JP2019003421A - Authenticity determination system, method and program for identity confirmation document - Google Patents

Abstract

To provide an authenticity determination system, a method and a program for an identity confirmation document that improve determination accuracy of a forged identity confirmation document.SOLUTION: An authenticity determination system for an identity confirmation document has a model recording unit and an authenticity determination unit. The model recording unit takes a determination object image as an input, which is constructed as a learning data input of a legitimate document drawing image and a forged document drawing image of an identity confirmation document, and stores a prediction model for calculating a forgery index indicating a degree to which the identity confirmation document displayed on the determination object image can be forged. The authenticity determination unit calculates the forgery index for the determination object image by inputting the determination object image to the prediction model.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a technique for determining the authenticity of identity verification documents such as a driver's license, passport, and insurance card.

  In modern society, it is required to present identification documents in various places. Official certificates such as driver's licenses, passports, and insurance cards are used as identification documents. In online transactions or contracts on websites of various services, it may be required to send an image obtained by photographing a personal identification document.

  On the other hand, in recent years, with the advancement of copiers and digital devices, it has become easy to finely forge personal identification documents. Such elaborate counterfeiting may be difficult to distinguish visually.

  Therefore, there is a need for a system that can be easily used in various situations and that determines the authenticity of the identity verification document with high accuracy.

  Patent Document 1 discloses a technique for determining whether a driver's license is true or false. The technique disclosed in Patent Document 1 compares an image of a specific portion of a driver's license with an image of a real driver's license, and performs authenticity determination based on the comparison result.

JP 2006-11866 A

  As described above, the technique disclosed in Patent Document 1 compares an image of a specific location in a driver's license with an image of a real driver's license, and makes a true / false determination based on the comparison result. For this reason, if the specific location is elaborately forged or the image at the specific location is unclear, the accuracy of the authenticity determination may be reduced.

  The objective of this invention is providing the technique which improves the determination precision of the forged identity verification document.

  An identity verification document authenticity determination system according to one embodiment of the present invention is constructed by using a valid image drawing image and a forged document graphic image of a personal identification document as learning data input, and using the determination target image as an input to the determination target image. A model recording unit that stores a prediction model for calculating a forgery index indicating the degree to which the displayed identity verification document can be forged, and the determination target image by inputting the determination target image into the prediction model A true / false determination unit that calculates the forgery index for the image.

  According to the present invention, forgery determination of an identity verification document is performed using a prediction model generated by learning, so that forgery determination with high accuracy is possible.

1 is an overall view of an identity verification document authenticity determination system according to Embodiment 1. FIG. 1 is a schematic diagram of a driver's license according to Embodiment 1. FIG. 5 is a flowchart of authenticity determination processing according to the first embodiment. FIG. 9 is an overall view of a personal identification document authenticity determination system according to a second embodiment. 10 is a flowchart of authenticity determination processing according to the third embodiment. 10 is a flowchart of classification processing according to the third embodiment. 10 is a flowchart of authenticity determination processing according to the fourth embodiment. The figure of the area information in the personal identification document which concerns on Example 4. FIG. The figure for demonstrating the mode of the authenticity determination of the forged driver's license which concerns on Example 5. FIG. FIG. 10 is a relationship diagram between an edge angle and the number of edges according to the fifth embodiment. The figure for demonstrating the mode of the authenticity determination of the other forged driver's license which concerns on Example 5. FIG.

  Several embodiments will be described below with reference to the drawings.

  FIG. 1 is an overall view of an identity verification document authenticity determination system in the present embodiment, and FIG. 2 is a schematic diagram of a driver's license. The identity verification document authenticity determination system 1 is a system for determining whether or not an identity verification document in an input image has been forged.

  There are several types of official certificates used as identity verification documents. Depending on the type of identity verification document, the display content or display position of the photo, name, address, origin, etc. will vary. For example, a face photograph is displayed on the driver's license (hereinafter, “license”) 5 (the right center blank in FIG. 2), and a face photograph is not displayed on the health insurance card (not shown). The display position of the name is different between the license 5 and the health insurance card.

  Furthermore, even if the same type of identity verification document is created, if the creator is different, the display may be different. For example, the display format or display content of the license 5 may differ depending on the public safety committees of the prefecture (Tokyo Metropolitan Public Safety Commission, Kanagawa Prefectural Public Safety Commission, etc.). For example, if the printing presses used in the Tokyo Metropolitan Public Safety Commission and Kanagawa Prefectural Public Safety Commission are different from each other, slight differences may occur in the display position, font, and character size of the name. Further, if the photographic equipment and the photographic environment are different, a slight difference may occur in the photographic image including the background portion.

  The identity verification document authenticity determination system 1 constructs a prediction model for determining the authenticity of an identity verification document by deep learning using a legitimate identity verification document image and a forged identity verification document image as learning data. . This prediction model takes an image of a personal identification document to be determined as an input, and outputs a forgery index indicating the degree to which the personal identification document can be forged. In building a prediction model, it is important to distinguish between a legitimate identity verification document and a forged identity verification document with high accuracy.

  For example, as described above, the font in the name field of the license 5 may be different depending on the public safety committee that is the creator. In the license 5 created by the Tokyo Metropolitan Public Safety Commission, the letter A is a valid letter, but in the license created by the Kanagawa Public Safety Commission, the letter B may be a valid letter. As described above, the mixture of the image difference due to forgery and the image difference due to the creation source may hinder improvement in accuracy of the authenticity determination.

  In this regard, the identity verification document authenticity determination system 1 according to the present embodiment adopts a configuration that takes into consideration that even if the identity verification document is a valid identity verification document, there may be differences in the images in each region if the origin is different. ing. That is, in the identity verification document authenticity determination system 1 of the present embodiment, a separate prediction model is prepared for each origin of the identity verification document. As a result, the characteristics of the legitimate identity verification document and the forged identity verification document for each creation source stand out, and the influence of image differences due to the creation source on the authenticity determination is reduced.

  As shown in FIG. 1, the identity verification document authenticity determination system 1 according to the present embodiment includes an authenticity determination unit 2, a learning unit 3, and a model recording unit 4.

  The authenticity determination unit 2 includes a classification determination unit 6, a model selection unit 7, and a true / false determination calculation unit 8.

  A determination target image is input to the classification determination unit 6 via a telecommunication line from a user terminal or the like. The classification determination unit 6 determines the classification to which the personal identification document displayed in the determination target image belongs from a plurality of classifications obtained by classifying various types of personal identification documents.

  Specifically, the classification discriminating unit 6 has a major classification discriminating unit 6A and a minor classification discriminating unit 6a, and a plurality of categories obtained by classifying the identity verification documents are determined in advance as major categories, and the major classifications are determined. A plurality of categories divided by attributes that may cause differences in the images of identity verification documents belonging to the category are determined in advance as small categories. Here, as an example, the classification determination unit 6 predetermines that the type of the personal identification document is a large classification and the creation source of the personal identification document is a small classification.

  The large classification determination unit 6A extracts a partial image indicating the type from the determination target image, and determines the large classification to which the personal identification document displayed in the determination target image belongs. For example, the major classification includes a license 5, a passport, a health insurance card (not shown), and the like.

  The small classification determination unit 6a determines a small classification in the large classification to which the personal identification document displayed in the determination target image belongs, according to the large classification determined by the large classification determination unit 6A. For example, the small classification is a creation source of the identity verification document. When the major classification is the license 5, the minor classification is, for example, a prefectural public safety committee (Tokyo Metropolitan Public Safety Commission, Kanagawa Public Safety Commission, etc.). When the major classification is a passport, the minor classification is, for example, an issuing country. When the major classification is a health insurance card, the minor classification is, for example, an insurer name.

  Returning to FIG. The model selection unit 7 selects a prediction model for small classification to which the identity confirmation document displayed in the determination target image belongs from among the prediction models 4a, 4b,... Recorded by the model recording unit 4.

  The true / false determination calculation unit 8 calculates the forgery index for the determination target image by inputting the determination target image of the classified personal identification document to the prediction model selected by the model selection unit 7.

  The learning unit 3 receives the legitimate document graphic image and the forged document graphic image of the personal identification document for each small classification as learning data. The learning unit 3 calculates a forgery index indicating the degree to which the identity verification document displayed in the determination target image can be forged for each small classification based on the input learning data for each small classification. Are constructed by deep learning. The counterfeit index may be an index having a plurality of values, for example, a true / false binary index. For example, the learning unit 3 uses the prediction model 4a for each small classification based on the learning data input for each small classification and the authenticity determination data of a known identity verification document separately input in addition to the learning data. 4b,... May be constructed by deep learning.

  The model recording unit 4 stores a plurality of prediction models 4 a, 4 b,... That are different for each small classification constructed by the learning unit 3.

  FIG. 3 is a flowchart of the authenticity determination process in the present embodiment.

  In the authenticity determination process, first, the authenticity determination unit 2 corrects the direction, resolution, angle, brightness, and the like of the input determination target image to be predetermined values (S301). Next, the large classification determination unit 6A extracts a partial image indicating the type from the corrected determination target image, and extracts which large classification the identity verification document displayed in the corrected determination target image belongs to. The determination is made based on the partial image. That is, the major classification determination unit 6A determines the major classification of the personal identification document from, for example, a license 5, a passport, a health insurance card, and the like. The small classification discriminating unit 6a extracts a partial image indicating the creation source from the determination target image according to the large classification discriminated by the large classification discriminating unit 6A, and the person displayed in the determination target image based on the extracted partial image It is determined to which small category the confirmation document belongs (S302). In other words, the minor classification determination unit 6a determines, for example, the origin of the personal identification document as the minor classification. The creation source is an example of an attribute that may cause a difference in the image of the personal identification document.

  Next, the model selection unit 7 selects a prediction model corresponding to the small classification determined by the classification determination unit 6 from the prediction models 4a, 4b,... Recorded by the model recording unit 4 for each small classification.

  Finally, the true / false determination calculation unit 8 inputs the determination target images of the identity verification documents classified by the large classification determination unit 6A and the small classification determination unit 6a to the prediction model selected by the model selection unit 7. A counterfeit index for the determination target image is calculated, and the authenticity of the license 5 displayed on the determination target image is determined (S303). If the forgery index is higher than a preset threshold value (for example, 10%), the authenticity determination calculation unit 8 estimates the forgery license 5.

  According to the present embodiment, the identity verification document authenticity determination system 1 includes the model recording unit 4 and the authenticity determination unit 2. The model recording unit 4 is constructed by using the legitimate document graphic image and the forged document graphic image of the identification document as learning data input, and the identification document displayed on the determination target image is forged by using the determination target image as an input. And stores a prediction model for calculating a counterfeit index indicating the degree that can be. The authenticity determination unit 2 calculates a counterfeit index for the determination target image by inputting the determination target image to the prediction model. Thereby, it is possible to determine the forgery of the identity confirmation document using the prediction model generated by learning, and based on various characteristics of the determination target image by learning the forged document drawing image forged in various modes. A highly accurate forgery determination is possible.

  The authenticity determination unit 2 determines a classification to which the identification document displayed in the determination target image belongs among a plurality of classifications obtained by classifying the identification document, and uses the classification determined as the determination target image and the prediction model. Thus, the forgery index of the determination target image is calculated. As a result, identification documents can be classified and counterfeit determination can be performed in consideration of classification, and identification verification of various types of identification documents can be performed with high accuracy.

  The authenticity determination unit 2 predetermines a plurality of classifications obtained by classifying identity verification documents as major classifications, and classifies a plurality of classifications classified by attributes that may cause differences in images of identification documents belonging to the major classification as minor classifications. A large classification and a small classification to which the identity verification document displayed in the determination target image belongs are determined. Then, the authenticity determination unit 2 calculates a forgery index for the determination target image using the determination target image, the large classification and the small classification to which the determination target image belongs, and the prediction model. As a result, the type of identity verification document is classified as major classification, the attribute that can cause a difference in the image to be judged is classified as minor classification, and the authenticity is judged by limiting the classification step by step from major classification to minor classification. Therefore, it is possible to determine the authenticity of various types of identity verification documents with high accuracy.

  Since the prediction models 4a, 4b,... Are provided for each classification, forgery determination of a plurality of categories can be performed with high accuracy by the prediction models 4a, 4b,.

  Since identity verification documents are usually created with the same equipment as long as the origin is the same, it is possible to make forgery determination with high accuracy by classifying identity verification documents by type and origin and considering them. It becomes.

  Since it has a learning unit that builds a prediction model using the legitimate document graphic image and counterfeit document graphic image of the identity verification document as learning data input, a highly reliable prediction model can be constructed with the input learning data, and high accuracy Counterfeit determination is possible.

  In this example, the identity verification document is the license 5, and the origin is the prefectural public security committee, so the license 5 created by the same public security committee has the same typeface and format. It is possible to accurately determine forgery by performing forgery determination in consideration of which prefectural public security committee has created the license 5.

  In the case of the license 5, since the “Personal Address” field may disappear after 2010 or the font may be changed, the sub-classification may be at the time of creating the personal identification document. Further, the minor classification may be whether the image is a color image (color image or black and white image). For this reason, in the case of a color image, the amount of information input to the prediction model can be increased to improve the determination accuracy. On the other hand, in the case of a black and white image, the amount of information input to the prediction model can be reduced. Capacity can be reduced. Further, the minor classification may be the presence or absence of a good mark or the presence or absence of an AT-limited mark.

  The identity verification document authenticity determination system 1 according to the second embodiment will be described. Each of the following embodiments including this embodiment corresponds to a modification of the first embodiment. In the first embodiment, a prediction model is constructed for each creation source of the identity verification document, and a forgery index is calculated using a prediction model corresponding to the creation source to which the determination target image belongs. As a result, the effect of image differences due to differences in creation sources on authenticity determination was reduced. On the other hand, in Example 2, the information indicating the creation source is used as an input to the prediction model, thereby reducing the influence of the image difference due to the creation source on the authenticity determination. Hereinafter, the second embodiment will be described focusing on differences from the first embodiment.

  FIG. 4 is an overall view of the identification document authenticity determination system according to the second embodiment.

  In the learning unit 23 of the present embodiment, the identity verification document for each information of the large classification (for example, license 5, passport, health insurance card, etc.) and the small classification (for example, the origin of the identification document) is valid. A document graphic image and a forged document graphic image are input as learning data. The learning unit 23 constructs a single prediction model 20 by deep learning based on the input learning data for each of the large classification information and the small classification information.

  The model recording unit 4 of this embodiment stores the prediction model 10 constructed by the learning unit 23. The prediction model 10 of this embodiment is different from the prediction models 4a, 4b,... Of the first embodiment in that large classification information and small classification information are input in addition to the determination target image. In the first embodiment, the model storage unit 4 stores a plurality of prediction models for each creation source. However, in the second embodiment, a single prediction model that can be commonly used for identity verification documents of all creation sources. 20 is memorized.

  The classification discriminating unit 6 of the present embodiment is the same as that of the first embodiment shown in FIG.

  The authenticity determination process of the present embodiment differs from that of the first embodiment shown in FIG. 3 only in step S303. In the authenticity determination process, in step S302, similar to the first embodiment, the large classification determination unit 6A extracts a partial image indicating the type from the corrected determination target image and displays the partial image on the corrected determination target image. Which major classification the identity verification document belongs to is determined based on the extracted partial image. Based on the major classification discriminated by the major classification discriminating section 6A, the minor classification discriminating section 6a extracts a partial image indicating the creation source from the judgment target image and displayed on the judgment target image of the personal identification document belonging to the major classification. It is discriminated based on the extracted partial image whether it was created by the creation source. Finally, in step S303, unlike the first embodiment, the true / false determination calculation unit 18 inputs the large classification and the small classification determined by the classification determination unit 6 and the determination target image to the prediction model 20 to determine the determination target. A counterfeit index for the image is calculated.

  According to the present embodiment, it is possible to construct the prediction model 20 having the classification as an input, and perform the forgery determination by the prediction model 20 with high accuracy.

  Since identity verification documents are usually created with the same equipment as long as the origin is the same, it is possible to make forgery determination with high accuracy by classifying identity verification documents by type and origin and considering them. It becomes.

  The identity verification document authenticity determination system 1 according to the third embodiment will be described. The process for correcting the determination target image in the first embodiment does not use the outer edge of the personal identification document. On the other hand, in the third embodiment, information on the outer edge of the identity verification document is used for processing for correcting the determination target image input to the authenticity determination unit 2 to a predetermined direction and angle. Therefore, in the third embodiment, it is determined whether or not the outer edge of the identification document can be detected in the determination target image. If the outer edge cannot be detected, the outer edge is estimated.

  The overall view of the identity verification document authenticity determination system 1 of the third embodiment is the same as that of the first embodiment shown in FIG. However, the authenticity determination process executed by the authenticity determination unit 2 according to the third embodiment uses a process for detecting the outer edge of the personal identification document, a process for estimating the outer edge of the personal identification document, and the outer edge of the personal identification document. It differs from that of Example 1 in that it includes processing.

  FIG. 5 is a flow of authenticity determination processing according to the third embodiment.

  The authenticity determination unit 2 of this embodiment determines whether or not the outer edge of the license 5 is displayed on the determination target image.

(Process to detect the outer edge of identity verification documents)
In the authenticity determination process, the authenticity determination unit 2 detects the outer edge of the identification document from the determination target image before image correction (S301), and whether the outer edge of the identification document is displayed in the determination target image. It is determined whether or not (S500). When the outer edge of the personal identification document is displayed (S500: YES), the process proceeds to the image correction process (S301).

(Process to estimate the outer edge of identity verification documents)
When the outer edge of the personal identification document is not displayed (S500: NO), the outer edge of the personal identification document is estimated (S504). For example, when a part of the outer edge of the identification document is not displayed, a part of the outer edge that is not displayed can be estimated based on most of the displayed outer edges. If most of the outer edge of the identity verification document is not displayed, the reference position is extracted from the displayed identity verification document, and the display position is displayed based on the relationship between the extracted reference position and the outer edge of the displayed part. It is possible to estimate the outer edge of the missing identity verification document.

(Processing using the outer edge of identity verification documents)
In the image correction process (301), the estimated outer edge of the identification document can be used as a reference line when correcting the orientation and angle of the determination target image input to the authenticity determination unit 2. For example, when the outer edge of the identification document displayed in the determination target image has a trapezoidal shape, keystone correction is performed. Note that when it is determined that the background displayed outside the outer edge of the determination target image is the same as the background when the counterfeit determination has been made in the past, the counterfeit index may be increased.

  Then, the large classification determination unit 6A determines the classification of the personal identification document displayed in the determination target image based on the image in the outer edge of the personal identification document displayed in the determination target image (S302). Hereinafter, the classification determination process of the present embodiment will be specifically described.

  FIG. 6 is a flowchart of the classification process in this embodiment.

  The large classification discriminating unit 6A according to the present embodiment discriminates the type of the identity verification document in which a part of the outer edge is not displayed in the judgment target image, and displays the origin of the identity verification document based on the identified type. Two or more areas of the area to be processed and the other areas are extracted. The small classification discriminating section 6a classifies the personal identification document by discriminating the creation source based on the area where the creation source of the personal identification document is displayed in the area extracted by the large classification discrimination section 6A. Determine.

  According to the present embodiment, if the outer edge of the identification document is not displayed in the determination target image, the outer edge can be estimated, and the direction and angle of the determination target image can be corrected using the estimated outer edge. Authenticity can be determined with high accuracy.

  Based on the type of identity verification document, the image to be judged is divided into a plurality of areas including at least whether or not the origin is displayed, and the origin is selected from the area where the origin of the identity verification document is displayed. Since the creation source can be determined by the determination procedure, the generation source can be determined with high accuracy.

  The identity verification document authenticity determination system 1 according to the fourth embodiment will be described. In the first embodiment, a prediction model for each creation source is provided, and a single counterfeit index is calculated. On the other hand, the fourth embodiment has a prediction model corresponding to a plurality of regions in the determination target image, calculates individual counterfeit indexes for the plurality of regions, and combines the calculated individual counterfeit indexes for each region. A counterfeit index of the determination target image is calculated.

  In the learning unit 43 of the present embodiment, the legitimate document graphic image of the identity verification document for each type of information of the identity verification document (for example, license 5, passport, health insurance card, etc.) and information of the origin of the identity verification document The counterfeit document graphic image is input as learning data. Based on the learning data for each of the large classification information and the small classification information that has been input, the learning unit 43 performs, for each type and each original identification document, a plurality of regions included in the identification document. Build prediction models corresponding to each.

  The model storage unit 44 of this embodiment stores a plurality of prediction models 44a, 44b,... Constructed by the learning unit 43.

  FIG. 7 is a flow of authenticity determination processing according to the fourth embodiment.

  The authenticity determination unit 2 of this embodiment holds in advance personal identification document area information indicating what kind of area exists in the personal identification document for each type of personal identification document.

  FIG. 8 is a diagram illustrating an example of the personal identification document area information. Referring to FIG. 8, the area information in the personal identification document stores the type of the personal identification document, the area names of a plurality of areas included in the type of personal identification document, and the coordinate information of these areas. Yes. For example, the license 5 includes a name area, a date of birth area, an address area, a delivery information area, an expiration date area, a condition area, a good driver display area, a driver's license number area, and a photograph area as shown in FIG. , A license acquisition date area, and a creation source area. Therefore, in FIG. 8, the area name and coordinate information of these areas are stored for the license 5.

  The name area is an area where the name of the license holder is displayed. The birth date area is an area in which the date of birth of the license holder is displayed. The address area is an area where the address of the license holder is displayed. The delivery information area is an area in which a license issuance date is displayed. The expiration date area is an area where the expiration date of the license is displayed. The condition area is an area where license conditions are displayed. The excellent driver display area is an area where a good mark is displayed for a good driver. The driver's license number area is an area where the license number is displayed. The photo area is an area where a face photograph of the license holder is displayed. The license acquisition date area is an area in which the acquisition date of the motorcycle license, the acquisition date of the automobile license, and the acquisition date of the two-type license are displayed. The creation source area is an area in which the name and mark of the public safety commission of the prefecture that created the license are displayed.

  The authenticity determination unit 2 determines the type of the identity verification document displayed in the determination target image, refers to the area information in the identity verification document based on the determined type, and includes a plurality of regions included in the type of identity verification document To know. Subsequently, the authenticity determination unit 2 extracts a plurality of acquired areas from the determination target image, calculates an individual forgery index with a prediction model selected for each of the plurality of areas, and combines the individual forgery indexes for each area. The counterfeit index of the determination target image is calculated.

  Note that the plurality of areas extracted from the determination target image include text description and graphic display of the personal identification document displayed in the determination target image. The text is, for example, a name. The drawing is, for example, a photograph.

  Furthermore, in the present embodiment, an example in which counterfeit indices of all regions are calculated using a prediction model corresponding to each region has been described, but the present invention is not limited to this. As another example, a forgery index may be calculated using a prediction model for some areas, and a counterfeit index may be calculated using another determination method for some other areas. Various other existing methods can be applied as other determination methods. For example, the image of the area may be compared with the image of that area of the real identity verification document, and the forgery index may be calculated based on the similarity of the images.

  In the authenticity determination process, first, the authenticity determination unit 2 determines the type of the identity verification document displayed in the determination target image, extracts a plurality of regions from the determination target image based on the determined type, and extracts the extracted plural One area is selected from these areas (S701). Next, the authenticity determination unit 2 selects a determination method used for the region (S702). In this embodiment, since the determination method is a method based on a prediction model, the authenticity determination unit 2 selects a prediction model.

  The authenticity determination unit 2 calculates an individual forgery index by the determination method (S703). Then, the authenticity determination unit 2 determines whether there is a remaining area (S704). If there is a remaining area (S704: YES), the process returns to S701 until there is no remaining area.

  When there is no remaining area (S704: NO), the authenticity determination unit 2 calculates the counterfeit index for the determination target image by combining the individual counterfeit indexes for each area (705). For example, the counterfeit index for the determination target image is calculated by weighting and adding the individual counterfeit index for each region.

  According to the present embodiment, since the counterfeit index of the determination target image can be calculated by combining the individual counterfeit indexes for each region, the counterfeit index for the determination target image can be calculated with high accuracy.

  The identity verification document authenticity determination system 1 according to the fifth embodiment will be described. In the first embodiment, when the authenticity determination process is started, the authenticity determination unit 2 corrects the direction, resolution, angle, brightness, and the like of the input determination target image to be predetermined values. On the other hand, in the fifth embodiment, the authenticity determination unit 2 executes a preliminary forgery determination process before starting the correction process shown in the first embodiment. The preliminary forgery determination process is a process for determining that the document is a forged identity verification document when there is an obvious cut and past mark in the determination target image. For a determination target image that has not been determined to be a personal identification document forged by preliminary counterfeit determination processing, the processing described in the first embodiment is further executed.

  Hereinafter, preliminary forgery determination processing will be described.

  FIG. 9 is a diagram for explaining a state of authenticity determination of a forged license according to the fifth embodiment.

  The authenticity determination unit 2 according to the present embodiment detects the cut and paste marks based on the inclination of the boundary line of the cut and paste marks, and calculates a forgery index for the determination target image based on the detected cut and paste marks.

  In the preliminary forgery determination process, first, when the upper and lower edge lines 11 and 12 are not displayed, the outer edge detection process, the estimation process, and the use process of the identification document are performed as in the third embodiment, and the identification process is performed. Estimate the outer edge of the document. Next, the authenticity determination unit 2 detects a straight line or a curve from the determination target image by edge detection processing. The image of the personal identification document has a line that should be originally included in the image depending on the type of the personal identification document. For example, if the identity verification document is a license 5, there are upper and lower edge lines 11 and 12 of the license 5 itself. In the name display field, there are upper and lower frame lines 13 and 14. The face photo also has upper and lower frame lines 15 and 16. In addition, the determination target image may include a line other than the line originally included in the identity verification document.

  The authenticity determination unit 2 adds the upper and lower edge lines 11 and 12 of the license 5, the upper and lower frame lines 13 and 14 of the name display field, and the upper and lower frame lines 15 and 16 of the face photograph to the determination target image displaying the license 5. When the inclined line segments 17 and 18 that are inclined with respect to the above are detected, the inclined line segments 17 and 18 are extracted. In this embodiment, it is possible to confirm two upper and lower inclined line segments 17 and 18 that incline to the lower right across the upper and lower frame lines 13 and 14 in the name display field.

  FIG. 10 is a diagram illustrating the relationship between the edge angle and the number of edges according to the present embodiment.

  Here, in the normal license 5, the upper and lower edge lines 11 and 12 and the upper and lower frame lines 13 to 16 extend in parallel to each other on the left and right. Therefore, it can be determined that there is a high possibility that the inclined line segments 17 and 18 that are not perpendicular to the left-right direction and greatly deviate from other angular distributions are clear cut and past marks. For the distortion of the inclined line segments 17 and 18, the inclined line segments 17 and 18 are extracted, the number of the extracted inclined line segments 17 and 18 is counted for each angle, and a counterfeit index for the determination target image is calculated. Thereby, the forgery of the license 5 including cut and past marks can be determined. The inclined line segments 17 and 18 may detect angles with respect to the vertical direction parallel to the left and right edge lines and the left and right frame lines.

  As a method for detecting an edge from an image, a method using a sobel filter or a Hough transform is generally known. Specifically, an edge is detected by converting the input image into a set of edge points using a sobel filter and performing a Hough transform on the converted set of edge points.

  Furthermore, as shown in FIG. 11, the authenticity determination unit 2 may detect cut and paste marks based on distortion of the edges of the cut and paste marks. For example, the authenticity determination unit 2 may detect two upper and lower curves 27 and 28 that are vertically distorted between the upper and lower edges 23 and 24 of the name display field and the name.

  Note that the counterfeit index of the determination target image may be calculated based on the length and thickness of the line segment, the position of the maximum or minimum, or the color difference from the background, without being limited to the angle or distortion of the line segment. Therefore, even if a license 5 including an inclined line segment or a curve is newly issued, this method can be applied.

  The above-described embodiments of the present invention are examples for explaining the present invention, and are not intended to limit the scope of the present invention only to those embodiments. Those skilled in the art can implement the present invention in various other modes without departing from the gist of the present invention.

DESCRIPTION OF SYMBOLS 1 ... Identity verification document authenticity determination system, 2 ... Authenticity determination part, 3 ... Learning part, 4 ... Model recording part, 4a, 4b ... Prediction model, 5 ... License, 20 ... Prediction model, 23 ... Learning part

Claims (15)

Indicates the degree to which the identity verification document displayed on the determination target image can be forged by using the determination target image as an input, constructed using the legitimate document graphic image and the forged document graphic image of the identification document as learning data input A model recording unit for storing a prediction model for calculating a counterfeit index;
An identity verification document authenticity determination system comprising: a true / false determination unit that calculates the forgery index for the determination target image by inputting the determination target image into the prediction model. The authenticity determination unit
Determining a classification to which the identification document displayed in the determination target image belongs among a plurality of classifications obtained by classifying the identification document;
Using the determination target image, the classification, and the prediction model, calculating the forgery index for the determination target image.
The identity verification document authenticity determination system according to claim 1. The authenticity determination unit
A plurality of categories obtained by classifying the identity verification documents are determined in advance as major categories,
A plurality of classifications divided by attributes that may cause differences in images of identity verification documents belonging to the major classification are determined in advance as minor classifications,
Determine the major classification and minor classification to which the identity verification document displayed in the judgment target image belongs,
The counterfeit index for the determination target image is calculated using the determination target image, the large classification and the small classification to which the determination target image belongs, and the prediction model.
The identity verification document authenticity determination system according to claim 2. The model storage unit stores a prediction model provided for each classification,
The authenticity determination unit
It is determined whether the identity verification document displayed in the determination target image is an image of the identity verification document included in any classification,
Using the determined prediction model of the classification, calculating the forgery index for the determination target image,
The identity verification document authenticity determination system according to claim 2. The type of the identity verification document is classified as a major classification, the origin of the identification document is classified as a minor classification,
The model storage unit stores a prediction model for each of the small classifications,
The authenticity determination unit
Extracting a partial image indicating the type from the determination target image, and determining which major classification the identity verification document represented in the determination target image belongs to,
A partial image indicating the creation source is extracted from the determination target image in accordance with the determined large classification, and based on the partial image, which subclass the identity verification document displayed in the determination target image belongs to Discriminate,
Calculating the forgery index for the determination target image, using the small classification prediction model to which the identity verification document displayed in the determination target image belongs;
The identity verification document authenticity determination system according to claim 4. The model storage unit stores a prediction model having the classification as an input,
The authenticity determination unit
Determine which classification the identity verification document displayed in the determination target image belongs to,
The determined classification is input to the prediction model, and the forgery index for the determination target image is calculated.
The identity verification document authenticity determination system according to claim 2. The type of the identity verification document is classified as a major classification, the origin of the identification document is classified as a minor classification,
The model storage unit stores a prediction model that receives the major classification and the minor classification,
The authenticity determination unit
Extracting a partial image indicating the type from the determination target image, and determining which major classification the identity verification document represented in the determination target image belongs to,
A partial image indicating the creation source is extracted from the determination target image in accordance with the determined large classification, and based on the partial image, which subclass the identity verification document displayed in the determination target image belongs to Discriminate,
Inputting the major classification and the minor classification to which the identification document displayed in the determination target image belongs to the prediction model, and calculating the forgery index for the determination target image;
The identity verification document authenticity determination system according to claim 6. The authenticity determination unit
It is determined whether an outer edge of the identity verification document is displayed on the determination target image,
If there is no outer edge, the outer edge is estimated,
Based on the image within the outer edge displayed on the determination target image, the classification of the identity verification document displayed on the determination target image is determined.
The identity verification document authenticity determination system according to claim 2. In the process of determining the classification, the authenticity determination unit includes:
Determine the type of identity verification document displayed in the determination target image,
Extracting a plurality of regions from the determination target image based on the determined type,
Based on an area where the origin of the identity verification document is displayed among the plurality of extracted areas, an attribute that may cause a difference in the image of the identity verification document is determined.
The identity verification document authenticity determination system according to claim 8. The authenticity determination unit
Determine the type of identity verification document displayed in the determination target image,
Extracting a plurality of regions from the determination target image based on the determined type,
An individual counterfeit index is calculated by a determination method selected for each of the plurality of areas,
Calculating the counterfeit index of the determination target image by combining the individual counterfeit indexes for each region;
The identity verification document authenticity determination system according to claim 2. A learning unit that constructs the prediction model using the legitimate document graphic image and the forged document graphic image of the identity verification document as learning data input;
The identity verification document authenticity determination system according to claim 1. The official document is a driver's license;
The origin is the prefectural public safety committee.
The identity verification document authenticity determination system according to claim 5 or 7. The authenticity determination unit
Detecting cut and paste marks of the identity verification document displayed in the determination target image, and calculating the forgery index for the determination target image based on the detected cut and paste marks,
The identity verification document authenticity determination system according to claim 1. The model recording means provided in the computer is constructed using the legitimate document graphic image and the forged document graphic image of the personal identification document as learning data input, and the personal identification document displayed on the determination target image is forged using the determination target image as an input. Record a predictive model for calculating counterfeit indicators that can be
An identity verification document authenticity determination method in which authenticity determination means included in a computer calculates the forgery index for the determination target image by inputting the determination target image to the prediction model. In the model recording means provided in the computer, the identification document displayed on the determination target image is forged using the determination target image as an input, which is constructed using the legitimate document graphic image and the forged document graphic image of the personal verification document as learning data input. Record a predictive model to calculate a counterfeit indicator that can be
An identity verification document authenticity determination program for causing an authenticity determination means provided in a computer to calculate the forgery index for the determination target image by inputting the determination target image into the prediction model.

Images