JP6865797B2 - Authenticity judgment method, information processing device, and program - Google Patents

Description

The present disclosure relates to an authenticity determination method, an information processing device, and a program.

When opening an account at a financial institution, you will be required to show your identity verification documents. When opening an account by mail, the applicant mails the original or a copy of the identity verification document together with the application documents to the financial institution. In addition to these methods, the introduction of a mechanism that allows applicants to open an account by sending an image of an identity verification document to a financial institution using a computer has recently been considered. When introducing this mechanism, it is necessary to have a technology to correctly determine the authenticity of the identity verification document from the sent image.

Regarding the determination of the authenticity of a certificate, in Patent Document 1 below, the registered character pattern of the characters included in the certificate is recorded in association with the issuance area and the issuance period of the certificate, and the registered character pattern is used for proof. A method for determining the authenticity of a calligraphy has been proposed. In this method, the characters contained in the certificate image to be verified are recognized, the issue area and the issue time of the certificate are specified, and the used character pattern specified using the certificate image is compared with the registered character pattern. The authenticity of the certificate is judged based on the result. In addition, the following Patent Document 2 proposes a method of determining the similarity between images by using an image feature amount.

Japanese Unexamined Patent Publication No. 2019-3504 Japanese Unexamined Patent Publication No. 2018-142189

By applying the above-mentioned proposed method, it is possible to determine that the certificate image in which the description portion where a different character pattern is used depending on the delivery area and the delivery period is processed is a forgery. However, if the certificate image is processed by using characters having the same character shape as the registered character pattern, the certificate image can be determined to be genuine, so that the determination accuracy is not sufficient.

In view of the above circumstances, according to one aspect of the present disclosure, an object of the present disclosure is to provide an authenticity determination method, an information processing device, and a program capable of more accurately determining the authenticity of a certificate image. It is in.

According to the first aspect of the present disclosure, the image of the certificate to be verified is acquired, the first part in which the information about the issuer is described is left, and the content is different for each proved person. Using a mask image that hides at least the different second part, masking the acquired certificate image to generate a judgment image, and masking the sample image for the pre-prepared certificate This includes determining the authenticity of the acquired certificate image based on the comparison result between each image of the image group and the determination image by referring to the storage unit that stores the image group masked by the image. An authenticity determination method is provided in which the processing is executed by a computer.

According to the second aspect of the present disclosure, the acquisition unit for acquiring the image of the certificate subject to the authenticity determination and the first part in which the information about the issuer is described are left, and the contents are for each proved person. For the masking part that generates a judgment image by masking the image of the acquired certificate using the mask image that hides at least the second part that is different from each other, and the sample image of the certificate prepared in advance. A judgment unit that determines the authenticity of the acquired certificate image based on the comparison result between each image in the image group and the judgment image by referring to the storage unit that stores the image group masked by the masked image. An information processing device having the above is provided.

According to the third aspect of the present disclosure, the image of the certificate to be verified is acquired, the first part in which the information about the issuer is described is left, and the content is different for each proved person. Using a mask image that hides at least the different second part, masking the acquired certificate image to generate a judgment image, and masking the sample image for the pre-prepared certificate This includes determining the authenticity of the acquired certificate image based on the comparison result between each image of the image group and the determination image by referring to the storage unit that stores the image group masked by the image. A program is provided that causes the computer to perform the process. Further, according to the fourth aspect of the present disclosure, a non-transitory computer readable storage medium in which the program is stored is provided.

According to the present disclosure, the authenticity of the certificate image can be determined more accurately.

It is a block diagram for demonstrating the system including the authenticity determination apparatus which concerns on 1st Embodiment of this disclosure. It is a block diagram for demonstrating the hardware which can realize the function of the authenticity determination apparatus which concerns on 1st Embodiment of this disclosure. It is 1st schematic diagram for demonstrating masking and pattern learning which concerns on 1st Embodiment of this disclosure. It is a 2nd schematic diagram for demonstrating masking and pattern learning which concerns on 1st Embodiment of this disclosure. It is a schematic diagram for demonstrating the generation of the glyph image DB which concerns on 1st Embodiment of this disclosure. It is a flow diagram for demonstrating the operation of the determination device generator which concerns on 1st Embodiment of this disclosure. It is a block diagram for demonstrating the function of the 1st determination part which concerns on 1st Embodiment of this disclosure. It is a flow figure for demonstrating the process of black-painted image determination which concerns on 1st Embodiment of this disclosure. It is a block diagram for demonstrating the function of the 2nd determination part which concerns on 1st Embodiment of this disclosure. It is a flow diagram for demonstrating the process of glyph image determination which concerns on 1st Embodiment of this disclosure. It is a block diagram for demonstrating the function of the 3rd determination part which concerns on 1st Embodiment of this disclosure. It is a chart for demonstrating the date condition DB which concerns on 1st Embodiment of this disclosure. It is a chart for demonstrating the relationship DB which concerns on 1st Embodiment of this disclosure. It is a 1st flow chart for demonstrating the process of the relationship determination which concerns on 1st Embodiment of this disclosure. It is a 2nd flow chart for demonstrating the process of the relationship determination which concerns on 1st Embodiment of this disclosure. It is a block diagram for demonstrating the function of the comprehensive determination part which concerns on 1st Embodiment of this disclosure. It is a figure for demonstrating the determination mode DB which concerns on 1st Embodiment of this disclosure. It is a chart for demonstrating the threshold value DB which concerns on 1st Embodiment of this disclosure. It is a 1st flow chart for demonstrating the process of the comprehensive determination which concerns on 1st Embodiment of this disclosure. It is a 2nd flow chart for demonstrating the process of the comprehensive determination which concerns on 1st Embodiment of this disclosure. It is a chart for demonstrating the template image DB which concerns on 2nd Embodiment of this disclosure. It is a flow figure for demonstrating the process of black-painted image determination which concerns on 2nd Embodiment of this disclosure. It is a chart for demonstrating the character image DB which concerns on 2nd Embodiment of this disclosure. It is a 1st flow chart for demonstrating the process of glyph image determination which concerns on 2nd Embodiment of this disclosure. It is a 2nd flow chart for demonstrating the process of glyph image determination which concerns on 2nd Embodiment of this disclosure.

An embodiment of the present disclosure will be described below with reference to the accompanying drawings. In the present specification and the drawings, elements having substantially the same function may be designated by the same reference numerals to omit duplicate description.

<First Embodiment>
Hereinafter, the first embodiment of the present disclosure will be described.

[1-1. system]
A system including an authenticity determination device according to the first embodiment will be described with reference to FIG. FIG. 1 is a block diagram for explaining a system including an authenticity determination device according to the first embodiment.

As shown in FIG. 1, the system according to the first embodiment includes an authenticity determination device 10, a determination device generation device 20, and a user device 30. The authenticity determination device 10 is an example of the authenticity determination device according to the first embodiment. The determination device generation device 20 is an example of the determination device generation device according to the first embodiment. The user device 30 is an example of the user device according to the first embodiment.

The authenticity determination device 10 determines the authenticity of the certificate image received for identity verification. The determination device generation device 20 generates a determination device and a DB (Database) used by the authenticity determination device 10 for authenticity determination of the certificate image, and provides the determination device and the DB to the authenticity determination device 10. The user device 30 is connected to the authenticity determination device 10 via the network NW, and provides the certificate image to the authenticity determination device 10 as an identity verification document.

(hardware)
Here, with reference to FIG. 2, the hardware capable of realizing the function of the authenticity determination device according to the first embodiment will be described. FIG. 2 is a block diagram for explaining hardware capable of realizing the function of the authenticity determination device according to the first embodiment.

The function of the authenticity determination device 10 can be realized by using the hardware of the information processing device shown in FIG. That is, the function of the authenticity determination device 10 can be realized by controlling the hardware of the information processing device shown in FIG. 2 using a computer program.

The information processing apparatus shown in FIG. 2 includes a processor 10a, a memory 10b, a display IF (Interface) 10c, a communication IF 10d, and a connection IF 10e.

The processor 10a may be a CPU (Central Processing Unit), a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), a GPU (Graphic Processing Unit), or the like. The memory 10b may be a ROM (Read Only Memory), a RAM (Random Access Memory), an HDD (Hard Disk Drive), an SSD (Solid State Drive), a flash memory, or the like.

The display IF10c is an interface for connecting a display device such as an LCD (Liquid Crystal Display) or an ELD (Electro-Luminescence Display). The communication IF10d is a communication interface for connecting to the network NW. The network NW is a communication network that includes at least a part such as a WAN (Wide Area Network), a wired LAN (Local Area Network), a wireless LAN, and a mobile communication network. The network NW may include a wired communication network or may include a wireless communication network.

The connection IF10e is an interface for connecting an external device. The connection IF10e may be a USB (Universal Serial Bus) port, an IEEE1394 port, a SCSI (Small Computer System Interface), or the like. An input IF such as a keyboard, a mouse, a touch panel, or a touch pad can be connected to the connection IF10e. A non-temporary portable recording medium 10f that can be read by a computer can be connected to the connection IF10e. The recording medium 10f may be a magnetic recording medium, an optical disk, a magneto-optical disk, a semiconductor memory, or the like.

The processor 10a may read the program stored in the recording medium 10f and store it in the memory 10b, and control the operation of the authenticity determination device 10 according to the program read from the memory 10b. The program that controls the operation of the authenticity determination device 10 may be stored in the memory 10b in advance, or may be downloaded via the communication IF 10d.

Although the hardware capable of realizing the functions of the authenticity determination device 10, the functions of the determination device generation device 20 and the user device 30 can also be realized by using the hardware of the information processing device shown in FIG. Therefore, the above description of the hardware will be incorporated, and detailed description of the hardware of the determination device generator 20 and the user device 30 will be omitted.

(Function of authenticity judgment device)
The function of the authenticity determination device 10 will be further described with reference to FIG. 1 again. As shown in FIG. 1, the authenticity determination device 10 includes a confirmation document receiving unit 11, a storage unit 12, a first determination unit 13, a second determination unit 14, a third determination unit 15, and a comprehensive determination unit 16.

The function of the confirmation document receiving unit 11 can be realized mainly by the processor 10a and the communication IF 10d described above. The function of the storage unit 12 can be realized mainly by the memory 10b described above. The functions of the first determination unit 13, the second determination unit 14, the third determination unit 15, and the comprehensive determination unit 16 can be realized mainly by the processor 10a described above. The functions of the first determination unit 13, the second determination unit 14, and the third determination unit 15 may be assigned to different information processing devices by using a plurality of information processing devices.

When the confirmation document reception unit 11 receives an application for a procedure (account opening procedure, etc.) that requires identity verification from the user device 30, it requests the user device 30 to present the identity verification document. In addition, the confirmation document reception unit 11 receives a certificate image presented as the identity verification document from the user device 30 in response to the request for presentation of the identity verification document. Then, the confirmation document receiving unit 11 stores the certificate image in the storage unit 12. Further, the confirmation document receiving unit 11 transmits a message according to the authenticity determination result of the certificate image to the user device 30.

The first determination unit 13 masks the certificate image stored in the storage unit 12 by the confirmation document receiving unit 11 to generate a black-painted image, and uses a pattern recognizer to generate a certificate based on the black-painted image. The authenticity of the image (hereinafter referred to as black-painted image judgment) is performed. The black-painted image is an example of a determination image. The pattern recognizer is obtained by having AI (Artificial Intelligence) learn a large number of template images generated by masking a large number of certificate images prepared in advance. The method of generating the pattern recognizer and the method of determining the black-painted image will be further described later.

The second determination unit 14 extracts an image of the characters or numbers described in the certificate (hereinafter referred to as a glyph image) from the certificate image stored in the storage unit 12 by the confirmation document receiving unit 11, and uses the glyph image DB. Then, the authenticity determination of the certificate image based on the glyph image (hereinafter referred to as the glyph image determination) is performed. The glyph image DB is a database that stores character and number figures (hereinafter, glyph images) extracted from a large number of certificate images prepared in advance. The method of generating the glyph image DB and the method of determining the glyph image will be further described later.

The third determination unit 15 identifies a plurality of mutually related description contents from the certificate image stored in the storage unit 12 by the confirmation document reception unit 11, and identifies them by using a DB showing the relationship between the description contents. By confirming the consistency between the described contents, the authenticity of the certificate image (hereinafter referred to as the relationship judgment) is determined. An example of a DB showing the relationship of the described contents and a method of determining the relationship will be further described later.

The comprehensive determination unit 16 comprehensively considers the results of the authenticity determination by the first determination unit 13, the second determination unit 14, and the third determination unit 15, and carries out the final authenticity determination (hereinafter, comprehensive determination). .. The method of comprehensive determination will be further described later. The result of the comprehensive determination is input to the confirmation document reception unit 11. The confirmation document receiving unit 11 transmits a message according to the result of the comprehensive determination to the user device 30.

(Function of judgment device generator)
Next, the function of the determination device generator 20 will be described.

The determination device generator 20 has an analysis unit 21 and a storage unit 22. The function of the analysis unit 21 can be realized mainly by using the hardware resources corresponding to the processor 10a described above. The function of the storage unit 22 can be realized mainly by using the hardware resource corresponding to the memory 10b described above.

The analysis unit 21 generates a pattern recognizer used by the first determination unit 13. In addition, the analysis unit 21 generates a glyph image DB used by the second determination unit 14. The storage unit 22 includes a certificate image group 22a including a large number of certificate images prepared in advance for a certificate used as an identity verification document, and a mask image 22b used when generating a black-painted image. Is stored. The mask image 22b is an image showing a portion to be masked. Hereinafter, the certificate image included in the certificate image group 22a is referred to as a sample image.

(How to generate a pattern recognizer)
When generating the pattern recognizer, the analysis unit 21 uses the mask image 22b to mask the sample image included in the certificate image group 22a. Hereinafter, the sample image after masking will be referred to as a template image. In addition, the analysis unit 21 causes AI to learn a large number of template images generated by masking to generate a pattern recognizer. Any AI technique available for image pattern recognition can be applied to the generation of the pattern recognizer. Further, as a learning method, known machine learning and deep learning can be arbitrarily applied.

Here, with reference to FIGS. 3 and 4, the masking and pattern learning according to the first embodiment will be further described with reference to specific examples. FIG. 3 is a first schematic diagram for explaining masking and pattern learning according to the first embodiment. FIG. 4 is a second schematic diagram for explaining masking and pattern learning according to the first embodiment.

FIG. 3 shows, as an example, a sample image 41 when a driver's license is used as an identity verification document. The sample image 41 has description fields 41a, ..., 41n including any of characters, numbers, photographs, and figures. The name of the person to be certified is written on the left side of the entry column 41a, and the date of birth is written on the right side. The address of the person to be certified is described in the entry field 41b. The date of issuance of the driver's license is described on the left side of the entry column 41c, and the reference number assigned according to the issuance date is described on the right side. This reference number may be assigned according to different standards for each issuing Public Safety Commission.

A photograph of the person to be proved is attached to the entry column 41d. The expiration date of the driver's license is described in the entry column 41e. The background color of the expiration date is determined according to the attributes of the certified person regarding the driver's license (the number of days elapsed since the license was obtained, the accident situation, the violation situation, etc.). For example, if the subject is a "good driver", the background color of the expiration date will be gold. In this case, a mark indicating "excellent driver" is described in the description column 41g. This mark is composed of the letters "excellent" and the border surrounding the letters. In addition, the background color will be green until the first renewal after obtaining the license. Otherwise, the background color will be blue.

In the entry column 41f, the conditions of the license and the like are described. For example, when the type of driver's license is "automatic limited license", "ordinary car is limited to AT car" is described in the entry column 41f. Further, in the entry column 41f, limited conditions such as "limited to a semi-medium-sized vehicle (5t)", "limited to a medium-sized vehicle (8t)", and "limited to a small two-wheeled AT vehicle" are described. Since the type of limited license may differ depending on the time of issuance, the content of the limited conditions may change depending on the time of issuance.

A 12-digit number is entered in the entry field 41h. Of the 12 digits, the 2 digits (1st and 2nd) from the left indicate the Public Safety Commission number. Therefore, the Public Safety Commission that issued the driver's license can be identified by referring to the first and second numbers of the numbers described in the entry column 41h. The next two digits (third and fourth digits) are numbers indicating the year of acquisition of the driver's license (last two digits of the Christian era). Therefore, the year in which the driver's license was first obtained can be specified by referring to the third and fourth numbers in the entry column 41h.

The next 6 digits (5th to 10th) are control numbers that can be set by the Public Safety Commission of each prefecture based on their own standards. The next one digit (11th) is the check digit. The last digit (12th) indicates the number of reissuances.

In the entry column 41i, a character string indicating that the driver's license is used is entered. The type of license is described in the entry column 41j. In the entry column 41k, the date of acquisition of the license for the corresponding category is entered. For categories that include multiple licenses, the date of acquisition of the first license acquired among the licenses included in that category is listed in the corresponding column.

The name of the Public Safety Commission is entered in the entry column 41m. The seal of the Public Safety Commission may be displayed in the entry column 41n. Therefore, the issuer of the driver's license can be specified by referring to the entry fields 41m and 41n. The driver's license issued by a specific Public Safety Commission may not display the seal of the Public Safety Commission.

In addition to the contents described in the above-mentioned entry column, the driver's license has a frame line or a pattern surrounding the entry column. The glyphs of letters and numbers described in the above-mentioned entry fields, the arrangement of letters and numbers, the contents described in a specific entry column, and the description method of the border (for example, the thickness of the line, the shape of the frame, The notation of the driver's license, such as the layout of the frame), may differ from issuer to issuer. Therefore, by letting AI learn the image of the part of the driver's license that contains information that can identify the issuer and the image of the part where the notation can change, it is possible to recognize different differences for each issuer. It is possible to generate a pattern recognizer that can be used.

As mentioned above, a driver's license contains many fields in which information about the person to be certified is described. Information about the certified person includes the name, date of birth, address, license number (number in the entry field 41h), license conditions, acquisition date, and type of license. In addition, the expiration date depends on the date of issuance and the attributes of the certified person regarding the driver's license (excellent driver, etc.). The information that depends on the proved person becomes noise when the pattern recognizer for recognizing the information that depends on the issuer is generated. Therefore, the analysis unit 21 masks the sample image 41 so as to eliminate as much information as possible depending on the person to be proved.

In the template image 42 illustrated in FIG. 3, while most of the information depending on the proved person is masked, "Tokyo" indicating the administrative division (prefecture) among the addresses described in the description column 41b. The characters "Miyako" are left unmasked. In the case of a driver's license, the issuer is the Public Safety Commission of the administrative division in which the person to be certified resides, so the characters indicating the administrative division can be used as information that can be used to identify the issuer. As in this example, positively leaving information about the issuer has an advantage in improving the recognition accuracy when generating a pattern recognizer that recognizes information depending on the issuer.

The same applies when a certificate other than a driver's license is used as an identity verification document. For example, when using an individual number card (my number card), the result is as shown in FIG. FIG. 4 illustrates a sample image 51 of the Individual Number Card and a template image 52 masking the sample image 51. In this example, the name of the person to be certified is described in the entry column 51a, and the address is entered in the entry column 51b. The date of birth of the person to be certified is described in the entry field 51c, and a mark indicating the type of certificate (individual number card) is described in the entry field 51d.

The gender of the person to be certified is described in the entry column 51e, and the expiration date is described in the entry column 51f. The expiration date of the Individual Number Card is set from the issue date to the 10th birthday. Therefore, the expiration date and the date of birth are the same. In the entry field 51g, characters indicating the mayor of the city, ward, town, or village that issued the Individual Number Card are entered. Therefore, the issuer who issued the Individual Number Card can be identified by referring to the entry field 51g.

The expiration date of the digital certificate may be described in the entry field 51h. The expiration date of the digital certificate is stated when the certified person issues the digital certificate. A photograph of the person to be proved is attached to the entry column 51j. The entry columns 51i and 51n are entry columns for items to be entered by the certified person or the issuer. The numbers in the fields 51k and 51m may include information about the proved person or the issuer.

In addition to the contents described in the above-mentioned entry column, the personal number card has a frame or pattern surrounding the entry column. Character shapes of letters and numbers described in the above-mentioned entry fields, arrangement of letters and numbers, contents described in specific entry columns, description method of borders (for example, line thickness, frame shape, frame The notation of the Individual Number Card, such as the layout) and the background pattern, may differ from issuer to issuer. Therefore, as in the case of a driver's license, by letting AI learn the image of the part containing information that can identify the issuer and the image of the part where the notation can change, different differences are recognized for each issuer. It is possible to generate a pattern recognizer that can be used.

The name, address, date of birth, gender, photo, and expiration date that depend on the date of birth are information that depends on the person to be certified. The description contents of the entry fields 51k, 51m, and 51n may also include information depending on the proved person. The expiration date of the digital certificate also depends on the behavior of the person to be certified (application for issuance of the digital certificate). This information becomes noise when generating a pattern recognizer for recognizing information depending on the issuer. Therefore, the analysis unit 21 masks the sample image 51 so as to eliminate such information as much as possible.

In the template image 52 illustrated in FIG. 4, most of the information depending on the proved person is masked, while among the addresses described in the description column 51b, the administrative division (city, ward, town, village) is shown. The characters "○○ ward, Tokyo" are left unmasked. In the case of an Individual Number Card, the issuer is the head of the administrative division in which the person to be certified resides, so the characters indicating the administrative division can be used as information that can be used to identify the issuer. As in this example, positively leaving information about the issuer has an advantage in improving the recognition accuracy when generating a pattern recognizer that recognizes information depending on the issuer.

As in the example shown in FIGS. 3 and 4, the analysis unit 21 masks the sample image included in the certificate image group 22a to generate a template image. Then, the analysis unit 21 causes AI to learn the template image to generate a pattern recognizer. In the above example, a driver's license and an individual number card are taken as an example, but the above method can also be applied to a driver's license, a passport, and the like. The analysis unit 21 provides the generated pattern recognizer to the first determination unit 13.

(Glyph image judgment)
Next, the generation of the glyph image DB according to the first embodiment will be described with reference to FIG. FIG. 5 is a schematic diagram for explaining the generation of the glyph image DB according to the first embodiment. FIG. 5 illustrates a driver's license sample image 41 and a glyph image DB 43 generated from a driver's license sample image group including the sample image 41.

As already explained, the notation of a driver's license can vary from issuer to issuer. In addition, the notation of the driver's license may change depending on the time of issuance. For example, the glyphs of letters and numbers described in a specific part may change depending on the issuer and the time of issuance. Therefore, the glyphs of letters and numbers described in a specific part can be used to determine the authenticity of a driver's license. The glyph image DB 43 is a database that stores information on the correct glyphs used in determining the authenticity. The issuer, delivery time, description location, characters / numbers, and images of the characters / numbers (character-shaped images) are stored in the character-shaped image DB 43 in association with each other.

The analysis unit 21 extracts a sample image of the driver's license from the certificate image group 22a, and recognizes the characters / numbers described in the predetermined description locations by using an arbitrary character recognition technique. For example, the character recognition technology used in OCR (Optical Character Recognition) can be used. In the case of the sample image 41 illustrated in FIG. 5, the analysis unit 21 recognizes the character "Tokyo" indicating the administrative division among the addresses described in the description column 41b, and puts it in the issuer column of the glyph image DB 43. Record. Further, the analysis unit 21 recognizes the delivery date described in the description column 41c, and records the time corresponding to the delivery date in the delivery time column of the glyph image DB 43. The analysis unit 21 may record the date of delivery in the delivery time column.

In addition, the analysis unit 21 recognizes the characters / numbers of the preset description location, and records the information indicating the description location in the description location column of the glyph image DB 43. Further, the analysis unit 21 records the recognition result of the characters / numbers described in the description location in the character / number column of the character image DB 43, and stores the character image of the character / number in the character image column of the character image DB 43. Record. In the example of FIG. 5, the number "5" of the date of birth described in the description column 41a is recognized, the "date of birth column" is in the description column, "5" is in the character / number column, and the glyph image. The character image is recorded in the column.

Similar to the above, in the example of FIG. 5, the character "Heisei" of the era name is recognized among the expiration dates described in the description column 41e, and the "expiration date column" and the characters are displayed in the description location column of the glyph image DB43. -"Heisei" is recorded in the number column, and the glyph image is recorded in the glyph image column. Information is also recorded in the glyph image DB 43 for the character "number" described in the description column 41h.

In the example of FIG. 5, for convenience of explanation, a thick frame is attached to the attention portion of the sample image 41, and the matter recorded in the glyph image DB 43 is illustrated for that portion, but other description portions are processed in the same manner. .. The description location to be processed may be all description locations or any description location selected in advance. For example, if the description portion of the portion having no background color is targeted for processing, character recognition becomes easy. Further, if the description place (for example, description column 41i) in which the same character is described is targeted for processing regardless of the person to be certified, the issuer, and the delivery time, that character is included in almost all sample images. Many glyph images can be obtained.

Here, the method of generating the glyph image DB 43 for the driver's license has been shown, but the same method can be applied to the other certificates to generate the glyph image DB. By applying this method, it is possible to generate a glyph image DB even for a certificate whose relationship between the issuer and issuance time and the glyph image is unknown. The analysis unit 21 provides the second determination unit 14 with a glyph image DB for each certificate that can be used as an identity verification document.

(Operation of the judgment device generator)
Next, the operation of the determination device generator according to the first embodiment will be described with reference to FIG. FIG. 6 is a flow chart for explaining the operation of the determination device generator according to the first embodiment.

(S101) The analysis unit 21 reads each sample image of the certificate image group 22a stored in the storage unit 22.

(S102) The analysis unit 21 masks each sample image using the mask image 22b stored in the storage unit 22, and generates a template image corresponding to each sample image. For example, as shown in FIGS. 3 and 4, the analysis unit 21 generates a template image in which information about the issuer is left and information depending on the proved person is masked. At the time of masking, a part of the information about the issuer in the sample image may be masked, or a part of the information depending on the proved person may be left.

(S103, S104) The analysis unit 21 causes AI to learn a large number of template images corresponding to a large number of sample images included in the certificate image group 22a to generate a pattern recognizer. Since the information about the issuer is left in the template image and the information that depends on the prover is masked, the pattern recognition that can recognize the difference in the notation that depends on the issuer for the notation that does not depend on the prover. A vessel is generated. The analysis unit 21 provides the generated pattern recognizer to the first determination unit 13.

(S105) The analysis unit 21 analyzes each sample image, identifies the issuer and the delivery time, and reads the characters / numbers of the preset description location (predetermined column). For example, the analysis unit 21 reads the address of the proved person by character recognition, and identifies the issuer from the characters indicating the administrative division of the address. In addition, the analysis unit 21 reads the delivery date by character recognition and specifies the delivery time corresponding to the delivery date. In addition, the analysis unit 21 reads the characters / numbers in the predetermined column by character recognition, and cuts out the character-shaped image of the characters / numbers from the sample image.

(S106, S107) The analysis unit 21 stores the issuer, the issuance time, the characters / numbers, and the character / number character-shaped images in association with the character-shaped image DB for each type of certificate. Then, the analysis unit 21 provides the glyph image DB to the second determination unit 14. When the process of S107 is completed, the series of processes shown in FIG. 6 is completed. As a modification, the processing groups of S102 to S104 and the processing groups of S105 to S107 may be exchanged, or these two processing groups may be executed in parallel.

(Functions of user equipment)
The function of the user device 30 will be described with reference to FIG. 1 again. As shown in FIG. 1, the user device 30 includes an imaging unit 31, a storage unit 32, a control unit 33, and a display unit 34.

The function of the image pickup unit 31 can be realized by using a camera unit having an optical system including a lens and an image pickup element such as a CCD (Charge-Coupled Device) sensor or a CMOS (Complementary Metal Oxide Semiconductor) sensor. The function of the storage unit 32 can be realized mainly by using the hardware resource corresponding to the memory 10b described above. The function of the control unit 33 can be realized mainly by using the hardware resources corresponding to the processor 10a described above. The function of the display unit 34 can be realized mainly by using the hardware resource corresponding to the display IF10c described above and the display device connected to the hardware resource.

The imaging unit 31 provides the camera function of the user device 30. For example, the imaging unit 31 is used to image a certificate used as an identity verification document. In this case, the imaging unit 31 images the certificate according to the control by the control unit 33, and stores the certificate image in the storage unit 32. Further, the imaging unit 31 may be used to image the face of the user who operates the user device 30. In this case, the imaging unit 31 stores the user's facial photograph in the storage unit 32.

The control unit 33 transmits the certificate image stored in the storage unit 32 to the authenticity determination device 10 in response to a user operation. At this time, the control unit 33 may transmit the user's face photograph together with the certificate image to the authenticity determination device 10. By transmitting the user's face photo together, the certifier sends the certificate image on the authenticity determination device 10 side based on the similarity between the face photo of the certificate image and the received user's face photo. You will be able to confirm that you have done so.

The display unit 34 may display a certificate image captured by the imaging unit 31 and a photograph of the user's face. Further, the display unit 34 may display a message transmitted from the authenticity determination device 10. For example, when applying for a procedure requiring identity verification, an information input form used for the application may be provided from the authenticity determination device 10 side, and the information input form may be displayed on the display unit 34. In addition, a message indicating the result of authenticity determination for the certificate image transmitted from the user device 30 may be displayed on the display unit 34.

The user device 30 is an information processing device used by the proved person to transmit the certificate image to the authenticity determination device 10 as an identity verification document, and the form thereof is arbitrary. For example, the user device 30 may be a mobile phone, a smartphone, a tablet terminal, a notebook PC (Personal Computer), a desktop PC, a game machine, or a store terminal such as an ATM (Automatic Teller Machine). In the example of FIG. 1, the number of user devices is one, but a plurality of user devices may be connected to the authenticity determination device 10 via the network NW.

The system according to the first embodiment has been described above. Hereinafter, the functions and operations of the first determination unit 13, the second determination unit 14, the third determination unit 15, and the comprehensive determination unit 16 included in the authenticity determination device 10 will be further described.

[1-2. 1st judgment unit (black-painted image judgment)]
First, the function of the first determination unit according to the first embodiment will be described with reference to FIG. 7. FIG. 7 is a block diagram for explaining the function of the first determination unit according to the first embodiment.

As shown in FIG. 7, the first determination unit 13 includes a storage unit 131, a target image acquisition unit 132, a masking unit 133, and a black-painted image determination unit 134.

The mask image 131a and the pattern recognizer 131b are stored in the storage unit 131. The mask image 131a is the same as the mask image 22b used by the determination device generator 20 when masking the sample image of the certificate image group 22a. The pattern recognizer 131b is a pattern recognizer generated by the determination device generator 20.

The target image acquisition unit 132 acquires a certificate image (hereinafter referred to as a target image) that is a target of authenticity determination from the confirmation document reception unit 11. The masking unit 133 masks the target image using the mask image 131a. Hereinafter, the target image after masking may be referred to as a black-painted image. The black-painted image determination unit 134 inputs the black-painted image generated by the masking unit 133 to the pattern recognizer 131b.

The pattern recognizer 131b recognizes the coincidence points and differences between the learned template image and the input black-painted image, and outputs a certainty degree C1 as a recognition result, which decreases as the degree of partial difference increases. To do. In the template image and the black-painted image, the part containing the information about the publisher is not masked, so that part is recognized as a coincidence point, and the difference between the template image and the black-painted image in other parts is the pattern recognition. Evaluated by vessel 131b. The certainty degree C1 output from the pattern recognizer 131b is input to the comprehensive determination unit 16.

Here, the flow of the black-painted image determination process according to the first embodiment will be described with reference to FIG. FIG. 8 is a flow chart for explaining the black-painted image determination process according to the first embodiment.

(S111) The masking unit 133 masks the target image acquired by the target image acquisition unit 132 using the mask image 131a stored in the storage unit 131, and generates a black-painted image corresponding to the target image. To do. The masking unit 133 outputs the generated black-painted image to the black-painted image determination unit 134.

(S112) The black-painted image determination unit 134 inputs the black-painted image generated by the masking unit 133 to the pattern recognizer 131b stored in the storage unit 131. Then, the black-painted image determination unit 134 outputs the certainty level output as the recognition result from the pattern recognizer 131b to the comprehensive determination unit 16.

(S113) The first determination unit 13 determines whether or not the processes of S111 and S112 (black-painted image determination) have been executed for all the target images. When all the target images have been processed, the series of processes shown in FIG. 8 ends. On the other hand, when there is an unprocessed target image, the process proceeds to S111, and the processes of S111 and S112 are executed for the unprocessed target image.

[1-3. Second Judgment Unit (Glyph Image Judgment)]
Next, the function of the second determination unit according to the first embodiment will be described with reference to FIG. 9. FIG. 9 is a block diagram for explaining the function of the second determination unit according to the first embodiment.

As shown in FIG. 9, the second determination unit 14 includes a storage unit 141, a target image acquisition unit 142, a character recognition unit 143, and a character shape determination unit 144.

The glyph image DB 141a and the similarity determination device 141b are stored in the storage unit 141. Like the above-mentioned glyph image DB43 corresponding to the driver's license, the glyph image DB 141a is a certificate in which the issuer, the delivery time, the description location, the characters / numbers, and the glyph images of the characters / numbers are stored in association with each other. It is a glyph image DB corresponding to a calligraphy.

The similarity determination device 141b is a program module that compares two images and outputs their similarity. The similarity determination device 141b may be generated by the determination device generator 20 using AI learning, or a general-purpose similarity determination device may be applied. As a method for determining the similarity, image features such as BRIEF (Binary Robust Independent Elementary Features), ORB (Oriented FAST and Rotated BRIEF), SIFT (Scale-Invariant Feature Transform), and SURF (Speeded-Up Robust Features) were used. There is a judgment method.

The target image acquisition unit 142 acquires the target image from the confirmation document reception unit 11. The character recognition unit 143 uses an arbitrary character recognition technique to read the characters / numbers in the description field that can be used to identify the issuer, and identifies the issuer from the characters / numbers. In addition, the character recognition unit 143 reads the characters / numbers in the description column that can be used to specify the delivery time by character recognition, and specifies the delivery time from the characters / numbers.

In addition, the character recognition unit 143 identifies the record of the character image DB 141a corresponding to the specified issuer and the delivery time. Then, the character recognition unit 143 reads characters / numbers from the corresponding portion of the target image by character recognition for the description portion included in the specified record, and cuts out the glyph image of the characters / numbers. Then, the character recognition unit 143 further extracts records corresponding to the read characters / numbers, and provides the character shape determination unit 144 with the information of the extracted records and the character shape image cut out from the target image.

The glyph determination unit 144 acquires the glyph image included in the corresponding record of the glyph image DB 141a based on the information of the record provided by the character recognition unit 143. Then, the glyph determination unit 144 inputs the glyph image provided by the character recognition unit 143 and the glyph image acquired from the glyph image DB 141a into the similarity determination device 141b. The similarity determination device 141b outputs the similarity between the two input glyph images.

The glyph determination unit 144 uses the similarity determination device 141b to calculate the similarity of the glyph images corresponding to all the description locations. The degree of similarity may be standardized so that it is 1 when it is the most similar and 0 when it is not the most similar. The glyph determination unit 144 calculates the certainty degree C2 based on the calculated similarity degree. For example, confidence C2 is given as a product of similarity corresponding to all descriptions. The glyph determination unit 144 provides the calculated certainty degree C2 to the comprehensive determination unit 16.

Here, with reference to FIG. 10, the flow of processing for the glyph image determination process according to the first embodiment will be described. FIG. 10 is a flow chart for explaining the process of determining the glyph image according to the first embodiment.

(S121) The character recognition unit 143 uses an arbitrary character recognition technique to read the characters / numbers in the description field that can be used to identify the issuer, and identifies the issuer from the characters / numbers. In addition, the character recognition unit 143 reads the characters / numbers in the description column that can be used to specify the delivery time by character recognition, and specifies the delivery time from the characters / numbers. In addition, the character recognition unit 143 identifies the record of the character image DB 141a corresponding to the specified issuer and the delivery time. Then, the character recognition unit 143 extracts all the description parts corresponding to the specified record.

(S122, S125) The processes of S123 and S124 are executed for each of the extracted description points. For example, the date of birth column, expiration date column, and number in the description location column of the glyph image DB 141a corresponding to the driver's license whose issuer is "Tokyo" and the issue date is "April-September 2004" When there is a description of column ..., the processes of S123 and S124 are repeatedly executed for each of these columns. When the processing is completed for all the described parts, the processing proceeds to S126.

(S123) The character recognition unit 143 reads characters / numbers from the description location of the corresponding target image, and cuts out a glyph image of the characters / numbers. Further, the character recognition unit 143 extracts a record corresponding to the read character / number from the character image DB 141a.

(S124) The glyph determination unit 144 acquires the glyph image included in the record of the glyph image DB 141a extracted by the character recognition unit 143 in S123. Then, the glyph determination unit 144 inputs the glyph image cut out from the target image by the glyph recognition unit 143 and the glyph image acquired from the glyph image DB 141a into the similarity determination device 141b, and the similarity between the two input glyph images. Calculate the degree. The degree of similarity may be standardized so that it is 1 when it is the most similar and 0 when it is not the most similar.

(S126) The glyph determination unit 144 calculates the certainty degree C2 based on the calculated similarity degree. For example, confidence C2 is given as a product of similarity corresponding to all descriptions. The glyph determination unit 144 provides the calculated certainty degree C2 to the comprehensive determination unit 16.

(S127) The second determination unit 14 determines whether or not the processes (character-shaped image determination) of S121 to S126 have been executed for all the target images. When all the target images have been processed, the series of processes shown in FIG. 10 ends. On the other hand, when there is an unprocessed target image, the process proceeds to S121, and the processes of S121 to S126 are executed for the unprocessed target image.

[1-4. Third Judgment Unit (Relationship Judgment)]
Next, the function of the third determination unit according to the first embodiment will be described with reference to FIG. FIG. 11 is a block diagram for explaining the function of the third determination unit according to the first embodiment.

As shown in FIG. 11, the third determination unit 15 includes a storage unit 151, a target image acquisition unit 152, a date / character recognition unit 153, and a consistency determination unit 154.

The date condition DB 151a and the relationship DB 151b are stored in the storage unit 151. The date condition DB 151a is a DB in which information indicating a pair of dates to be compared and a future-to-post relationship of dates is stored for a plurality of dates included in one certificate. The relationship DB 151b is a DB in which information indicating the presence / absence of notation or the condition for changing the notation method is stored for the characters, numbers, dates, and figures included in one certificate.

Here, the date condition DB according to the first embodiment will be further described with reference to FIG. FIG. 12 is a chart for explaining the date condition DB according to the first embodiment.

As shown in FIG. 12, the date condition DB 151a has a certificate type column, two date columns, and a date condition column. Information indicating the type of certificate is described in the certificate type column. In the two date fields, the types of dates to be compared are described. Date conditions describe the relationship between two dates.

In the case of a driver's license, as already explained with reference to FIG. 3, the date of issuance, the expiration date, the date of acquisition, the year of acquisition, and the date of birth can be obtained.

The year of acquisition is the year in which the driver's license was first obtained. The acquisition date is the date on which the first license included in the corresponding category was acquired. Therefore, the year of acquisition is the same as the year described in the column of acquisition date, or the year of acquisition is earlier (past). The date of issuance is the date on which the driver's license was first issued or the date of renewal. Therefore, the acquisition date and the delivery date are the same day, or the acquisition date is earlier (past). The expiration date is set on the day after a predetermined period from the date of issuance. Therefore, the delivery date is earlier (past) than the expiration date. A driver's license is, of course, issued after the birth of the person to be proved. Therefore, the date of birth is before (past) the year of acquisition.

In the date condition DB 151a shown in FIG. 12, the above relationship is shown in the date condition column. The date described in the date (X) column is X, the date described in the date (Y) column is Y, and if X is in the past, X <Y is described, and whether X and Y are the same or not. When X is in the past, it is written as X ≦ Y. In addition, when comparing only the year without comparing the month and day, (year) is written in the date condition column. The same applies to other certificate types. The value of T in the personal number card column is 5 (when the age at the time of issuance is 19 years or younger) or 10 (when the age at the time of issuance is 20 or more). As will be described later, if the above conditions are satisfied for the two dates to be compared, it is determined to be "true", and if not, it is determined to be "false".

Next, the relationship DB according to the first embodiment will be described with reference to FIG. FIG. 13 is a chart for explaining the relationship DB according to the first embodiment.

As shown in FIG. 13, the relationship DB 151b has a certificate type column and two reference points / condition columns. Information indicating the type of certificate is described in the certificate type column. In the reference location / condition column, information indicating the reference location and information indicating the conditions relating to the description contents of the reference location are described. If both of the conditions described in the two reference points / condition columns are satisfied, it is determined as "true", and if it is not satisfied, it is determined as "false".

In the case of a driver's license, the notation of limited conditions may differ depending on the time of acquisition. For example, an AT-only ordinary car license was created on November 1, 1991, and prior to this date, the driver's license does not contain any limiting conditions indicating that it is AT-only. In addition to the limited conditions, in the case of a good driver, the background color of the expiration date will be gold and a mark indicating a good driver will be displayed. Therefore, if one of them is missing, it is probably suspected to be an art forgery. FIG. 13 shows an example of a driver's license, but the same applies to other certificates.

See FIG. 11 again. The target image acquisition unit 152 acquires the target image from the confirmation document reception unit 11. The date / character recognition unit 153 reads the date from the target image by character recognition. Further, the date / character recognition unit 153 reads the information of each reference part by character recognition for the reference part shown in the relationship DB 151b. The consistency determination unit 154 confirms that the date read by the date / character recognition unit 153 satisfies all the date conditions of the corresponding date condition DB 151a. Further, the consistency determination unit 154 confirms that the information read by the date / character recognition unit 153 satisfies all the conditions of the corresponding relationship DB 151b.

If there is a date that does not satisfy the date condition, or if the information read from the reference location does not satisfy the condition of the relationship DB 151b (hereinafter, the relationship condition), the consistency determination unit 154 determines that the target image is an art forgery. .. When it is determined that the target image is an art forgery, the consistency determination unit 154 may notify the operator of the authenticity determination device 10 of an error message. The message may include, for example, that there was a date that did not meet the date condition, or that there was information that did not meet the relationship condition. The consistency determination unit 154 notifies the comprehensive determination unit 16 of the determination result.

As an application example, the consistency determination unit 154 may read a plurality of description contents capable of identifying the same target from the target image and collate the plurality of description contents. For example, in the case of the driver's license illustrated in FIG. 3, the description contents in the entry fields 41m and 41n represent the issuer. In addition, among the addresses described in the entry field 41b, the characters indicating administrative divisions can be used to identify the issuer. In addition, the first two digits of the number described in the entry field 41h can be used to identify the issuer.

In the case of this example, the consistency determination unit 154 may specify the issuers corresponding to the entry fields 41b, 41h, 41m, and 41n, respectively, and determine whether the specified issuers are all the same. If they are all the same, it is determined that the description contents are consistent, and if at least one issuer is different, it is determined that the description contents are inconsistent. As described above, various application examples can be conceived from the configuration of the third determination unit 15 described above, and such application examples naturally belong to the technical scope of the first embodiment.

Here, with reference to FIGS. 14 and 15, the flow of the process of the relationship determination according to the first embodiment will be described. FIG. 14 is a first flow chart for explaining the process of determining the relationship according to the first embodiment. FIG. 15 is a second flow chart for explaining the process of determining the relationship according to the first embodiment.

(S131, S132) The date / character recognition unit 153 selects the record of the date condition DB 151a corresponding to the certificate type of the target image. Further, the date / character recognition unit 153 reads the date from the target image by character recognition.

(S133) The consistency determination unit 154 determines whether or not the date read by the date / character recognition unit 153 matches the date condition of the selected record. If it matches the date condition, the process proceeds to S134. On the other hand, if it does not match the date condition, the process proceeds to S141.

(S134) The third determination unit 15 determines whether or not all the records of the target date condition DB 151a have been selected. When there are a plurality of records corresponding to the certificate type of the target image and there is a record not selected by the date / character recognition unit 153 in S131, the process proceeds to S131. On the other hand, when all the target records are selected, the process proceeds to S135.

(S135) The consistency determination unit 154 determines that the target image satisfies the date condition (condition described in the date condition DB 151a), and proceeds to the process in S136.

(S136, S137) The date / character recognition unit 153 selects the record of the relationship DB 151b corresponding to the certificate type of the target image. Further, the date / character recognition unit 153 recognizes characters / numbers from the description field of the target image corresponding to the reference part of the selected record by character recognition, and reads the information described in the reference part. For example, when the reference location is the date of birth column, the date / character recognition unit 153 recognizes characters and numbers from the date of birth column of the target image and reads the date information indicating the date of birth.

(S138) The consistency determination unit 154 matches the information of the reference location read by the date / character recognition unit 153 with the two conditions (conditions A and B) corresponding to the two reference locations described in the selected record. Determine whether or not to do so. That is, the consistency determination unit 154 determines that the conditions A and B are both satisfied, or that the conditions A and B are not satisfied together, and that the conditions A and B are not satisfied, and that the conditions A and B are not satisfied. If it is determined to match, the process proceeds to S139. On the other hand, if it is determined that they do not match, the process proceeds to S141.

For example, the certificate type is a driver's license, the reference part corresponding to condition A (background color = gold) is the "expiration date" column, and the reference part corresponding to condition B (with "excellent" mark) is "license". Consider the case under the "Conditions, etc." column. In this case, when the background color in the "Expiration date" column is gold and there is an "excellent" mark under the "License conditions, etc." column, or when the background color in the "Expiration date" column is green or blue, " When there is no "excellent" mark under the "license conditions, etc." column, the consistency determination unit 154 determines that the conditions A and B are consistent. At other times, the consistency determination unit 154 determines that the conditions A and B are not consistent.

(S139) The third determination unit 15 determines whether or not all the records of the target relationship DB 151b have been selected. When there are a plurality of records corresponding to the certificate type of the target image and there is a record not selected by the date / character recognition unit 153 in S136, the process proceeds to S136. On the other hand, when all the target records are selected, the process proceeds to S140.

(S140) The consistency determination unit 154 determines that the target image satisfies the relationship condition (condition described in the relationship DB 151b), and proceeds to the process in S142. Since the process proceeds to S136 when the target image satisfies the date condition, the arrival of the process at S140 means that the target image satisfies both the date condition and the relationship condition. In this case, the consistency determination unit 154 notifies the comprehensive determination unit 16 that the target image is “true” as a result of the relationship determination.

(S141) The consistency determination unit 154 determines that the target image is an art forgery, and proceeds to the process of S142. If the target image does not satisfy any of the date conditions, or if the target image does not satisfy any of the relationship conditions, the process proceeds to S141. Therefore, the fact that the process has reached S141 indicates that the target image has reached S141. It means that the described information is inconsistent. In this case, the consistency determination unit 154 notifies the comprehensive determination unit 16 that the target image is “false” as a result of the relationship determination.

(S142) The third determination unit 15 determines whether or not the processing (relationship determination) after S131 has been executed for all the target images. When there are a plurality of target images to be determined, if the processing for the undetermined target image is not completed, the processing proceeds to S131, and the processing after S131 is executed for the undetermined target image. On the other hand, when the processing is completed for all the target images, the series of processing shown in FIGS. 14 and 15 ends.

The processing groups of S131 to S135 related to the determination of the date condition and the processing groups of S136 to S140 related to the relational condition may be executed in different order, or these two processing groups may be executed in parallel. Good.

[1-5. Comprehensive judgment unit]
Next, the function of the comprehensive determination unit according to the first embodiment will be described with reference to FIG. FIG. 16 is a block diagram for explaining the function of the comprehensive determination unit according to the first embodiment.

As shown in FIG. 16, the comprehensive determination unit 16 includes a storage unit 161, a determination result acquisition unit 162, a determination processing unit 163, and a message output unit 164.

The storage unit 161 stores the determination mode DB 161a and the threshold value DB 161b. The determination mode DB 161a is a DB in which information regarding a combination of determination processes used for authenticity determination of the target image is stored. The threshold value DB 161b is a DB in which information on the threshold value and the determination condition used when determining the authenticity of the target image is stored.

Here, the determination mode DB according to the first embodiment will be described with reference to FIG. FIG. 17 is a chart for explaining the determination mode DB according to the first embodiment.

As shown in FIG. 17, the determination mode DB 161a contains information regarding a plurality of determination modes. In the example of FIG. 17, it corresponds to a combination capable of black-painted image determination corresponding to the first determination unit 13, glyph image determination corresponding to the second determination unit 14, and relationship determination corresponding to the third determination unit 15. Seven determination modes are shown.

Mode 1 is a determination mode that utilizes all the determination results of the black-painted image determination, the glyph image determination, and the relationship determination. When the determination mode is set to mode 1, the comprehensive determination unit 16 comprehensively considers all the determination results output from the first determination unit 13, the second determination unit 14, and the third determination unit 15. To determine the authenticity of the target image. Mode 1 is a determination mode for making the most rigorous determination.

Mode 2 is a determination mode that uses only the determination results of the black-painted image determination and the relationship determination. When the determination mode is set to mode 2, the comprehensive determination unit 16 determines the authenticity of the target image by comprehensively considering the determination results output from the first determination unit 13 and the third determination unit 15. .. Mode 2 is a highly reliable determination mode because it combines the determination results of the relationship determination that detects a logical contradiction and makes a determination.

Mode 3 is a determination mode that uses only the determination results of the glyph image determination and the relationship determination. When the determination mode is set to mode 3, the comprehensive determination unit 16 determines the authenticity of the target image by comprehensively considering the determination results output from the second determination unit 14 and the third determination unit 15. .. Mode 3 is also a highly reliable determination mode because it combines the determination results of the relationship determination that detects a logical contradiction and makes a determination.

Mode 4 is a determination mode that uses only the determination results of the black-painted image determination and the glyph image determination. When the determination mode is set to mode 4, the comprehensive determination unit 16 determines the authenticity of the target image by comprehensively considering the determination results output from the first determination unit 13 and the second determination unit 14. .. The mode 4 is a determination mode applicable to a certificate in which the relationship between the described contents is unknown, and is a determination mode in which sufficient reliability is ensured by combining two types of image determinations.

Mode 5 is a determination mode that uses only the determination result of the black-painted image determination. When the determination mode is set to mode 5, the comprehensive determination unit 16 determines the authenticity of the target image based on the determination result output from the first determination unit 13. Mode 5 is a determination mode in which processing load is low and high-speed processing is possible because only one type of determination result is used. Further, as in mode 4, it can be applied to a certificate in which the relationship between the described contents is unknown.

Mode 6 is a determination mode that uses only the determination result of the glyph image determination. When the determination mode is set to mode 6, the comprehensive determination unit 16 determines the authenticity of the target image based on the determination result output from the second determination unit 14. Mode 6 is a determination mode in which processing load is low and high-speed processing is possible because only one type of determination result is used. Further, as in mode 4, it can be applied to a certificate in which the relationship between the described contents is unknown.

Mode 7 is a determination mode that uses only the determination result of the relationship determination. When the determination mode is set to mode 7, the comprehensive determination unit 16 determines the authenticity of the target image based on the determination result output from the third determination unit 15. Mode 7 is a determination mode in which processing load is low and high-speed processing is possible because only one type of determination result is used. In addition, since a logical contradiction is detected and a judgment is made, it is possible to distinguish an art forgery with a high degree of certainty.

The determination mode may be set in advance, or may be set manually or automatically depending on the type of certificate and the type of procedure. For example, mode 1 is selected when the degree of similarity between the photograph of the certificate image received from the user device 30 and the photograph of the person who is separately received is less than a predetermined value, and mode 2 is selected when the degree of similarity is greater than or equal to the predetermined value. Any one of ~ 4 may be selected. At this time, mode 4 may be selected in the case of a certificate whose relationship between the described contents is unknown. Further, when a plurality of certificate images are provided, mode 1 is selected for the determination of one certificate image, and one of modes 2 to 7 is selected for the determination of the other certificate image. You may.

Next, the threshold value DB according to the first embodiment will be described with reference to FIG. FIG. 18 is a chart for explaining the threshold value DB according to the first embodiment.

As shown in FIG. 18, the threshold value DB 161b describes the determination conditions regarding the certainty degree C1 output from the first determination unit 13 and the certainty degree C2 output from the second determination unit 14. In the example of FIG. 18, in addition to the judgment conditions for making a judgment using the certainty levels C1 and C2 individually, the judgment conditions for making a judgment using the evaluation values based on the certainty levels C1 and C2 are shown. .. Further, in the example of FIG. 18, in addition to the determination condition, a condition (visual inspection condition) for outputting a message requesting a visual inspection by the operator is shown. The judgment condition and the inspection condition using the evaluation value may be omitted.

In the determination condition for determining by using the certainty degrees C1 and C2 individually, the certainty degree C1 is compared with the threshold value TH1. When C1 ≧ TH1, the result of the black-painted image determination is “true”. When the certainty C1 is represented by a value in the range 0 to 1, the threshold TH1 may be set to, for example, 0.9 (corresponding to 90% certainty). Similarly, the confidence C2 is compared to the threshold TH2. When C2 ≧ TH2, the result of the glyph image determination is “true”. When the certainty C2 is represented by a value in the range 0 to 1, the threshold TH2 may be set to, for example, 0.9 (corresponding to 90% certainty).

When setting the inspection condition, the condition TH1a <C1 <TH1 can be set for the certainty degree C1. The threshold value TH1a is set to a value slightly smaller than the threshold value TH1. For example, when the threshold TH1 is 0.9, the threshold TH1a is set to 0.85. Similarly, for the certainty C2, the condition TH2a <C2 <TH2 can be set. The threshold value TH2a is set to a value slightly smaller than the threshold value TH2. For example, when the threshold TH2 is 0.9, the threshold TH2a is set to 0.85.

When the image determination is combined, there is a possibility that low confidence levels C1 and C2 are output due to the sharpness of the certificate image and various noises. Therefore, when the certainty levels C1 and C2, which are slightly smaller than the threshold values TH1 and TH2, are output, a visual inspection is requested to avoid the risk that the genuine certificate image is finally judged as a forgery. be able to. Instead of requesting the operator for visual inspection, a mechanism may be adopted in which a message requesting re-sending of the certificate image is sent to the user device 30 when the inspection conditions are satisfied.

When making a judgment using the evaluation values based on the certainty levels C1 and C2, in the example of FIG. 18, it is required that the certainty levels C1 and C2 are equal to or higher than the threshold values TH1b and TH2b, respectively, as a premise. The threshold values TH1b and TH2b may be equal to or lower than the threshold values TH1 and TH2, respectively. FIG. 18 shows a determination condition using the product of the confidence levels C1 and C2 as the evaluation value (see the following equation (1)) and a determination condition using the weight average of the confidence levels C1 and C2 as the evaluation value (see the following). (See Equation (2)) and.

C1 * C2 ≧ TH3
… (1)

(K1 * C1 + K2 * C2) / (K1 + K2) ≧ TH3
… (2)
However, K1> 0 and K2> 0.

When the judgment conditions given by the above formula (1) are adopted, when the conditions of C1 ≧ TH1b, C2 ≧ TH2b, and the formula (1) are satisfied, the judgment result based on the black-painted image judgment and the glyph image judgment is ". It becomes "true". When the judgment conditions given by the above formula (2) are adopted, when the conditions of C1 ≧ TH1b, C2 ≧ TH2b, and the formula (2) are satisfied, the judgment result based on the black-painted image judgment and the glyph image judgment is ". It becomes "true".

The inspection conditions may also be set when the determination conditions of the above formula (1) or formula (2) are adopted. In this case, the inspection condition when the determination condition of the equation (1) is adopted is given by the following equation (3), and the inspection condition when the determination condition of the equation (2) is adopted is the following equation (4). ). The threshold values TH3a and TH3b are set to values slightly smaller than the threshold values TH3. For example, when the threshold value TH3 is 0.9, the threshold values TH3a and TH3b are set to 0.85. In addition, when the inspection condition is satisfied, a mechanism may be adopted in which a message requesting re-sending of the certificate image is sent to the user device 30 instead of a request for visual inspection.

TH3a <C1 * C2 <TH3
… (3)

TH3b <(K1 * C1 + K2 * C2) / (K1 + K2) <TH3
… (4)

See FIG. 16 again. The determination result acquisition unit 162 acquires the certainty degree C1 from the first judgment unit 13, the certainty degree C2 from the second judgment unit 14, and the relationship determination result (“true” / “” from the third judgment unit 15. Get fake "). The determination processing unit 163 makes a final authenticity determination for the target image by using all or selectively the results of the black-painted image determination, the glyph image determination, and the relationship determination according to the set determination mode. ..

For example, in the case of mode 1, the determination processing unit 163 first refers to the result of the relationship determination, and if the result of the relationship determination is "false", determines that the target image is an art forgery. When the result of the relationship determination is "true", the determination processing unit 163 refers to the determination condition of the threshold value DB161b and confirms that the certainty degrees C1 and C2 satisfy all the target determination conditions.

For example, when the determination is made by using the certainty degrees C1 and C2 individually, the determination processing unit 163 confirms that the certainty degree C1 satisfies C1 ≧ TH1 and the certainty degree C2 satisfies C2 ≧ TH2. If any of the determination conditions is not satisfied, the determination processing unit 163 determines that the target image is an art forgery. On the other hand, when all the determination conditions are satisfied, the determination processing unit 163 determines that the target image is genuine.

Further, when the evaluation value based on the confidence levels C1 and C2 (either of the above equations (1) and (2)) is used for determination, the determination processing unit 163 satisfies that the confidence level C1 satisfies C1 ≧ TH1b. In addition, it is confirmed that the confidence level C2 satisfies C2 ≧ TH2b, and the condition of the target equation in the above equation (1) or equation (2) is satisfied. If at least one condition is not satisfied, the determination processing unit 163 determines that the target image is an art forgery. On the other hand, when all the conditions are satisfied, the determination processing unit 163 determines that the target image is genuine.

When the inspection condition is satisfied, the determination processing unit 163 may output a visual inspection request message to the operator or a re-send message of the certificate image to the user device 30 via the message output unit 164.

The message output unit 164 outputs a message to be transmitted to the user device 30 according to the determination result by the determination processing unit 163. For example, the message output unit 164 sends a message notifying that the procedure cannot be continued or a message prompting the user to provide the correct certificate image when the target image is determined to be forgery in the final authenticity determination. Output. These messages are sent to the user device 30 via the confirmation document receiving unit 11.

Here, the flow of the process of the comprehensive determination according to the first embodiment will be described with reference to FIGS. 19 and 20. FIG. 19 is a first flow chart for explaining the processing of the comprehensive determination according to the first embodiment. FIG. 20 is a second flow chart for explaining the processing of the comprehensive determination according to the first embodiment.

(S151) The determination processing unit 163 refers to the determination mode DB 161a and determines the determination process to be executed according to the setting of the determination mode.

When the determination mode is mode 1, all the processes after S152 are executed. When the determination mode is mode 2, the processes of S156 to S158 relating to the certainty C2 are skipped. When the determination mode is mode 3, the processes of S153 to S155 relating to the certainty C1 are skipped. When the determination mode is mode 4, the process of S152 regarding the result of the relationship determination is skipped.

When the determination mode is mode 5, the processing of S152, S156 to S158 regarding the certainty degree C2 and the result of the relationship determination is skipped. When the determination mode is mode 6, the processing of S152, S153 to S155 regarding the certainty degree C1 and the result of the relationship determination is skipped. When the determination mode is mode 7, the processes of S153 to S161 relating to the certainty degrees C1 and C2 are skipped.

(S152) The determination processing unit 163 determines whether or not the date condition and the relationship condition are satisfied based on the result of the relationship determination output from the third determination unit 15. If the result of the relationship determination is "true", the date condition and the relationship condition are satisfied, and if the result of the relationship determination is "false", at least one of the date condition and the relationship condition is satisfied. Not satisfied. If the date condition and the relationship condition are satisfied, the process proceeds to S153. On the other hand, if the date condition and the relationship condition are not satisfied, the process proceeds to S163.

(S153) The determination processing unit 163 determines whether or not the condition of C1 ≧ TH1 is satisfied. If the condition of C1 ≧ TH1 is satisfied, the process proceeds to S156. On the other hand, if this condition is not satisfied, the process proceeds to S154.

(S154) The determination processing unit 163 determines whether or not the condition of C1> TH1a is satisfied. When the process reaches S154, the condition of C1 <TH1 is satisfied, and in S154, the inspection condition for the certainty degree C1 shown in FIG. 18 is substantially determined. If the condition C1> TH1a is satisfied, the process proceeds to S155. On the other hand, if this condition is not satisfied, the process proceeds to S163.

(S155) The determination processing unit 163 sets a flag of “visual inspection required”. A "resend required" flag for requesting the user device 30 to resend the certificate image may be set instead.

(S156) The determination processing unit 163 determines whether or not the condition of C2 ≧ TH2 is satisfied. If the condition of C2 ≧ TH2 is satisfied, the process proceeds to S159. On the other hand, if this condition is not satisfied, the process proceeds to S157.

(S157) The determination processing unit 163 determines whether or not the condition of C2> TH2a is satisfied. When the process reaches S157, the condition of C2 <TH2 is satisfied, and in S157, the inspection condition for the certainty degree C2 shown in FIG. 18 is substantially determined. If the condition C2> TH2a is satisfied, the process proceeds to S158. On the other hand, if this condition is not satisfied, the process proceeds to S163.

(S158) The determination processing unit 163 sets a flag of “visual inspection required”. A "resend required" flag for requesting the user device 30 to resend the certificate image may be set instead.

(S159) The determination processing unit 163 determines whether or not there is a flag setting of "visual inspection required". If the "resend required" flag is set, it is determined whether or not the flag is set. If there is a flag setting, the process proceeds to S160. On the other hand, if there is no flag setting, the process proceeds to S162.

(S160) The determination processing unit 163 outputs a message requesting the operator to perform a visual inspection via the message output unit 164. Then, the determination processing unit 163 receives the execution result of the visual inspection by the operator. When the "re-send required" flag is set, the determination processing unit 163 sends a message requesting the user device 30 to re-send the certificate image via the message output unit 164, and FIGS. 19 and 20 show. The series of processing shown in is completed.

(S161) The determination processing unit 163 confirms whether or not the target image is determined to be an art forgery by visual inspection. If the target image is determined to be forgery by visual inspection, the process proceeds to S163. On the other hand, when the target image is determined to be genuine by the visual inspection, the process proceeds to S162.

(S162) The determination processing unit 163 determines that the target image is “genuine”. In this case, the confirmation document reception unit 11 proceeds with the processing related to the applied procedure. Further, the series of processes shown in FIGS. 19 and 20 is completed.

(S163) The determination processing unit 163 determines that the target image is “art forgery”. In this case, the processing related to the applied procedure is interrupted, and a message notifying that the procedure cannot be continued and a message prompting to provide the correct certificate image are sent to the user device 30 via the confirmation document reception unit 11. Be done. Further, the series of processes shown in FIGS. 19 and 20 is completed.

The examples of FIGS. 19 and 20 assume a method of individually using the above-mentioned confidence levels C1 and C2, but when applying a method of using evaluation values based on the certainty levels C1 and C2, , S156 to S158 are omitted, the determination condition of S153 is replaced with the above formula (1) or formula (2), and the inspection condition of S154 is replaced with the above formula (3) or formula (4). .. Further, the determination mode is set to mode 1 or mode 4.

Further, the process related to the inspection condition may be omitted. Further, the processing groups of S153 to S155 relating to the determination of the certainty C1 and the processing groups of S156 to S158 relating to the determination of the certainty C2 may be executed in different order, or these two processing groups may be executed in parallel. You may.

[1-6. Modification example]
Next, a modified example of the above-described embodiment will be described.

(About transformation of black-painted image judgment)
In the above-mentioned explanation of the black-painted image determination, a method of masking the information mainly dependent on the proved person while leaving the information about the issuer is described. The masking method according to the first embodiment is not limited to this, and for example, a method of masking all other characters and numbers while leaving information and a border about the issuer (hereinafter, method A) may be applied. Different issuers may have different printers or printing machines that print the certificate, which can lead to different border notations. While it is difficult for humans to recognize such differences, AI is good at accurately recognizing graphic patterns that do not include letters and numbers. Therefore, even when the method A is applied, a pattern recognizer that can be used for determining the authenticity of the certificate can be generated.

Further, in the above explanation, the template image was trained by AI to generate a pattern recognizer, but information on the delivery date or delivery time corresponding to each template image is given to AI as teacher data, and by supervised learning. A method for generating a pattern recognizer (hereinafter referred to as method B) may be applied. In this case, when determining the target image, information regarding the delivery date or delivery time read from the target image by character recognition is input to the pattern recognizer. In the case of the method B, since the template images are accurately classified with respect to the delivery date or the delivery time, the recognition accuracy may be improved.

In the above method B, information on the delivery date or delivery time is given to the AI as teacher data, and further, information on the issuer (information on the delivery location) is given to the AI as teacher data to generate a pattern recognizer (hereinafter,). , Method C) may be applied. In this case, when determining the target image, information on the delivery date or delivery time read from the target image by character recognition and information on the issuer (information on the delivery location) are input to the pattern recognizer. In the case of method C, since the template image is accurately classified by the delivery date or delivery time and the issuer (delivery place), the recognition accuracy may be improved.

The part to be masked as in the above method A can be changed as appropriate. Further, as in the above methods B and C, information readable by character recognition can be given to AI as teacher data, and a pattern recognizer can be generated by supervised learning. Naturally, such a modification also belongs to the technical scope of the first embodiment.

(About the transformation of glyph image judgment)
In the above description of the glyph image determination, the glyph image DB 141a is used to evaluate the similarity of the glyphs, but the glyph image determination may be performed using a pattern recognizer. For example, by preparing a template image that leaves only the specified area of the certificate including the delivery date, issuer (delivery place), and predetermined characters, and letting AI learn the template image, the delivery date and issuer (delivery place). ), A pattern recognizer that recognizes the character shape of a predetermined character can be generated. By using this pattern recognizer, it is possible to determine a glyph image.

Also, in the glyph image determination, a method of generating a pattern recognizer by giving information about the issue date or time and the issuer (delivery place) to AI as teacher data and letting AI learn the glyph of a predetermined character ( Hereinafter, method D) can be applied. In this case, when determining the target image, information regarding the delivery date or delivery time read from the target image by character recognition is input to the pattern recognizer. In the case of the method D, since the template images are accurately classified with respect to the delivery date or the delivery time, the recognition accuracy can be improved. Naturally, such a modification also belongs to the technical scope of the first embodiment.

(About deformation of authenticity judgment device)
In the above description, the authenticity determination device 10 including all of the first determination unit 13, the second determination unit 14, and the third determination unit 15 has been illustrated, but a modification in which a part thereof is omitted is possible.

The authenticity determination device 10 omitting the first determination unit 13 performs at least one of the glyph image determination and the relationship determination. The authenticity determination device 10 omitting the second determination unit 14 performs at least one of the black-painted image determination and the relationship determination. The authenticity determination device 10 omitting the third determination unit 15 performs at least one of the black-painted image determination and the glyph image determination. The authenticity determination device 10 omitting the first determination unit 13 and the second determination unit 14 performs the relationship determination. The authenticity determination device 10 omitting the first determination unit 13 and the third determination unit 15 performs the glyph image determination. The authenticity determination device 10 omitting the second determination unit 14 and the third determination unit 15 performs black-painted image determination.

As described above, the form of the authenticity determination device 10 can be appropriately deformed, and such a modification naturally belongs to the technical scope of the first embodiment.

(Transformation of system)
In the system shown in FIG. 1, the authenticity determination device 10 and the determination device generator 20 are separate devices, but the authenticity determination device 10 and the determination device generator 20 are the same computer or one computer or the same device generator 20. It may be configured by one distributed processing system in which a plurality of computers are connected. Further, the first determination unit 13, the second determination unit 14, the third determination unit 15, and the comprehensive determination unit 16 may be configured by separate computers, or a part of them may be a single computer. It may be configured and the rest may be configured on another computer. Further, the function of each storage unit included in the authenticity determination device 10 and the determination device generation device 20 may be realized by the network NW or cloud storage on another network.

As described above, the form of the system according to the first embodiment can be appropriately modified, and such a modified example naturally belongs to the technical scope of the first embodiment.

<2. Second Embodiment>
Next, a second embodiment of the present disclosure will be described. It should be noted that detailed description of the portion overlapping with the above-described first embodiment will be omitted.

(Black-painted image judgment)
First, the black-painted image determination according to the second embodiment will be described. The black-painted image determination according to the second embodiment is different from the first embodiment described above, in that each template image that can be generated from the certificate image group 22a using the mask image 22b and the black that is generated from the target image. It is compared with the filled image and performed based on the comparison result.

In the second embodiment, the first determination unit 13 refers to the template image DB231a described later and uses a pattern recognizer for determining the similarity between the images to resemble the black-painted image and each template image. Determine the degree. The template image DB 231a may be generated by the determination device generation device 20, or may be stored in advance in the storage unit 131 of the first determination unit 13. Further, it should be noted that the pattern recognizer according to the second embodiment is different from the pattern recognizer 131b according to the first embodiment described above.

The template image DB 231a contains, for example, information as shown in FIG. FIG. 21 is a chart for explaining the template image DB according to the second embodiment of the present disclosure. As shown in FIG. 21, the template image DB 231a stores the issuer, conditions, delivery date, and image data of the template image (tP190410_01a, etc.) in association with each other. In the example of FIG. 21, a plurality of template images are associated with one combination of issuer, condition, and delivery date, but one template image is associated with one combination. May be good. Further, the conditions described in the condition column may be various conditions included in the relationship DB 151b illustrated in FIG. For example, conditions such as whether or not the driver is a good driver and whether or not the driver is a "quasi-medium-sized vehicle" license type can be the target. Further, in the template image DB 231a, a plurality of template images are associated with each set of the issuer, the condition, and the delivery date.

The first determination unit 13 according to the second embodiment executes the process according to the flow illustrated in FIG. FIG. 22 is a flow chart for explaining the black-painted image determination process according to the second embodiment of the present disclosure.

(S201) The masking unit 133 masks the target image using the mask image 131a to generate a black-painted image.

(S202, S206) The black-painted image determination unit 134 changes the parameter p for specifying the template image included in the template image DB231a from 1 to NP (NP is the number of template images), and is between S202 and S206. Repeat the process. However, when the condition of S205 is satisfied, the process proceeds to S208 in the middle of the iterative process.

(S203) The black-painted image determination unit 134 reads out the p-th template image from the template image DB 231a.

(S204) The black-painted image determination unit 134 calculates the degree of similarity between the black-painted image and the read template image by using the pattern recognizer. The pattern recognizer used here may be a determination device that determines the degree of similarity between images based on image features such as BRIEF, ORB, SIFT, and SURF. In this case, the black-painted image determination unit 134 inputs the black-painted image and the template image to the determination device, and acquires the similarity output from the determination device.

(S205) The black-painted image determination unit 134 determines whether or not the similarity calculated in S204 is equal to or greater than a predetermined threshold value. For example, when the similarity is standardized so that the similarity is 1 when the images to be compared are the same and 0 when they are completely different, the predetermined threshold value is a value within the range of 0 to 1. For example, it is set to 0.9). If the similarity is equal to or greater than a predetermined threshold value, the process proceeds to S208. On the other hand, if the similarity is not equal to or higher than a predetermined threshold value, the process proceeds to S206. When the process proceeds to S206, the process returns to S202 when p is less than NP, and proceeds to S207 when p is NP.

(S207) The black-painted image determination unit 134 sets the result of the black-painted image determination as false and provides the determination result to the comprehensive determination unit 16. In the example of FIG. 22, when the black-painted image is not similar to any of the template images included in the template image DB231a, the process reaches S207. Then, when the process of S207 is reached, the determination result is false. When the process of S207 is completed, the process proceeds to S209.

(S208) The black-painted image determination unit 134 makes the result of the black-painted image determination true, and provides the determination result to the comprehensive determination unit 16. In the example of FIG. 22, when the black-painted image is similar to any of the template images included in the template image DB231a (when the similarity is equal to or more than the threshold value), the process reaches S208. Then, when the process of S208 is reached, the determination result is regarded as true.

(S209) The first determination unit 13 determines whether or not the above-mentioned black-painted image determination has been performed for all the target images. If there is an unprocessed target image, the process proceeds to S201. On the other hand, when the black-painted image determination is completed for all the target images, the series of processes shown in FIG. 22 is completed.

As described above, in the second embodiment, all the template images prepared in advance in the template image DB231a and the black-painted image are collated, and sufficient similarity between any of the template images can be confirmed. In this case, the result of the black-painted image determination is true.

When the template image corresponding to the black-painted image is specified as a result of the black-painted image determination, the issuer, the condition, and the set of the delivery date corresponding to the template image can be specified from the contents of the template image DB231a. The specified issuer, conditions, and date of issue can be used for other judgments. For example, by confirming the consistency between the issuer, condition, and issue date described in the target image and the issuer, condition, and issue date specified based on the template image DB231a, the accuracy of the authenticity determination can be improved. Can be enhanced.

(Glyph image judgment)
Next, the glyph image determination according to the second embodiment will be described. In the glyph image determination according to the second embodiment, unlike the glyph image determination according to the first embodiment described above, the glyph image DB241a, which will be described later, is used without specifying the issuer and the delivery time from the target image. Judgment is made based on the similarity of images. The glyph image DB 241a contains, for example, information as shown in FIG. 23. FIG. 23 is a chart for explaining the glyph image DB according to the second embodiment of the present disclosure.

In the example of FIG. 23, the glyph image DB 241a according to the second embodiment stores the issuer, the condition, the date of delivery, the description location, and the image data of the glyph image (tC190410_01a, etc.) in association with each other. The conditions described in the condition column may be various conditions included in the relationship DB 151b illustrated in FIG. For example, conditions such as whether or not the driver is a good driver and whether or not the driver is a "quasi-medium-sized vehicle" license type can be the target. Further, in the glyph image DB 241a, a plurality of glyph images are associated with each set of the issuer, the condition, and the delivery date.

In the entry field of the certificate, specific characters and numbers representing the era name, date, etc. are used. Therefore, for the characters / numbers that can be described in each entry column, a plurality of glyph images corresponding to each set of the issuer, the condition, and the date of issuance are prepared and stored in the glyph image DB 241a. The glyph image stored in the glyph image DB241a is compared with each glyph image extracted from the target image, and if all the glyph images are sufficiently similar, the target image can be presumed to be true. In the glyph image determination according to the second embodiment, the determination is performed using this mechanism. In the second embodiment, the character recognition unit 143 of the second determination unit 14 may be omitted.

Hereinafter, the process of determining the glyph image according to the second embodiment will be further described with reference to FIGS. 24 and 25. FIG. 24 is a first flow chart for explaining the process of determining the glyph image according to the second embodiment of the present disclosure. FIG. 25 is a second flow chart for explaining the process of determining the glyph image according to the second embodiment of the present disclosure. It is assumed that the glyph image DB 241a is stored in the storage unit 141 of the second determination unit 14.

(S211 and S218) The processing between S211 and S218 is executed for each of the target description points. The target description locations are all the description locations described in the glyph image DB 241a. In S211 the description points are sequentially selected, and the processes up to S218 are executed for the selected description points. For example, the process between S211 and S218 is executed for the "expiration date column", the process between S211 and S218 is executed for the "date of birth column", and the other described parts are similarly processed in sequence. Will be executed.

(S212, S217) The process between S212 and S217 is executed for each glyph image of the glyph image DB241a corresponding to the description location selected in S211. In S212, the parameter q (q = 1 to NQ; NQ is the number of target glyph images) for specifying the target glyph image is sequentially selected, and each value of the parameter q is between S212 and S217. The process is executed. That is, all the glyph images of characters / numbers that can be described in the selected description location are collated with the glyph images cut out from the description fields of the corresponding target images.

(S213) The glyph determination unit 144 extracts the glyph image from the corresponding description location of the target image. In the following, for the sake of simplicity, the glyph of the glyph image extracted from the glyph image may be referred to as a "glyph".

(S214) The glyph determination unit 144 acquires all the glyph images that can be used in the corresponding description location from the glyph image DB 241a. In the following, for the sake of simplicity, the glyph of the glyph image acquired from the glyph image DB241a may be referred to as a “reference glyph”.

(S215) The glyph determination unit 144 inputs each of the glyph image extracted from the target image and the glyph image acquired from the glyph image DB241a into the similarity determination device, and between the target glyph and each reference glyph. Calculate the similarity of. That is, in the second embodiment, all the reference glyphs and the target glyphs are compared. For example, in the case of the driver's license shown in FIG. 3, a string of numbers in which numbers 0 to 9 are arranged is described in the entry field 41h. In this case, the reference glyph is a glyph of 0 to 9 that can be described in the description column 41h, and the glyph determination unit 144 is extracted from all the glyphs (reference glyph) of 0 to 9 and the same description column of the target image. Compare the glyphs of each number (target glyphs).

(S216) The glyph determination unit 144 compares the degree of similarity between the target glyph and each reference glyph with a predetermined threshold value. When there are a plurality of target glyphs and there are a plurality of reference glyphs to be compared, it is first determined whether or not there is a reference glyph whose similarity is equal to or higher than the threshold value for one target glyph (determination A). Further, from the results of the determination A for all the target glyphs, it is determined that there is a reference glyph whose similarity is equal to or higher than the threshold value (determination B).

When it is determined in the above determination B that there is a reference glyph whose similarity is equal to or higher than the threshold value for all the target glyphs, the character shape determination unit 144 determines that the similarity equal to or higher than the threshold value is obtained for each target glyph. When the similarity equal to or higher than the threshold value is obtained for each target glyph, the process proceeds to S217. When the process proceeds to S217, if the parameter q is less than NQ, the process returns to S212, and if the parameter q is NQ, the process proceeds to S218. Further, when the processes of S212 to S217 are completed for all the described parts, the process proceeds to S219. On the other hand, if no similarity equal to or higher than the threshold value is obtained for any of the target glyphs, the process proceeds to S220.

(S219) The glyph determination unit 144 makes the result of the glyph image determination true, and provides the determination result to the comprehensive determination unit 16. In the examples of FIGS. 24 and 25, when the existence of a similar reference glyph for each of the target glyphs is confirmed for one description location, and further, the processing reaches S219 when the confirmation is taken for all the description locations. To do. Then, when the process of S219 is reached, the determination result is regarded as true. When the process of S219 is completed, the process proceeds to S221.

(S220) The glyph determination unit 144 sets the result of the glyph image determination as false and provides the determination result to the comprehensive determination unit 16. In the examples of FIGS. 24 and 25, when the existence of a similar reference glyph for each of the target glyphs cannot be confirmed for any of the description locations, the process reaches S220. Then, when the process of S220 is reached, the determination result is false.

(S221) The second determination unit 14 determines whether or not the above-mentioned glyph image determination has been performed for all the target images. If there is an unprocessed target image, the process proceeds to S211. On the other hand, when the glyph image determination is completed for all the target images, the series of processes shown in FIGS. 24 and 25 is completed.

As described above, in the second embodiment, the glyphs (reference glyphs) of all the glyph images prepared in advance in the glyph image DB 241a and the glyphs (target glyphs) of the target image are collated with each other at each description location. When sufficient similarity between the target glyph and any of the reference glyphs can be confirmed, the result of the glyph image determination is true. Even if the result of the glyph image determination is false, it is possible to specify the description location including the glyph image of the target glyph that does not have the corresponding reference glyph, so specify which description column is forged. Can be done. In addition, it can be presumed that the description contents are correct for other description points.

The black-painted image determination and the glyph image determination according to the second embodiment have been described above. Since the elements other than the black-painted image determination and the glyph image determination are the same as those in the first embodiment described above, detailed description thereof will be omitted here. In addition, only one of the black-painted image determination and the glyph image determination according to the second embodiment may be modified so as to replace the corresponding element of the first embodiment, and various types described in the description of the first embodiment may be modified. Transformations may be applied. In this way, the technical elements of the first and second embodiments may be arbitrarily combined, or may be further modified. Such modifications also naturally belong to the technical scope of the embodiments according to the present disclosure.

<3. Addendum>
Hereinafter, additional notes will be disclosed with respect to the technical matters relating to the embodiments of the present disclosure. The contents of the appendices shown below are not intended to limit the technical scope of the present disclosure, and of course, modifications, applications, and specific examples that can be conceived by those skilled in the art based on the contents of the present disclosure are also included in the present disclosure. It belongs to the technical scope.

(Appendix 1) Obtaining an image of the certificate that is the target of authenticity judgment,
Analyzing the acquired image of the certificate, identifying the issuer and the issue time, and extracting the image of the characters or numbers in the predetermined entry field.
The image of the character or number extracted from the image of the certificate obtained by referring to the database that associates the issuer, the time of issuance, and the image of the character or number in the predetermined entry field, and the database. An authenticity determination method in which a computer executes a process including determining the authenticity of an image of the acquired certificate by using a determination device for determining the similarity of the characters or numbers extracted from the image.

(Appendix 1A) Obtaining an image of a certificate that is subject to authenticity judgment,
Analyzing the acquired image of the certificate, identifying the issuer and the issue time, and extracting the image of the characters or numbers in the predetermined entry field.
For a plurality of sample images prepared in advance for the certificate, the issuer and the issue time are given as teacher data, and a judgment device obtained by pattern learning the image of characters or numbers in the predetermined entry field is specified. An authenticity determination method in which a computer executes a process including determining the authenticity of the acquired image of the certificate by using the issued issuer and issuance time and the extracted image of the characters or numbers. ..

(Appendix 1B) Obtaining an image of the certificate that is the target of authenticity judgment,
Analyzing the acquired image of the certificate to extract an area containing information that can be used to identify the issuer and the issue time, and an image of characters or numbers in a predetermined entry field.
A determination device obtained by pattern-learning an image of the region and an image of characters or numbers in the predetermined entry field for a plurality of sample images of the certificate, an extracted image of the region, and the above. An authenticity determination method in which a computer executes a process including determining the authenticity of an image of the obtained certificate using an image of characters or numbers.

(Appendix 2) Analyzing the acquired image of the certificate and acquiring a plurality of dates from a plurality of first entry fields in which the dates are described.
With reference to the first information indicating the future relationship of the dates described in the plurality of first entry fields, whether the acquired future relationship of the plurality of dates is consistent with the future relationship indicated by the first information. It further includes determining whether or not the image of the obtained certificate is forgery if it is confirmed that the information is inconsistent.
The authenticity determination method according to Appendix 1, 1A and 1B.

(Appendix 3) Analyzing the acquired image of the certificate to detect the predetermined description content that differs depending on the issuance time among the description contents of the matters certified by the certificate.
It is confirmed whether or not the specified issuance time and the detection result for the predetermined description are consistent, and if it is confirmed that they are not consistent, the obtained image of the certificate is an art forgery. Further including determining that there is,
The authenticity determination method according to Appendix 1, 1A, 1B or 2.

(Appendix 4) The acquired image of the certificate is analyzed, and the characters or numbers are extracted from each of the plurality of second entry fields in which the characters or numbers that can be used to identify the issuer are described. It is confirmed whether or not the issuers corresponding to the plurality of second entry fields match each other, and if it is confirmed that they do not match, it is determined that the obtained image of the certificate is fake. Including,
The authenticity determination method according to any one of Supplementary notes 1 to 3.

(Appendix 5) The acquisition unit that acquires the image of the certificate that is the target of authenticity judgment,
An analysis unit that analyzes the acquired image of the certificate, identifies the issuer, the issue time, and the characters or numbers in the predetermined entry fields, and extracts the image of the characters or numbers.
The image of the characters or numbers extracted from the acquired image of the certificate by referring to the database that associates the issuer, the time of issuance, and the image of the characters or numbers in the predetermined entry field, and the database. An information processing apparatus having a determination unit for determining the authenticity of an image of the acquired certificate by using a determination device for determining the similarity of the character or number image extracted from the image.

(Supplementary Note 6) The information processing apparatus according to Supplementary Note 5, further executing the process according to the authenticity determination method according to any one of Supplementary notes 1 to 4.

(Appendix 7) A program for causing a computer to further execute the process according to the authenticity determination method according to any one of the appendices 1 to 4.

(Appendix 8) A non-transitory computer readable storage medium in which the program described in Appendix 7 is stored.

(Appendix 9) Obtaining an image of the certificate that is the target of authenticity judgment,
Mask the acquired image of the certificate by using a mask image that leaves the first part where information about the issuer is described and at least hides the second part whose contents are different for each proved person. Generating an image for judgment and
Obtained by using a first determination device obtained by pattern learning an image group masked with the mask image on a sample image of the certificate prepared in advance, and the determination image. An authenticity determination method in which a computer executes a process including determining the authenticity of an image of the certificate.

(Appendix 10) Extracting the image of the issuer, the issue time, and the characters or numbers in the predetermined entry fields from the acquired image of the certificate.
The issuer, the issue time, and the image of the character or number extracted from the acquired image of the certificate by referring to the database associated with the image of the character or number in the predetermined entry field, and the image of the character or number extracted from the database. Further including determining the authenticity of the acquired image of the certificate by using a second determination device for determining the similarity of the letters or numbers to the image.
The authenticity determination method described in Appendix 9.

(Appendix 11) Analyzing the acquired image of the certificate and acquiring a plurality of dates from a plurality of first entry fields in which the dates are described.
With reference to the first information indicating the future relationship of the dates described in the plurality of first entry fields, whether the acquired future relationship of the plurality of dates is consistent with the future relationship indicated by the first information. It further includes determining whether or not the image of the obtained certificate is forgery if it is confirmed that the information is inconsistent.
The authenticity determination method according to Appendix 10 or 11.

(Appendix 12) The image of the obtained certificate is analyzed to specify the issuance time, and among the description contents of the matters proved by the certificate, the description is different depending on the issuance time. Detecting the content and
It is confirmed whether or not the specified issuance time and the detection result for the predetermined description are consistent, and if it is confirmed that they are not consistent, the obtained image of the certificate is an art forgery. Further including determining that there is,
The authenticity determination method according to any one of Supplementary Notes 9 to 11.

(Appendix 13) The acquired image of the certificate is analyzed, and the characters or numbers are extracted from each of the plurality of second entry fields in which the characters or numbers that can be used to identify the issuer are described. It is confirmed whether or not the issuers corresponding to the plurality of second entry fields match each other, and if it is confirmed that they do not match, it is determined that the obtained image of the certificate is fake. Including,
The authenticity determination method according to any one of Appendix 9 to 12.

(Appendix 14) An acquisition unit that acquires an image of a certificate that is subject to authenticity judgment,
Mask the acquired image of the certificate by using a mask image that leaves the first part where information about the issuer is described and at least hides the second part whose contents are different for each proved person. A masking unit that generates a judgment image and
Obtained by using a first determination device obtained by pattern learning an image group masked with the mask image on a sample image of the certificate prepared in advance, and the determination image. An information processing device having a determination unit for determining the authenticity of the image of the certificate.

(Appendix 15) Obtaining an image of a certificate that is subject to authenticity judgment,
Mask the acquired image of the certificate by using a mask image that leaves the first part where information about the issuer is described and at least hides the second part whose contents are different for each proved person. Generating an image for judgment and
Obtained using a first determination device obtained by pattern learning an image group masked with the mask image on a sample image of the certificate prepared in advance, and the determination image. A program that causes a computer to perform processing including determining the authenticity of an image of the certificate.

A non-transitory computer readable storage medium in which the program described in Appendix 15 is stored.

Although preferred embodiments of the present disclosure have been described above with reference to the accompanying drawings, the present disclosure is not limited to such examples. It is clear that a person skilled in the art can come up with various modifications or modifications within the scope of the claims, which naturally belong to the technical scope of the present disclosure.

10 Authenticity judgment device 11 Confirmation document reception unit 12, 22, 32, 131, 141, 151, 161 Storage unit 13 First judgment unit 14 Second judgment unit 15 Third judgment unit 16 Comprehensive judgment unit 20 Judgment generator generator 21 Analysis Part 22a Certificate image group 22b, 131a Mask image 30 User equipment 31 Imaging unit 33 Control unit 34 Display unit 41, 51 Sample image 42, 52 Template image 43, 141a, 241a Glyph image DB
131b Pattern recognizer 132, 142, 152 Target image acquisition unit 133 Masking unit 134 Black-painted image judgment unit 141b Similarity judgment unit 143 Character recognition unit 144 Character shape judgment unit 151a Date condition DB
151b Relationship DB
153 Date / character recognition unit 154 Consistency judgment unit 161a Judgment mode DB
161b Threshold DB
162 Judgment result acquisition unit 163 Judgment processing unit 164 Message output unit 231a Template image DB
NW network

Claims (10)

Obtaining an image of the certificate that is the target of authenticity judgment,
Mask the acquired image of the certificate by using a mask image that leaves the first part where information about the issuer is described and at least hides the second part whose contents are different for each proved person. Generating an image for judgment and
With reference to the first storage unit that stores the image group masked by the mask image with respect to the sample image of the certificate prepared in advance, each image of the image group and the determination image The similarity is compared, and when the similarity is equal to or higher than the threshold value for any of the images in the image group, the determination result for the acquired image of the certificate is true, and all the images in the image group have the similarity. An authenticity determination method in which a computer executes a process including executing a first authenticity determination process in which the determination result for the acquired image of the certificate is false when the degree is less than the threshold value. The authenticity determination method according to claim 1, wherein the first portion includes characters that can be used to identify the issuer and figures having different shapes for each issuer. The mask image according to claim 1 or 2, wherein at least a part of a portion where a specific background color is used when a specific matter is proved by the certificate is set to remain at the time of the masking. Authenticity judgment method. The mask image according to claim 3, wherein at least a part of a portion in which a specific character is described remains at the time of the masking when the specific matter is proved by the certificate. Authenticity judgment method. Extracting the image of the issuer, the issue time, and the characters or numbers in the predetermined entry fields from the acquired image of the certificate.
With reference to the second storage unit that associates the issuer, the issue time, and the image of the character or number in the predetermined entry field, the image of the character or number extracted from the image of the obtained certificate and the image of the character or number. Based on the comparison result with the image of the character or number extracted from the second storage unit, the second authenticity determination process for the acquired image of the certificate is executed.
Further including determining that the acquired image of the certificate is forgery when at least one of the first authenticity determination process and the second authenticity determination process is determined to be false. ,
The authenticity determination method according to any one of claims 1 to 4. Analyzing the acquired image of the certificate and acquiring a plurality of dates from a plurality of first entry fields in which the dates are described,
With reference to the first information indicating the future relationship of the dates described in the plurality of first entry fields, whether the acquired future relationship of the plurality of dates is consistent with the future relationship indicated by the first information. It further includes determining whether or not the image of the obtained certificate is forgery if it is confirmed that the information is inconsistent.
The authenticity determination method according to any one of claims 1 to 5. The image of the acquired certificate is analyzed to specify the issuance time, and among the description contents of the matters proved by the certificate, the predetermined description contents that differ depending on the issuance time are detected. That and
It is confirmed whether or not the specified issuance time and the detection result for the predetermined description are consistent, and if it is confirmed that they are not consistent, the obtained image of the certificate is an art forgery. Further including determining that there is,
The authenticity determination method according to any one of claims 1 to 6. The acquired image of the certificate is analyzed, and the letters or numbers are extracted from each of the plurality of second entry fields in which the letters or numbers that can be used to identify the issuer are described, and the plurality of second entries are extracted. It further includes checking whether or not the issuers corresponding to the entry fields in the above match each other, and if it is confirmed that they do not match, it is determined that the obtained image of the certificate is fake.
The authenticity determination method according to any one of claims 1 to 7. The acquisition unit that acquires the image of the certificate that is the target of authenticity judgment,
Mask the acquired image of the certificate by using a mask image that leaves the first part where information about the issuer is described and at least hides the second part whose contents are different for each proved person. A masking unit that generates a judgment image and
With reference to the storage unit that stores the image group masked by the mask image with respect to the sample image of the certificate prepared in advance, the similarity between each image of the image group and the determination image is determined. By comparison, when the similarity is equal to or higher than the threshold value for any of the images in the image group, the determination result for the acquired image of the certificate is true, and the similarity is the threshold value for all the images in the image group. An information processing apparatus having a determination unit that falsely determines the determination result for the acquired image of the certificate when the value is less than or equal to. Obtaining an image of the certificate that is the target of authenticity judgment,
Mask the acquired image of the certificate by using a mask image that leaves the first part where information about the issuer is described and at least hides the second part whose contents are different for each proved person. Generating an image for judgment and
With reference to the storage unit that stores the image group masked by the mask image with respect to the sample image of the certificate prepared in advance, the similarity between each image of the image group and the determination image is determined. By comparison, when the similarity is equal to or higher than the threshold value for any of the images in the image group, the determination result for the acquired image of the certificate is true, and the similarity is the threshold value for all the images in the image group. A program that causes a computer to execute a process including false determination result for the image of the obtained certificate when the value is less than.

Images