WO2015022872A1

WO2015022872A1 - Method for automatically determining authenticity of individual article using micro contour shape of printing as identifier

Info

Publication number: WO2015022872A1
Application number: PCT/JP2014/070503
Authority: WO
Inventors: 平山貞宏; 三輪忠仁; 尾形廣秋
Original assignee: 株式会社偽物識別技術研究所
Priority date: 2013-08-11
Filing date: 2014-08-04
Publication date: 2015-02-19

Abstract

[Problem] There is no method which is for automatically determining the authenticity of an article and which can be executed by an average person using a camera-equipped cell phone. [Solution] The color of an ink (for which the difference between an attribute such as the color luminance and that of a background color is equal to or greater than a fixed value) is specified, and a fine unique shape is printed in that color as an identification mark. A computer is used to binarize a digital image of the identification mark automatically using, as a threshold value, a value which is near the central value of the luminance (or the like) corresponding to two peaks appearing in a histogram of the luminance (or the like) of that digital image, and a micro contour shape which is equal to or less than the shape control limit of that identification mark is used as an identifier. The identifier of an identification mark to be compared is compared with the identifier of a stored legitimate identification mark, and the degree of matching therebetween is represented with a unique statistical index calculated automatically by a computer, thereby enabling the authenticity of the identification mark and an individual article to be determined with high precision by an average person using a camera-equipped cell phone.

Description

Automatic authenticity judgment method for individual articles using printed micro contour as identifier

The present invention relates to an automatic authenticity determination method or automatic recognition method for individual articles that can be easily executed by an ordinary imaging device by an ordinary person using a micro-contour shape of a print that cannot be copied because a person cannot control the shape as an identifier. .

A micro-identification mark is printed on an article or its certificate, and a micro-contour shape of printing that cannot be copied because a person cannot control the shape is used as an identifier. The identifier and the article identification code (manufacturing number) , Serial number, etc.) and storing them in a storage device, when the goods are shipped together with the identification mark, and then it is desired to determine the authenticity of the shipped goods, the article identification code and the identification mark Is read by a general daily imaging device (such as a camera-equipped mobile phone) and transmitted to the authenticity judging person via the Internet, and the authenticity judging person compares the received article identification code and the identification mark with those stored. An authenticity determination system configured to determine the authenticity of an article according to the degree of coincidence and return the result is disclosed or patent-patented.

The above patent application is registered as Japanese Patent No. 4775727. Japanese Patent Application No. 2011-103830 The above patent application is a divisional application of Patent Document 1. PCT / JP2011 / 065915 The above patent application was transferred to Japan and registered as Japanese Patent No. 5071592. Japanese Patent Application No. 2011-282547

The authenticity determination methods of Patent Documents 1 to 4 in which a minute identification mark is printed and only the shape of a micro contour line below the shape control limit is used as an identifier does not describe an automatic authentication method. In order to instantly determine the authenticity of the huge number of identification marks required by the general person's authenticity requester and send back the results, a method using the naked eye of the human being makes a huge number of determination personnel at the authentication center. Must be prepared. Collation and judgment with the naked eye of the enlarged image until the shape characteristics of the microscopic outline can be read sufficiently is inherently very accurate, but humans sometimes make errors, and in the judgment with the naked eye, a quantitative index for judgment should be provided Is also impossible. Therefore, it is highly desirable to calculate a quantitative index of the degree of coincidence by a computer and to automate the judgment based on it.

However, as for the authenticity determination method using the micro-contour shape of the printed identification mark as an identifier, no automated method has been proven or disclosed yet. Moreover, there is no disclosure of an automatic authentication method using the identifier that can be executed by ordinary people with daily popular devices. There are many problems that must be solved technically in order to automate the authentication. That is, the work of imaging the identification mark and storing the microscopic contour shape in the database as a master image is performed by a specialist, so the imaging equipment and imaging environment (quality of irradiation light, irradiation angle, position in the image of the identification mark, size) , Resolution) or imaging technology (good or bad focus) (hereinafter, the shooting environment and shooting technology are collectively referred to as shooting conditions) can be made stable with high quality, but the general public is popular. When imaging the identification mark with an imaging device (for example, a camera-equipped mobile phone), the quality of the imaging device is low, the quality of the captured image varies depending on the device, the imaging environment varies, and the imaging technology is also general. Inferior. In order to capture microscopic contours, images are captured by optical and electronic enlargement, so the image on the display screen is constantly vibrated greatly due to camera shake, and it is difficult for ordinary people to focus accurately. It is.

Therefore, the image quality of the identification mark imaged by a general person and transmitted to the authenticity judgment person is low, the imaging light quality, the irradiation angle, the brightness, the image position and the size are various, and the focus is often not clear. That is, the image quality is low because the quality of the imaging conditions is low and not stable. These must be automatically checked against the stored master image to determine whether or not they match. That is not technically easy. The characteristics of the captured image will be described below.

(1) Individual characteristics of printed matter.
Please refer to FIG. As an example of a micro printed matter, there are three printed matter by an ink jet printer having a length of about 1 mm. The printed microscopic contour shape, which is the identifier of the present application, can be visually observed that even if the printed material has exactly the same shape, the printed material has different individual characteristics as long as the printed material is different. These microscopic contours have randomness that cannot be controlled by humans. If the printing of the identification mark is viewed microscopically, random irregularities, enclaves (small islands), blurring, etc. are unavoidable at the boundary of the ink region. In addition, as shown in FIG. 6, the brightness intensity changes continuously in each boundary region.

(2) Change in captured image depending on the imaging model and imaging conditions.
Please refer to FIG. 8, FIG. 9, and FIG. FIGS. 9 and 10 show images obtained by picking up the upper part of the printed matter (vertical and horizontal length of about 1 mm) in FIG. 8 with various camera-equipped mobile phones.
Even if the same individual printed matter is imaged, the contour shape differs little by little depending on the model of the camera-equipped mobile phone, and there are various image quality, both good and bad. Further, the brightness, color, shading, size, position, and imaging angle of the identification mark on the screen are not exactly the same.

(3) The contour area is gradation. Changes due to the binarization threshold of the contour shape.
Further, when the identification mark is viewed microscopically in the boundary region, the color (for example, luminance) is gradation, so that the specific boundary position cannot be determined and the contour shape cannot be determined. In other words, in the microscopic contour region of the image obtained by capturing the printed identification mark, the color (for example, luminance) continuously and gently changes for each light receiving cell. Therefore, the shape of the contour varies greatly depending on which color value (for example, luminance value) is used as the threshold value to define the contour. Accordingly, the microscopic contour shape cannot be uniquely determined. A part of the identification mark of FIG. 4 is enlarged and shown in FIG. It can be seen that the brightness of the contour region does not change in a tomographic manner but changes while gradually gradation.
FIG. 7 shows a histogram of the brightness of the image and the number of light receiving cells in FIG. 6, and changes in the threshold value of luminance and changes in contour shape when binarized. The left contour diagram is, for example, a luminance value 110 as a threshold, and the right contour diagram is a contour when, for example, 155 is taken. It shows that the contour shape changes significantly if the threshold value changes.

As described above, by comparing the image with unstable image quality captured by a general person with a general imaging device and the image with good image quality stored in the storage device, and by taking the binarization threshold value for producing the contour shape It is not easy to collate images with greatly changing contour shapes and automatically determine whether the contour shapes match or not with high accuracy by a computer. All the above-mentioned problems must be solved in order to automate the collation determination by a computer. In order to solve these problems, the present application has invented a solution using the following series of steps.

Table of contents Overview of automatic authentication method for solving problems.
2. Automatic generation method of binarized micro image.
3. Automatic collation method of binarized micro image and automatic judgment method of coincidence / mismatch.

1. An overview of a computer-based automatic authentication method for solving problems.
The problem is solved by a series of steps (1) to (9) described below.

(1) A step of printing the identification mark in a color that is clearly different from the background color.
Please refer to FIG. 12 (printing of an identification mark on an article certificate). A fine mark having an arbitrary shape is printed on the target article or its certificate in a color that is clearly different from the background color, and this is used as an identification mark. If the color of the mark is not clearly different from the background color, the color change in the microscopic outline of the identification mark of the printed matter is too gradual. Therefore, as described in paragraph 0005 above, the imaging conditions (imaging environment and imaging technology) ) And, as described in paragraph 0009 above, the microscopic contour shape as an identifier changes greatly depending on how to set the threshold for binarization of the image, and automatic determination with high accuracy cannot be performed.
However, even if the color is clearly different from the background color, the identification mark is not clearly different from the background color on the microscopic image obtained by capturing it as described above with reference to FIG. Accordingly, it is necessary to quantitatively define what is “the color of the identification mark is clearly different from the background color”, which is the core of the present application, rather than sensory. The definition is defined by the image of the identification mark taken as follows.
Note that the background color may be printed in a dark color and the identification mark may be a color having a large color difference from that in the printing. As an example, the identification mark may be a blank area of ink having an arbitrary shape in a dark color.

Quantitative definition of colors that are distinctly different from the background color.
An example will be described with reference to FIG. 11 (upper half of horizontal No. 1 printing, histogram, definition of clear color difference). FIG. 11 shows the original image of the upper half part of FIG. 4 (horizontal No. 1 printing, length of about 1 mm), a histogram of its luminance value and the number of pixels, and a binarized contour diagram with a specific threshold value. Yes. If the luminance value expressed in RGB format is used for the color information of the image, the value can be expressed with a minimum of 1 and a maximum of 255. In the histogram of the luminance value and the number of pixels, if the identification mark is printed in a color that is clearly different from the background color, there will always be two peaks and a low pan bottom between them. The bright background color is the right peak, the dark identification mark color is the left peak, and the pan bottom part is the gradation part of the boundary area between the background and the identification mark. The degree of clearness from the background color is greater as the difference in luminance values corresponding to the peaks of the two peaks is larger and the width of the bottom of the pan is longer and lower. Therefore, the degree of definition can be quantitatively defined by the distance between the two peaks and the width of the bottom. For example, in the histogram of the brightness value and the number of pixels of the identification mark to be printed, the difference between the brightness values corresponding to the peaks of two peaks is 60 or more, and the bottom of the pan The difference between the luminance values at the two ends (the luminance value immediately before and immediately after the absolute value of the luminance gradient suddenly changes greatly) should be 30 or more. In order to make it more precise, the height of the pan bottom is limited to be low. The identification mark is printed in such a color.

By the identification mark printed as described above, as described later in paragraph number 0028,
(a) Regardless of imaging model
(b) Regardless of the imaging conditions (imaging environment and imaging technology)
(c) As long as the threshold for binarization is near the center of the luminance value corresponding to the peaks of the two peaks,
I discovered that I can make high-precision authentication. As long as the threshold when binarizing to obtain a microscopic contour shape is near the center of the luminance value corresponding to the peak of two peaks, important contour features are retained no matter where it is taken. Therefore, the determination accuracy does not drop. These three tolerances (a), (b), and (c) are characteristics brought about by setting the color difference between the background color and the identification mark as described in paragraph 0014. It was discovered that the identification mark has the characteristic. This characteristic is a crucial requirement for authenticity determination with an unstable image captured by a general person as in the present application.

Note that when the identification mark is imaged and converted into electronic color information, the color information used for the histogram reference is not limited to the RGB luminance value. Color attributes such as saturation and lightness may be used as color information. For each color information, a histogram similar to the histogram of the luminance value and the number of pixels described above is obtained, and in the histogram, the length and height of the width of the bottom and the width of the bottom are defined, and the “clearness with respect to the background color” is defined. “Different colors” may be defined. The specified values to be set depend on the determination accuracy required for coincidence / mismatch.

(2) A step of imaging the regular identification mark.
See FIG. 3 (capturing and storing an image of an identification mark on a certificate). The identification mark printed in the step (1) is imaged by an imaging device capable of recognizing the features of the micro contour shape, and a primitive electronic image is acquired.

(3) A step of automatically binarizing the identification mark and storing / registering it.
See FIG. 13 (Regular Identification Information Automatic Registration). The original electronic image captured in the step (2) is automatically binarized by “2. Binarized image automatic generation method” described later in paragraph No. 0026 to obtain a “binarized micro image”.
Next, both the original electronic image and the binarized micro image, or the latter, are stored and registered in the electronic storage device as regular identification information of the identification mark.

(4) A step of inputting an individual article identification code.
See FIG. 3 (capturing and storing an image of an identification mark on a certificate). The individual article identification code (article serial number, manufacturing number, etc.) of the target article is stored in the electronic storage device in association with the regular identification information stored in the electronic storage device in step (3).

(5) Step of inputting the image of the identification mark to be verified and the individual article identification code to be verified.
See Fig. 1 (Conceptual diagram of automatic authentication system). An ordinary person who wants to determine the authenticity of an article takes an image with a terminal device with camera capable of recognizing the feature of the microscopic contour shape of the identification mark to be verified, or acquires a reading original electronic image with an electronic reading device such as a scanner. Further, the individual article identification code to be verified corresponding to the identification mark to be verified is read or inputted by the terminal device.

(6) Transmitting step.
See FIG. 1 (conceptual diagram of the automatic authentication system). The general person side transmits the collated individual article identification code and the original electronic image of the collation identification mark input in step (5) to the authenticity determination person side.

(7) A receiving step.
The authenticity determination side receives the transmission information of step (6) above.

(8) Automatic collation / determination step.
See Fig. 1 (Conceptual diagram of No. 11 automatic authentication system). The authenticity judge side automatically binarizes the original electronic image of the verification identification mark received in the step (7) by paragraph number 0026 according to “2. Automatic generation method of binarized image” described later. A micro-image ". Then, it automatically collates with the binarized micro image of the regular identification mark stored in the electronic storage device corresponding to the received individual article identification code to be verified, and determines the coincidence or inconsistency. A method of automatically collating a binarized micro image and a method of automatically determining coincidence / mismatch by a computer will be described in “3. Automatic method of collating binarized micro image and method of automatically determining coincidence / mismatch” described later in paragraph 0030. . See FIG. 14 (Automatic image collation / judgment method).

Alternatively, the general person side transmits only the verified individual article identification code input in step (5) to the authenticity determination person side, and the corresponding regular identification mark stored in the electronic storage device in step (3) corresponding thereto. As shown in paragraph number 0028 below, the computer automatically generates a binarized micro image from the original electronic image of the verification identification mark obtained in step (5) above. As described in paragraph 0030, both binarized micro images are automatically collated, and the coincidence / mismatch is automatically judged.
In other words, the automatic generation of the above-described collated binarized micro images and the automatic collation and automatic determination of coincidence / non-coincidence may be performed on the authenticity determination center side, or may be performed on the general person side with software obtained in advance. That is to say.

(9) A step of notifying the result.
See FIG. 1 (conceptual diagram of the automatic authentication system). The authenticity of the target article is determined based on the result of determination of coincidence / non-coincidence in step (8), and is returned or displayed.

2. A method for automatically generating a binarized micro image by a computer.
Please refer to the figure below.
Fig. 15 Threshold values at both ends of the bottom of the histogram and binarized micro image,
Fig. 16 Low, middle and high threshold values near the center of the bottom of the histogram and a binarized micro image,
Fig. 17 Minimum area that does not overlap Same printing and different printing.

With respect to the luminance values of all the pixels of the original electronic image of the identification mark, all the pixels are binarized by automatically setting a threshold value to obtain a binarized micro image.
An automatic threshold setting method for binarization will be described.
The identification mark is printed in a color that is clearly different from the background color. If the background area is large enough, the original electronic image will not show anything other than the identification mark and the background. Alternatively, if the background area is not sufficiently large and the identification mark and the object other than the background are captured, only the identification mark is captured through the process of extracting only the identification mark from the identification mark template. The histogram of the luminance and the number of pixels of the image always has two and only two peaks as described in paragraph 0014 above. The valley between the two peaks is a flat pan bottom with a long width and a very low height.

In FIG. 15, FIG. 16, and FIG. 17, only the upper half of the horizontal bar identification mark is cut out to simplify the explanation, but of course the same applies to the whole. In FIG. 15, the threshold value is set near both end points of the pan bottom (the luminance is 93 in the upper diagram and the luminance 174 in the lower diagram), but the outline shape clearly loses its individual characteristics. In FIG. 16, a low luminance value (middle part 119) and a high luminance value near the median value of the two luminance values corresponding to the peaks of the two peaks (upper figure 137) and to the extent that the features of the contour shape are not significantly lost. The three contours (lower figure 149) are taken as threshold values to obtain respective contour shapes. The width of the low luminance value and the high luminance value is as large as 30 and wide, but the feature of the contour shape is not lost significantly. However, there are some changes in the contour shape. Therefore, whether or not the contour shape is effective for determining the coincidence / non-coincidence of the printed matter even if the contour shape slightly changes is verified as follows.

In FIG. 17, two binarized micro images of a low luminance value (119) and a high luminance value (149) with a large threshold are compared for the binarized micro image of No. 1 printing. The minimum area where they do not overlap is defined as the difference area SI (filled portion). SI (space between identical) is a minimum difference area where micro-binarized images do not overlap when binarization is performed with greatly different threshold values in the case of the same printing. On the other hand, with respect to another printed solid identification mark (No. 2 printing), the threshold value is set near the central value of the two peaks (99) to obtain a binarized micro image, and the identification mark (No. 1 printing) of FIG. ), The threshold value is taken around the median value of the two peaks (137) to obtain a binarized micro image, and the two are compared. The minimum non-overlapping area is obtained as a difference area SD (space between different). Comparing both different areas SI and SD, in the case of the same printing solids, even if the threshold fluctuates considerably around the median of the peaks of the two peaks, the size of the difference area is different between different printing solids We found that it was significantly smaller than the area. That is, SI << SD. It has been found that this relational expression is maintained even if the threshold value changes as long as it is near the median value of the two peaks. In other words, the automatic threshold setting may be the median of the luminance values corresponding to the peaks of two peaks, and statistics such as a method for deriving the luminance values in the vicinity thereof, such as a discriminant analysis method for deriving a threshold value that maximizes the interclass variance An automatic threshold automatic derivation method may be used. And if the minimum area that does not overlap, that is, the difference area is measured and it is a value less than a certain value (SIT) determined by trial, it is SI and it is judged as the same printing solid, or it is a certain value determined by trial ( It was found that if the value is greater than or equal to (SDT), it is SD and can be determined as a separate print object. Judgment is easy because SIT and SDT differ greatly.
Based on the findings as described above, an arbitrary luminance value in the vicinity of the median value of the two peaks in the histogram is automatically set as a threshold, and the original image is automatically binarized to obtain a binarized micro image.
The relational expression and judgment criteria for the above different areas are between images with different imaging models, imaging environments (quality of irradiation light, irradiation angle, position of identification mark in image, size, resolution) or imaging technology (good or bad focus) It was proved by the trial that it was maintained even in the verification.
If the focus is poor, the same print solid is determined as another print solid. Therefore, if the image quality is determined and the focus is below a certain level, it is necessary to re-image and check and determine again.

The large difference between SI and SD is due to the difference in the contour shape of the binarized micro image. In the case of comparison between the same printing solids, there is little change in the contour shape even if the imaging model, imaging conditions and threshold value are different, and in the case of comparison between different printing, the difference in contour shape is different from the model, imaging conditions and threshold value. It was originally found to be very large regardless of how it was taken. The difference may be expressed by a minimum area that does not overlap as described above, or, as described in paragraph 0030 “A method for automatically matching binarized micro images and a method for automatically determining coincidence / mismatch”, other quantitative indicators such as You may express by the correlation coefficient of two outline shapes. In the case of the same printed solid, the correlation coefficient is very high, and in the case of different printed solids, the correlation coefficient is very low. This makes it possible to determine whether the printed solids match or not, regardless of the type of imaging and imaging conditions (imaging environment and imaging technology), and if the binarization threshold is near the center value of the two peaks, Nevertheless, it was discovered that it can be carried out with high accuracy. This discovery is a decisive indispensable discovery in an authenticity determination method that uses an image that is unavoidably unstable because the image is taken by an ordinary person using a general device (mobile phone with camera).

3. An automatic collation method of binarized micro images and a method of automatic determination of coincidence / mismatch by a computer.
As described above, the binarized micro images of the identification mark are always different from each other. There cannot be an exact match. The determination of the coincidence / non-coincidence is performed based on the degree of approximation or difference between the contour shapes. If the printed matter of the identification mark is the same, the closeness of the contour shape of the two binarized micro images is very high, while if the printed matter of the identification mark is different, the contour shape of the two binarized micro images is different. Sex is very high. This is as described in paragraph 0028. As described in paragraph 0029, the quantitative index of the approximation or the difference may be any quantity representing similarity such as the minimum area that does not overlap, the correlation coefficient of the contour shape, and RMS (Root Mean Square). These indices are automatically calculated by a computer, and when they exceed a certain numerical value or smaller than a certain numerical value, it is determined that the two printed identification marks match or do not match.

This natural phenomenon that the degree of difference is greatly different between the same printing and different printing is due to the printing characteristics (randomness without unevenness of the microscopic contour, gradation characteristics of the boundary area between the ink and the printing medium) and digital imaging. This is based on the optical characteristics of the device (relationship between resolution, focus failure, binarization threshold, etc. and contour shape). Utilizing this natural phenomenon, the automatic authentication method of the present application was invented. This is a surprising natural phenomenon. Since reading by a general device by a general person is a precondition, the model of the imaging device changes, the imaging condition (imaging environment and imaging technology) changes, the resolution and luminance of the captured image change, and the threshold for binarization of luminance Therefore, it may be common sense to think that the accuracy of the judgment of coincidence / non-coincidence cannot be maintained because the feature of the contour shape changes greatly. However, when an identification mark is printed as described in paragraph 0014, a threshold is automatically set as described in paragraph 0028, and an arbitrary quantitative index representing the difference in contour shape is calculated, this natural phenomenon I found out that

Various methods have been developed to combat the vast array of things in the world. Most of these methods are methods for preventing counterfeiting by attaching an identification mark that has been subjected to advanced technology that is difficult to counterfeit. However, even if a very high level of anti-counterfeiting technology is applied to the genuine article, it is meaningless for the purpose of preventing counterfeit distribution among ordinary people. This is because it is impossible for an ordinary person who does not have the knowledge to distinguish between the authenticity of the identification mark and the authenticity to determine the authenticity. To make an authenticity decision, you must bring the item to an expert. Even if an authenticity judgment is made by an expert, an authenticity judgment is still possible. However, the general public is bothersome and not usually done.
In order to get rid of the vast amount of treasures in the world, it is necessary for ordinary people to be able to determine the authenticity of all articles automatically, easily, instantaneously, almost in a habit, with high accuracy, using general-purpose devices. is there. If an authentication system with these abilities becomes widespread, the vast amount of the world's treasures can no longer exist and should be exterminated dramatically.

The conceptual diagram of an automatic authentication system. Certificate. Capture and record the image of the identification mark on the certificate (acquisition and storage of master information) Example of identification mark (No.1 horizontal printing, 1mm length) Inherent feature of micro contour shape. Compare contour shapes of different printed solids with the same identification mark shape. Smartphone + 2x lens. No.1, NoNo.2, No.3 identification mark from the top. Original image from the left, enlarged trimming, and micro image after binarization Gradient of micro outline area Change of contour shape by histogram and threshold of gradation image Reverse U type identification mark An image of an individual printed with the same identification mark by a mobile phone with multiple models. Part 1 An image of an individual printed with the same identification mark by a mobile phone with multiple models. Part 2 Upper half of horizontal one identification mark (Fig. 4) Histogram Definition of clear color difference Printing the identification mark and serial number on the product certificate Automatic registration of regular identification information Automatic image collation / judgment method Threshold values at both ends of the bottom of the histogram and binarized micro image Low, medium and high threshold values near the center of the bottom of the histogram and binarized micro image Minimum area that does not overlap The same printing and separate printing Primitive electronic image of identification mark and reference angle mark Histogram and threshold of primitive electron image and primitive binarized micro image Binary micro image of identification mark and reference angle mark Filling the binarized micro image of the identification mark and reference angle mark Setting the reference angle and removing the reference angle mark Physical quantity measurement and identifier to be matched Printing identification marks and reference angle marks Angular distance method Automatic registration of regular identification information

Several modes for carrying out the invention will be described below. Of course, embodiments are not limited to the methods described herein. In the following description, the reading device for the regular identification mark or the identification mark to be collated is mainly a digital camera (including a digital camera attached to a mobile phone), but the reading device may be a scanner.

First, a manufacturer or distributor of a target article or a contractor (referred to as a trader) entrusted with the article first places a micro-identification mark having a color clearly different from the background color as shown in FIG. 2 (certificate). Or print on the certificate. The width of the minute identification mark should be about 0.2 to 1 mm, and if it is long, the probability of reproducing the features of the random contour shape that cannot be controlled by humans can be made almost infinitely small, and it can be easily observed visually. The presence of minute identification marks can be searched. The size of the background zone around the identification mark is preferably such that when the identification mark is imaged by a reading device, no image other than the identification mark is imaged on the display screen. A width of about 3 times is desirable. Then, there is no worry that some dark colors other than the identification mark (color that becomes black when automatic binarization) are captured on the screen of the reading device.

Select printing inks that have strong water resistance, light resistance and aging stability. In addition, a combination of printing ink, printing medium, and printing machine is selected so that ink bleeds, protrusions, and omissions occur moderately in the micro contour portion.
There are various types of printing machines such as offset printing, gravure printing, letterpress printing, screen printing, ink jet printing, laser printing, etc., and any of them may be used, but the magnification required for the lens of the reader is microscopic depending on each. To be able to recognize the difference in contour shape.
Furthermore, the individual article identification code (manufacturing number or the like) of the target article is printed on this print medium (certificate).
The certificate may be printed electronically with the URL of the destination authenticator who transmits the identification mark of the present application.

Next, as shown in FIG. 3 (capturing and recording the image of the identification mark on the certificate), the contractor is near the center of the printing ink blank zone in the thick frame printed on the certificate 33. A minute identification mark and a part of the surrounding background zone are magnified by a 32 lens, and an enlarged image is captured by a 31 digital camera and read. Regarding the overall magnification of the lens and the digital camera zoom, it is necessary to have a total magnification that can read the features of the micro outline shape that cannot be controlled by the person of the identification mark.
Rather than obtaining the original electronic image by the above method, the basic shape of the identification mark is registered in advance, and only the identification mark and its peripheral region are cut out from the original electronic image as a template and used as the original electronic image. It is also good.

Next, the read original electronic image of the identification mark is transmitted to the server 35 of FIG. 3 and recorded in the storage device (database) 36, which is used as the original electronic image of the normal identification mark. At this time, the individual article identification code of the target article is simultaneously input and stored in the database.

A binary micro image is automatically extracted from the original electronic image of the read regular identification mark by automatically setting a threshold value as described in paragraph 0028 above. In the case of automatic binarization, if the height of two peaks in the luminance histogram becomes unbalanced and one peak becomes lower and the luminance position of the peak becomes unclear, the area around the original image is reduced. Trim to decrease. Both the original image and the binarized micro image or the latter are stored in the database as identification information of the regular identification mark.

This certificate is attached to the target item. The attaching method may not be physically connected. Thereafter, the target article is shipped to the market together with the certificate and distributed to the market. When an ordinary person wants to know whether the target article is manufactured by an authorized manufacturer, the authenticity of the article can be determined by the following procedure.
This will be described with reference to FIG. 1 (conceptual diagram of the automatic authentication system).

First, the general person reads the individual article identification code printed on the target article or the certificate attached thereto. Next, the micro-identification mark is enlarged to a proper magnification with the zoom function of the digital camera and, if necessary, with a simple lens function attached to the digital camera, and then it is copied to the screen of the mobile phone and expanded until the micro contour shape can be recognized. To do. The image is transmitted to the authentication center together with the individual article identification code.

The original electronic image of the verification identification mark received at the authentication center is binarized by automatically setting a threshold value by computer processing as described in paragraph 0028 above, and the binary of the verification identification mark Obtain a micro image. Alternatively, the basic shape of the identification mark may be registered in advance, and only the identification mark and the peripheral area in the vicinity thereof may be cut out from the original electronic image using the template as a template and automatically binarized as the original electronic image.
The binarized micro image and the binarized micro image of the regular identification mark stored in the database are subjected to matching processing by a computer, and a difference between them is calculated.
In the following, a method for matching contour shapes and determining whether or not they match by a computer will be described.

Matching and determination are performed according to the following series of steps (1) to (8).
(1) A binary image of a collation binary image is obtained by automatically binarizing the original image of the collation identification mark. On the other hand, a binarized micro image obtained by automatically binarizing regular identification marks is called a master binarized micro image. In the binarized image, the identification mark is black and the periphery is white. A micro-image obtained by automatically binarizing an identification mark is composed of a peninsula, a bay, a small island, a rock, a lake, and a pond. And the land is uniformly black and the water surface is uniformly white.
(2) The center of gravity (the center of gravity of the land area) of each of the black pixels of the master binarized micro image and the checked binarized micro image is obtained and the center of gravity is overlapped.
(3) Enlarging / reducing the total area of the black region of the binarized micro image to be collated to the same area as the entire black region of the master binarized micro image. The area of a region is the number of pixels in the region. This makes the size of the identification mark “almost” identical.
(4) The center of gravity of the two binarized micro images with the center of gravity overlap is used as an axis, and an arbitrary direction is set to an angle of 0 degrees. Or rotate left. The area of the black area of both images that does not overlap is determined as DS, and when the rotation angle is θ (i), it is defined as DS (θ (i)). As described in Example 3 in paragraph 0049, a reference angle mark may be printed to set a reference angle, and the reference angles of the two images may be matched.

(5) Continue to calculate each DS (θ (i)) in increments of Δθ, and let θ (min) be the angle at which the minimum value is obtained. DS (θ (min)) is obtained. Thus, the minimum value of the non-overlapping areas was roughly determined.
(6) However, if the position and size of the center of gravity are finely adjusted, an even smaller DS can be obtained. About the binarized micro image to be verified, the center of gravity position (X, Y axis) and size (size of the area of the black region) are changed in a small range with a constant increment, and for each case, θ (min ) In the small angle range around), calculate (5) above, and set the minimum value among all DS cases as “the smallest area that does not overlap” as minDS. A value obtained by dividing this minDS by the area of the master binarized micro image is defined as minR. This is the “difference or difference of micro contour regions”.

(7) The minR between the same printing individuals is minR (I), and that between different printing individuals is minR (D).
As described in paragraph 0028 above, even if the imaging device and imaging conditions change, the change in minR (I) is within a small range, and minR (D) is much larger than minR (I). minR (I) << minR (D). This facilitates automatic determination of coincidence / non-coincidence. As described in paragraph 0028 above, SIT and SDT as threshold values for determining whether the prints are the same or different are obtained as empirical values. Since minR (I) <SIT << SDT <minR (D), it can be determined whether minR (I) or minR (D) based on the obtained difference minR. That is, it is possible to determine whether the print solid is the same print solid or another print solid.

(8) Based on the result of the automatic determination, the authenticity of the target article is determined and returned.
With the above method, it can be determined with a very high probability whether or not the identification mark corresponding to the article has been properly printed. If the image received by the Authenticity Judgment Management Center and the master image can be determined to be the same as the binarized micro image, “the article with this identification mark is genuine”, or if it cannot be determined to be the same “ It is very likely that the item with the mark is not genuine. " The ordinary person can receive it and obtain the result of the authenticity judgment at the professional level.

By constructing the above system, even ordinary people can easily determine the authenticity of the identification mark at the expert level. As a result, the authenticity of the article with the identification mark of the present application can be determined with extremely high accuracy. If it is determined that the identification mark is genuine, it is confirmed that it is unique in the world, so it is almost certain that the article to which the identification mark is attached is also genuine. This is because there is no fake trader who purchases a large number of genuine articles to obtain a genuine identification mark and sells the purchased genuine certificate with a large number of goods. That's because you never get it.

The above explanation is mainly based on the assumption that a mobile phone with a digital camera is used. However, since its function is to capture, transmit, receive and display identification marks, a digital camera, a microscope, a USB microscope, etc. The same function can be obtained even if used.

The other operations are the same as those in the first embodiment except that the method for matching the binarized micro image of the regular identification mark and the binarized micro image of the identification mark to be compared is different from that of the first embodiment. Various methods of contour shape matching by a computer other than the first embodiment are disclosed and put into practical use. For example, in all combinations obtained by changing the size, horizontal / vertical movement, and rotation angle in a constant step, the sum of absolute values of the difference in y-axis values corresponding to certain x-axis values of both contour shapes The minimum difference is obtained as an index of the difference by calculating by a statistical processing method such as sum of squares, correlation coefficient, and the like. Alternatively, feature points such as end points, cusps, corner points, and bending points of both contour shapes are calculated by a computer, and they are compared to find inconsistent points, and the values obtained by quantifying the number and characteristics of the differences are used as a difference indicator. Get as. If the obtained difference index is within a certain value, it is automatically determined as a match, and if it exceeds a certain value, it is automatically determined as a mismatch.

The authenticity determination is performed according to the following series of steps (1) to (15).
(1) Print of identification mark and reference angle mark Please refer to FIG. Similar to the first embodiment or the second embodiment, a minute mark having an arbitrary shape and another arbitrary minute mark inside or near the mark on the target article or its certificate are displayed on the background. Print in a color that is distinctly different from the color. When it is provided inside the mark, it becomes an ink blank area. One mark is an identification mark and the other mark is a reference angle mark. Alternatively, instead of providing the reference angle mark inside and outside the identification mark as described above, a characteristic shape such as a concave portion or a convex portion may be provided on a part of the edge of the identification mark and printed as a reference angle mark. The following description explains the case where the reference angle mark is printed in the vicinity of the identification mark and the outside thereof. When the identification mark has any other shape, the method for obtaining the reference angle is described in paragraph number 0056, which will be described later, but the other operations are the same as when the reference angle mark is printed in the vicinity of the identification mark and the outside thereof. .

(2) Imaging of identification mark See FIG. The identification mark and the reference angle mark of (1) above are enlarged with a lens so that the features of individual micro-outline shapes that cannot be controlled by those persons can be recognized and imaged with a digital camera, or directly read electronically. An image is taken to obtain a primitive electronic image of the identification mark and the reference angle mark.

(3) Threshold setting and automatic binarization See FIG. Since the identification mark and the reference angle mark are printed in a color that is clearly different from the background color as described in the above (1), the histogram of the luminance and the number of pixels of the pixel of the original electronic image as described in paragraph 0014 above. Always have two and only two mountains. The median value of the luminance values corresponding to these two peaks or the luminance value in the vicinity thereof is automatically calculated as a threshold value by any general method, and the threshold value is automatically calculated as described in paragraph 0028 above. To automatically binarize the original image and obtain the original binarized micro image.

(4) Generation of identification mark binarized micro image from primitive binarized micro image See FIG. The primitive binarized micro image obtained in the above (3) is a binarized identification mark and reference angle mark, but a peninsula, a bay, a small island, and a pond are scattered as black areas. Among them, only the maximum area area (for example, the identification mark here) and the second area area (for example, the reference angle mark here) are extracted, and the identification mark binarized micro image and the reference angle mark 2 are extracted. A valued micro image is obtained.

(5) Generation of “Filling Mark Binarized Micro Image (Identifier)” and “Filling Reference Angle Mark Binarized Micro Image” See FIG. In the printing technique, it is inevitable that a blank or a light-colored zone such as a pond exists inside the binarized micro image of the identification mark and the binarized micro image of the reference angle mark. Therefore, it has a significant influence on the accuracy of determination by the angular distance method described later. Accordingly, the inside of the binarized micro image of the identification mark and the binarized micro image of the reference angle mark are all filled. As a result, a “filled identification mark binarized micro image” and a “fill reference angle mark binarized micro image” are obtained. The former is an identifier.

(6) Setting of reference angle See FIG. In the above (5), an identifier (filled identification mark binarized micro image) and a filled reference angle mark binarized micro image are obtained. For automatic determination of coincidence / non-coincidence, the characteristics of the above-described filled binary micro image must be represented by some physical method. The representation method may be an angular distance method (polar coordinate method) or an orthogonal coordinate method. In the angular distance method, a numerical sequence l _{1 to n} = l (θ _{1 to n} ) of an angle and a distance from the origin is an identifier. In the orthogonal coordinate method, (Y _{1 to n} , X _{1 to m} ) points are assigned as 1 when the points are on the outline, and 0 when there are no points, and the numerical sequence becomes an identifier. In any method, if the origin of the coordinates and the reference angle or the orthogonal coordinate axis are determined, the representation and the identifier can be easily verified.
For this purpose, first, the area barycentric position (B) of only the “fill reference angle mark binarized micro image” is obtained. Next, the area centroid position (A) of only the identifier (filled identification mark binarized micro image) is obtained, and the straight line connecting A and B is set as the reference angle of the identifier.

Alternatively, the reference angle can be obtained by another method. As described in paragraph 0049 above, a reference angle mark of a color distinctly different from the identification mark is printed inside or outside the identification mark area, or a blank area of printing ink is set and imaged in the same manner as described above. Then, automatic binarization is performed, and a maximum area area is extracted to obtain an identification mark binarized micro image. Next, a blank area having a maximum area as a reference angle mark is extracted from the image area, and the barycentric position is obtained. Next, all blanks or light-colored zones (including reference angle marks) in the area of the image are filled to obtain a filled identification mark binarized micro image (identifier). The barycentric position of the identifier is obtained, and the barycentric position of the reference angle mark is connected to the barycentric position to obtain the reference angle.

In addition, when a characteristic shape such as a concave portion or a convex portion is provided on a part of the edge of the identification mark and printed, the distance by the above-mentioned angular distance method is the shortest distance angle, the longest distance angle, or a change in distance. The reference angle may be obtained based on an angle having a large value.
There may be other methods for obtaining the reference angle, but any method may be used as long as the reference angle is obtained.

(7) Measurement of physical quantity to be collated and extraction of identifier The following method is called the angular distance method, and the physical quantity measurement and identifier extraction method will be described. For the orthogonal coordinates, the identifier described in the paragraph number 0054 is extracted. The two identifiers of the orthogonal coordinates are collated, and an index representing the similarity between them is calculated by a statistical method to determine coincidence / mismatch.
See FIG. In a solidified identification mark binarized micro image (identifier), rotate from the center of gravity A to the right or left and start from the center of gravity A at each angle (θ _i ), starting from the reference angle and increasing the minute angle. Measure the distance l (el) (θ _i ) to the contour line encountered.
If the contour shape consists of a complex peninsula or bay, and the bay bites along the contour line or the peninsula exists along the contour line, there will be multiple contour points encountered on the line from the center of gravity A, Measure the distance to the first encountered contour and ignore the others. Since the regular identifier to be stored and the identifier to be collated are similar in shape, they can be ignored. The above measurement is measured in a range of 360 degrees or a specific angle, and is defined as l (θ _i ). This number string represents the contour-shaped waveform almost continuously, and is an identifier.

In order for a computer to be able to match two micro contour shapes, the size of both contour shapes must be the same. However, there is no strict constant size for the micro contour shape. This is because the shape of each printed micro identification mark is random, and the contour region has a gradation of brightness, so there is no fixed size. Strictly speaking, it is impossible to equalize the size of individual printed matter. Further, the stored regular identification mark was taken by a contractor, whereas the verification identification mark was taken by an ordinary person with a camera-equipped mobile phone with an appropriately enlarged image. Different. Even in the same printed matter, the size of the image is different for each shot. Furthermore, the size of the binarized image also changes depending on the threshold value when binarizing the captured original image.
As a method of solving these problems, this l (θ _i ) is divided by the average value and indexed. By doing so, it is possible to collate / determine the coincidence / disagreement by calculating the degree of similarity of the contour shape regardless of the size of the identification mark.

(8) Storage of regular identification information See FIG. Next, both the original electronic image of the identification mark of (2) and the reference angle mark and the identifier consisting of the numerical sequence of l (θ _i ) of (7), or the latter, are electronically stored as normal identification information of the identification mark. Store / register in the device.

(9) Input of individual article identification code See FIGS. 24 and 25. The individual article identification code (serial number, product number, etc.) of the target article is stored in the electronic storage device in association with the regular identification information stored in the electronic storage device in (8) above.

(10) Imaging of verification target identification mark and input of individual verification target identification code Refer to FIG. 1 (conceptual diagram of automatic authentication system). The general public who wants to determine the authenticity of an article captures each collated identification mark and reference angle mark (both are collectively referred to as a minute mark) with a terminal device that can recognize the microscopic contour shape of each original electronic image. To get. Further, a collated individual article identification code corresponding to the collation identification mark is input to the terminal device.

(11) Transmission See FIG. 1 (conceptual diagram of automatic authentication system). The general person side transmits the original electronic image of the individual article identification code to be collated and the minute mark (the collation identification mark and the reference angle mark) input in (10) to the authenticity judgment person side through the Internet.

(12) The reception authenticity determination side receives the original electronic image of the collated individual article identification code and the minute mark (the collation identification mark and the reference angle mark) of (11) above.

(13) Automatic generation of identifier of verification identification mark See FIG. 1 (conceptual diagram of automatic authentication system). The authenticator side uses the computer processing described in (3) to (7) above from the original electronic image of the verification identification mark and the reference angle mark received in (12) above to obtain a numerical sequence of l (θ _i ). Get. The numerical sequence is divided by the average value and indexed. This numerical string is an identifier of the identification mark to be verified.

(14) Automatic verification and determination Regular identifier stored in the electronic storage device described in (8) above corresponding to the identifier of the identification mark to be verified obtained in (13) above and the received individual article identification code to be verified The identifier of the identification mark is automatically collated, and the match / mismatch is determined. The automatic verification method by computer and the automatic determination method of coincidence / non-coincidence are as follows.

Numerical sequence of the identifier of the collation identification mark (13), wherein the l (theta _i) was used as a X _i, numeric columns corresponding the identifier of the authorized identification mark according to (8) l (theta _i) Let Y _i be. When calculating the statistical numerical values representing at degree of coincidence in the shape of X _i of the shape and Y _i, the as described in (3), regardless of changes in the imaging model and imaging conditions, also Threshold Auto Set methods Regardless of this, it is possible to accurately determine whether the identification marks match or not. Any method may be used as long as the statistical value is a statistical index representing the degree of difference such as the correlation coefficient between X _i and Y _i , the root mean square of the difference between the two, or the sum of absolute values of the difference between the two. When the identification marks are different from those when the identification marks are the same, as described in the paragraph No. 0028 above, they are clearly different regardless of changes in the imaging conditions and regardless of the automatic threshold setting method. It is easy to determine the discrepancy.

The reason why the coincidence / non-coincidence can be determined with high accuracy in this way is that the identification mark and the reference angle mark are printed in a color that is clearly different from the background color as described in (1) above, and (3) to (7) This is because computer processing was performed. Further, in order to further increase the accuracy of coincidence / non-coincidence determination and reduce the amount of computer processing, before calculating the statistical index, the identifier (numerical string of l (θ _i )) is Fourier-transformed within a range where the accuracy of determination does not deteriorate. You may perform the process which smooths a fine unevenness | corrugation by conversion and moving average conversion. In addition, since the reference angle changes minutely depending on the imaging condition and the degree of matching of the focus, a small adjustment of the reference angle may be tried for the adjustment to increase the statistical index representing the degree of matching.

As described above, the general person side does not transmit both the captured image and the individual product identification code to be verified to the authenticator side, but only the individual product identification code input in (10) above is sent to the authenticator side. The identifier (l (θ _i )) of the regular identification mark stored in the electronic storage device of (8) is received and the corresponding identification mark and reference angle mark acquired in (10) are received. The identifier l (θ _i ) of the identification mark to be verified is automatically created from the electronic image by the method described in the above (3) to (7), both identifiers are automatically verified, and the coincidence mismatch is determined by the above statistical index. Automatic determination may be performed.
That is, the automatic generation of the identifier of the identification mark to be verified and the automatic determination of coincidence and coincidence with the automatic verification may be performed on the authenticity determination center side or on the general public side.

In addition, when the above-described statistical index for determining the coincidence mismatch is completely matched when the matching identification mark and the corresponding regular identification mark are matched, the identifier of the verification identification mark is an identifier based on a newly captured image. Instead, it is considered that an identifier based on a past captured image was stolen and transmitted to the authenticator. This is because images that completely match cannot be captured. In that case, the computer is programmed to issue a warning instead of a match.

(15) Notification of results See FIG. 1 (conceptual diagram of No. 12 automatic authentication system). Based on the result of the determination of coincidence / non-coincidence in (14) above, the authenticity of the identification mark and the target article is determined and returned or displayed.

Through the above steps (1) to (15), the variety of imaging model performance, which is the “problem to be solved by the invention”, and the diversity of image quality due to changes in imaging conditions (image brightness, light quality) , Size, angle, focus, etc.) has been solved, and automatic determination can be performed with very high accuracy. This diversity problem solution is an indispensable condition for ordinary people to authenticate with everyday equipment.

Currently, brand goods such as banknotes, medicines, electronic parts, bags, wallets, clothes, watches, glasses, home appliances, passports, identification cards, cash cards, credit cards, licenses, car verifications, diplomas It is said that the damage caused by counterfeit products such as securities, jewelery certificates, livestock pedigrees, pet pedigrees, etc. has reached $ 1 trillion worldwide. Enormous costs are spent on the prevention measures. According to the present invention, these counterfeits cannot exist.
The feature of the present invention is that the identification mark for authenticity determination can never be duplicated, and that an ordinary person can easily determine with an ordinary everyday device. For example, in the case of pharmaceuticals, if the packaging is almost the same as the real product, and if the form and chemical composition of the drug are almost the same, even an authorized manufacturer cannot immediately distinguish between the real product and the bag. Even in that case, if the micro-identification mark according to the present invention is printed or attached to the packaging box, the authenticity can be discriminated quickly. As a result, counterfeit goods cannot exist and the huge cost of extermination is unnecessary.
In addition, the feature of the present invention is that it can be executed only with devices that are used by ordinary people or devices that are available at a very low cost, and as a business operator on the authenticity judgment side, there is no need for completely new technology development and almost no capital investment. Therefore, it is possible to execute immediately with little cost.
However, an article that is manufactured and sold as a counterfeit product is not effective because it is known that the product is a counterfeit product.

11: Mobile phone with camera 12: Lens 13: Certificate 14: Identification mark 15: Server of the authentication center (executes automatic contour generation / automatic verification / automatic determination)
16: Storage device of authenticity judgment center 17: Electronic reader 18: Individual article serial number 21: Certificate 22: Ink-free zone and outer frame 23 as a mark for identification mark 23: Identification mark 31: Digital camera or mobile phone with camera Telephone 32: Lens 33: Certificate 34: Certificate identification mark 35: Authentication judgment center server 36: Authentication judgment center storage device 37: Electronic reader

Claims

An automatic authenticity determination method using a computer of individual articles,
Specify a clear color such that the difference between the background color and the brightness value of the color and other attribute values of the color is a certain value or more, and print a minute shape having an arbitrary shape as the identification mark with the color, An automatic authenticity determination method, wherein a micro contour shape having a shape control limit or less of the identification mark is used as an identifier.
A printed matter as an identification mark used for automatic authenticity determination according to claim 1.
The digital image of the printed identification mark according to claim 1, wherein the luminance value of the color and other attribute values of the color and the central value of the luminance value and attribute value for the two vertices appearing in the histogram consisting of the number of pixels corresponding to the color value. The value is automatically derived by a computer, the value is automatically set as a threshold value for binarization, and based on this, the digital image is automatically binarized by the computer, and the micro contour below the shape control limit is obtained. An automatic authentication method characterized by using a shape as an identifier.
In order to determine the identity and dissimilarity of the two printed materials from which the identifiers according to claim 3 are derived, when various mathematical methods are used as a method of comparing the two identifiers and determining whether they match or not, Use the fact that the printed material is the same or different, regardless of the imaging model, imaging conditions, and sharpness of image quality, for the mathematical indicators used (correlation coefficient, non-overlapping areas of difference, etc.) An automatic authentication method characterized by that.