (Zmax)) ^N
As a result, the probability P to be obtained is obtained. Here, Φ (x) is a cumulative distribution function of a standard normal distribution.

By using this value, it is possible to adjust parameters in a short time. Also, according to experiments, the performance of discrimination is better than that of the conventional standard. FIG. 10 shows a table showing the results of the experiment in which the discrimination was performed 50 times for the real product and 50 times for the fake product. It should be noted that the average value and standard deviation of the values obtained by the experiment to be judged as genuine are μ ₀ and σ ₀ , and the average value and the standard deviation of the values obtained by the experiment to be judged as fake are μ ₁ and σ _{When 1} was set, | μ ₀ −μ ₁ | / (σ ₀ + σ ₁ ) was defined as “distance between genuine distribution and fake distribution”.

  In the example of this experiment, the real and the fake are well separated regardless of which index is used, but as shown in FIG. 10, the probability P that the fake comparative image brings such an observation result is excellent. I understand that.

  As described above, in the past, the threshold was determined by accumulating many experiments, but in this embodiment, by using the Fisher's Z transformation and the cumulative distribution function of the normal distribution, there are many more than the conventional one. The correct threshold can be obtained without conducting an experiment.

  As described above, in the present embodiment, since the reference data for proving the authenticity can be reduced from the reference image at a high compression rate, the reference data is converted into a QR code or the like. It is possible to enter not only the data code but also the magnetic stripe.

  In the present embodiment, the reference data is QR-coded and added to the printed matter 13. However, the reference data may be associated with the identification information of the printed matter 13 and stored in a database for management. In this case, instead of the code image extraction unit 34 and the code data restoration unit 35, a means for reading the reference data from the database is provided.

  Further, in the present embodiment, the printed matter that is a paper medium has been described as an example of a solid having a unique characteristic that can be read with randomness. A paper fingerprint on a part of the surface of the paper medium was used as a characteristic unique to the paper medium. However, the paper fingerprint existing inside the paper medium, not the surface of the paper medium, may be read out. Further, the present embodiment can be applied not only to a paper medium as long as the solid has characteristics unique to the solid. Further, unique features may be read from the whole rather than from a portion of the solid.

  1 CPU, 2,49 ROM, 3,48 RAM, 4,45 Hard disk drive (HDD), 5 HDD controller, 6 Mouse, 7 Keyboard, 8 Display, 9 Input / output controller, 10 Network controller, 11 Internal bus, 17 Network , 20 image processing device, 21 image acquisition unit, 22 reference image extraction unit, 23 reference data generation unit, 24 encoding unit, 25 print processing unit, 30 authenticity determination device, 31 image acquisition unit, 32 comparison image extraction unit, 33 Comparison data generation unit, 34 code image extraction unit, 35 code data restoration unit, 36 determination unit, 37 result output unit, 41 CPU, 42 address data bus, 43 operation panel, 44 scanner, 46 printer engine, 47 network interface (I / F), 50 Part media interface (I / F).

Claims (12)

Acquisition of the whole or a part of solid image data having unique characteristics that can be read with randomness and expressed in two dimensions of W × H pixels (W and H are both positive integers). Means,
Generating means for generating reference data having identification of the solid from the acquired image data;
Have
The generation unit includes a first data compression unit that compresses the acquired image data,
The first data compression unit includes:
The acquired image data is converted into a sequence of one-dimensional pixels having a length of W × H ,
The converted sequence of one-dimensional pixels is randomly rearranged using pseudo-random numbers ,
Divided into groups of each compression degree N that is preset parallel beauty changed sequence of numbers,
Ru obtains the sum of the pixel values included in the group for each said group,
An image processing apparatus.   The image processing apparatus according to claim 1.
  The generation unit includes a second data compression unit that further compresses the image data compressed by the first data compression unit,
  The second data compression unit generates reference data by compressing each group of data values obtained by the summation by the first data compression unit according to a predetermined rule.
  An image processing apparatus. The image data of all or part of the solid body that have a readable unique features with random properties W × H pixels image data (W, H are both positive integers) is represented by a two-dimensional A comparison image acquisition means for acquiring the image as a comparison image for authenticity determination ;
When the acquired comparison image is converted into a sequence of one-dimensional pixels having a length of W × H, and the converted sequence of one-dimensional pixels is used as reference data that is used as a reference for determining the authenticity of the solid. rearranged randomly by using the same pseudo-random number that was used to divide a sequence for changing parallel beauty, a group of each same compression degree N that was used when the reference data is created, a comparison data generating means for generating comparison data by obtaining the sum of the pixel values included in the group for each said group,
Correlation calculating means for calculating a correlation between the reference data and the comparison data,
And authenticity determination means for performing the calculated phase function thus determining the authenticity of the previous SL solid body,
An authenticity determination device characterized by comprising: In the authenticity determination device according to claim 3 ,
The authenticity determination device, wherein the comparison image acquisition unit acquires the comparison image while shifting the comparison image smaller than one pixel when sequentially acquiring the comparison images from the image data of the entire solid surface to be determined. In the authenticity determination device according to claim 3 ,
When the authenticity determination means performs authenticity determination by comparing the correlation value calculated by the correlation calculation means with a preset threshold value, the threshold value is a sequence obtained from the true solid image data. An authenticity determination device characterized by being determined by performing a quantitative evaluation based on a correlation value obtained by comparing a number sequence obtained from solid image data to be determined. The image processing apparatus according to claim 2 .
The second data compression unit, data values for each group, 0 if Kere smaller than a predetermined threshold, in accordance with a predetermined rule that is converted to 1 if the threshold value or more, the data values of each group, respectively 1 An image processing apparatus which converts data into bit length data. The image processing apparatus according to claim 2 .
When the quartiles of the data values of each group are Q1, Q2, and Q3, the second data compression unit is 0 if the data value of each group is smaller than Q1, Q1 or more, and (Q1 + Q3) / Each group data value is converted into 2-bit length data according to a predetermined rule of 1 if less than 2, 1 if (Q1 + Q3) / 2 or more and 2 if less than Q3, or 3 if Q3 or more. An image processing apparatus. The computer,
Acquisition of the whole or a part of solid image data having unique characteristics that can be read with randomness and expressed in two dimensions of W × H pixels (W and H are both positive integers). Means ,
Generating means for generating reference data having identification of the solid from the acquired image data;
Function as
The generation unit includes a first data compression unit that compresses the acquired image data,
The first data compression unit includes:
The acquired image data is converted into a sequence of one-dimensional pixels having a length of W × H ,
The converted sequence of one-dimensional pixels is randomly rearranged using pseudo-random numbers ,
Divided into groups of each compression degree N that is preset parallel beauty changed sequence of numbers,
Ru obtains the sum of the pixel values included in the group for each said group,
A program characterized by that. The computer,
The image data of all or part of the solid body that have a readable unique features with random properties W × H pixels image data (W, H are both positive integers) is represented by a two-dimensional A comparison image acquisition means for acquiring a comparison image for authenticity determination ,
When the acquired comparison image is converted into a sequence of one-dimensional pixels having a length of W × H, and the converted sequence of one-dimensional pixels is used as reference data that is used as a reference for determining the authenticity of the solid. rearranged randomly by using the same pseudo-random number that was used to divide a sequence for changing parallel beauty, a group of each same compression degree N that was used when the reference data is created, comparison data generating means for generating comparison data by obtaining the sum of the pixel values included in the group for each said group,
Correlation calculating means for calculating a correlation between the reference data and the comparison data,
Authenticity judging means for performing authenticity determination of the calculated phase function to thus before Symbol solid body,
Program to function as .   An image processing device;
  An authenticity determination device;
  Have
  The image processing apparatus includes:
  Acquisition of the whole or a part of solid image data having unique characteristics that can be read with randomness and expressed in two dimensions of W × H pixels (W and H are both positive integers). Means,
  Generating means for generating reference data having identification of the solid from the acquired image data;
  Have
  The generation unit includes a first data compression unit that compresses the acquired image data,
  The first data compression unit includes:
  The acquired image data is converted into a sequence of one-dimensional pixels having a length of W × H,
  The converted sequence of one-dimensional pixels is randomly rearranged using pseudo-random numbers,
  The rearranged number sequence is divided into groups for each preset compression degree N,
  For each group, find the sum of the pixel values included in the group,
  The authenticity determination device includes:
  The whole or a part of solid image data having unique characteristics that can be read with randomness, and image data expressed in two dimensions of W × H pixels (W and H are both positive integers) Comparison image acquisition means for acquiring a comparison image;
  The acquired comparison image is converted into a sequence of one-dimensional pixels having a length of W × H, and the converted sequence of one-dimensional pixels is the same as that used when the reference data is created. The random number is rearranged by using a random number, and the rearranged number sequence is divided into groups each having the same degree of compression N that was used when the reference data was created, and pixels included in the group for each group Comparison data generating means for generating comparison data by calculating the sum of values;
  Correlation calculating means for calculating a correlation between the reference data and the comparison data;
  Authenticity determination means for determining the authenticity of the solid according to the calculated correlation;
  Output means for outputting the result of the authenticity determination by the authenticity determining means;
  The system characterized by having.   An image processing device;
  An authenticity determination device;
  Have
  The image processing apparatus includes:
  Acquisition of the whole or a part of solid image data having unique characteristics that can be read with randomness and expressed in two dimensions of W × H pixels (W and H are both positive integers). Means,
  Generating means for generating reference data having identification of the solid from the acquired image data;
  Means for placing the generated reference data in a data code or magnetic stripe;
  Have
  The generation unit includes a first data compression unit that compresses the acquired image data,
  The first data compression unit includes:
  The acquired image data is converted into a sequence of one-dimensional pixels having a length of W × H,
  The converted sequence of one-dimensional pixels is randomly rearranged using pseudo-random numbers,
  The rearranged number sequence is divided into groups for each preset compression degree N,
  For each group, find the sum of the pixel values included in the group,
  The authenticity determination device includes:
  The whole or a part of solid image data having unique characteristics that can be read with randomness, and image data expressed in two dimensions of W × H pixels (W and H are both positive integers) Comparison image acquisition means for acquiring a comparison image;
  Reference data acquisition means for acquiring the reference data by decoding the data code or the magnetic stripe;
  The acquired comparison image is converted into a sequence of one-dimensional pixels having a length of W × H, and the converted sequence of one-dimensional pixels is the same as that used when the reference data is created. The random number is rearranged by using a random number, and the rearranged number sequence is divided into groups each having the same degree of compression N that was used when the reference data was created, and pixels included in the group for each group Comparison data generating means for generating comparison data by calculating the sum of values;
  Correlation calculating means for calculating a correlation between the reference data and the comparison data;
  Authenticity determination means for determining the authenticity of the solid according to the calculated correlation;
  The system characterized by having.   An image processing device;
  An authenticity determination device;
  Database means;
  Have
  The image processing apparatus includes:
  Acquisition of the whole or a part of solid image data having unique characteristics that can be read with randomness and expressed in two dimensions of W × H pixels (W and H are both positive integers). Means,
  Generating means for generating reference data having identification of the solid from the acquired image data;
  Means for storing the generated reference data in the database means in association with the fixed identification information;
  Have
  The generation unit includes a first data compression unit that compresses the acquired image data,
  The first data compression unit includes:
  The acquired image data is converted into a sequence of one-dimensional pixels having a length of W × H,
  The converted sequence of one-dimensional pixels is randomly rearranged using pseudo-random numbers,
  The rearranged number sequence is divided into groups for each preset compression degree N,
  For each group, find the sum of the pixel values included in the group,
  The authenticity determination device includes:
  The whole or a part of the image data of the solid subject to authenticity determination having unique characteristics that can be read having randomness, and expressed in two dimensions of W × H pixels (W and H are both positive integers). A comparison image acquisition means for acquiring the image as a comparison image;
  Reference data acquisition means for acquiring the reference data by reading from the database means;
  The acquired comparison image is converted into a sequence of one-dimensional pixels having a length of W × H, and the converted sequence of one-dimensional pixels is the same as that used when the reference data is created. The random number is rearranged by using a random number, and the rearranged number sequence is divided into groups each having the same degree of compression N that was used when the reference data was created, and pixels included in the group for each group Comparison data generating means for generating comparison data by calculating the sum of values;
  Correlation calculating means for calculating a correlation between the reference data and the comparison data;
  Authenticity determination means for determining the authenticity of the solid according to the calculated correlation;
  The system characterized by having.

Images