JP2006053608A - Authentication system for imprint - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an authentication system for an imprint further improving reliability by extracting a feature point in a frequency component of an imprint image and checking on the feature point. <P>SOLUTION: Authenticating imprint image data taken in by a scanner 1 and identification data inputted by an input part 8 are acquired. An information generation part 10 executes a process for calling a binary one-dimensional template and registering imprint image data corresponding the identification data, an alignment process for performing alignment based on correlation between the called registering imprint image data and the authenticating imprint image data, a process for performing DCT conversion, a process for calculating a DCT coefficient, a process for finding a DCT code, and a process for generating binary one-dimensional information corresponding to the number of blocks from the DCT code. An authentication part 9 compares and authenticates the binary one-dimensional template corresponding to the identification data generated by the information generation part 10 generating registration information, and the binary one-dimensional information generated by the information generation part 10 generating authentication information. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an imprint authentication system, and more particularly to an imprint authentication system that further improves reliability by extracting feature points in frequency components of an imprint image and checking the feature points.

  Conventionally, a bank seal is visually checked against a seal of a slip and a seal imprinted on a passbook when a deposit seal is stamped on a seal column of a passbook. However, there is an increasing problem that the bankbook is stolen, the registered seal stamped in the bankbook is forged, and the savings are withdrawn.

  For this reason, in recent years, the method of not displaying the seal stamp on the passbook has begun to be adopted, but from the user's side, when trying to make a bank transaction with the passbook, it has become inconvenient because it does not know which seal stamp. If the user pushes the wrong seal on the slip, the transaction becomes impossible, and in order to avoid this, a plurality of seals may be brought to the bank for authentication.

  However, it is troublesome to visually perform such a plurality of seal authentications, which may lead to complicated banking work.

  For this reason, Patent Document 1 proposes a passbook issuing system, a passbook, and an imprint collation system that can digitize seal stamp data and rationalize office work that has been collated visually. In Patent Document 1, there is a description that a bar code (one-dimensional code) may be used as a label code printing format, but there is no disclosure of a method for creating a one-dimensional code, which is impossible.

  In Patent Document 2, an embedded bitmap image is created by embedding keywords (password, seal date, document information) in a solid white or black bitmap image created separately from an imprint image. The registered seal image is ANDed to create a seal image. However, since this technique is an image embedding technique that does not employ image compression, there is a drawback that the data capacity is increased.

Patent Document 3 is a method in which an account number or the like is used as a bar code, and the bar code and a seal image are associated and registered in a database, and has the same drawback as Patent Document 2 in that the seal image is compared.
Japanese Patent Laid-Open No. 10-208048 JP 2000-20718 A JP-A-9-319807

  An object of the present invention is to provide a seal authentication system that further improves reliability by extracting feature points in frequency components of a seal image and checking the feature points.

  Other problems of the present invention will become apparent from the following description.

  The above problems are solved by the following invention.

(Claim 1)
A scanner having an image capturing unit for capturing a seal imprint, an authentication server having an input unit for inputting identification data, an information generating unit for generating registration information and authentication information, and an authentication unit for authenticating a seal image,
An information generation unit that generates registration information in the authentication server acquires registration imprint image data captured by the scanner and identification data input by the input unit, and the information generation unit acquires the acquired registration. A process for dividing the seal image data into a plurality of blocks, a process for performing DCT conversion for each block, a process for calculating a DCT coefficient, and the DCT coefficients up to a predetermined number from the upper absolute value of the DCT coefficient A process for obtaining a one-to-one address (address) and a DCT code, and a process for generating a binary one-dimensional template for the number of blocks from the DCT code,
Means for storing the generated binary one-dimensional template corresponding to the number of blocks, the acquired identification data, and the acquired registration imprint image data in association with each other;
An information generation unit for generating authentication information in the authentication server acquires the imprint image data for authentication captured by the scanner and the identification data input by the input unit, and the information generation unit converts the identification data into the identification data. Processing for calling the corresponding registration imprint image data and binary one-dimensional template, alignment processing for performing alignment based on the correlation between the called registration imprint image data and authentication imprint image data, and the alignment processing Processing for dividing the obtained authentication imprint image data into a plurality of blocks, processing for DCT conversion for each block, processing for calculating a DCT coefficient, and a predetermined number from the top of the absolute value of the DCT coefficient A process of obtaining an address (address) and a DCT code corresponding to a DCT coefficient on a one-to-one basis, and generating binary one-dimensional information for the number of blocks from the DCT code Run the process,
The authentication unit compares the binary one-dimensional template corresponding to the identification data generated by the information generation unit that generates registration information and the binary one-dimensional information generated by the information generation unit that generates authentication information. Imprint authentication system characterized by authentication.

(Claim 2)
The imprint authentication system according to claim 1, wherein the binary one-dimensional template and the binary one-dimensional information for each block generated by the information generation unit include the number of pixels of the imprinted pixels in the configuration.

(Claim 3)
When the number of pixels with imprints in the binary one-dimensional template and / or binary one-dimensional information is equal to or less than the threshold, the authentication unit removes the target from the authentication target and performs the authentication using only the blocks larger than the threshold. The seal stamp authentication system according to claim 2, wherein the system is executed.

(Claim 4)
3. The authentication unit compares the binary one-dimensional template and the binary one-dimensional information, compares the number of pixels, and further authenticates using fuzzy inference based on the comparison result. Or the imprint authentication system of 3 description.

(Claim 5)
Means for storing the generated binary one-dimensional template for the number of blocks, the acquired identification data, and the acquired registration imprint image data in correspondence with a data table in the authentication server or the authentication server; The seal stamp authentication system according to claim 1, wherein the seal stamp authentication system is a data table in a provided database server.

(Claim 6)
6. The data table is configured by associating the generated binary one-dimensional template corresponding to the number of blocks, the acquired identification data, and the acquired registration imprint image data. Seal imprint authentication system.

(Claim 7)
The data table is configured by associating the generated binary one-dimensional template for the number of blocks, the acquired identification data, the acquired registration imprint image data, and user attributes. The seal stamp authentication system according to claim 5.

(Claim 8)
The data table includes a main data table and a sub data table, and the main data table includes the identification data, a sub data table name corresponding to the identification data, and registered seal image data. The seal imprint authentication system according to claim 5.

(Claim 9)
The sub-data table is information for generating the identification data, the block number of each block divided into the plurality of blocks corresponding to the identification data, the number of imprint pixels of each block divided into the plurality of blocks, and the registration information 9. The seal imprint authentication system according to claim 8, comprising a binary one-dimensional template generated by the generation unit.

(Claim 10)
Means for classifying storage in the sub-data table in the data table (1) a configuration corresponding to an outer peripheral shape of an imprint including at least three outer imprint shapes of a round mark, a square mark, and an oval mark; A configuration that distinguishes readings from the edge of the character part and arranges them in the order of the Japanese syllabary, or (3) a configuration that corresponds to a typeface of an imprint that includes at least three types of imprints, a typeface, a typeface, and an old typeface The seal stamp authentication system according to claim 8 or 9, wherein the seal stamp authentication system is any one of the above.

  INDUSTRIAL APPLICABILITY The present invention can provide a seal authentication system that further improves reliability by extracting feature points in frequency components of a seal image and checking the feature points. Specifically, to ensure safety in real estate transactions and money handling, which is the most important issue for both individuals and corporations, it is quick and accurate whether the seal used is genuine or counterfeit. A system that can be identified can be provided, and it is highly effective in places where authentication is performed using seal stamps.

  Hereinafter, the best mode for carrying out the present invention will be described.

  FIG. 1 is a functional block diagram showing an example of an imprint authentication system according to the present invention, and FIG. 2 is a functional block diagram showing another example of an imprint authentication system according to the present invention.

  The example of FIG. 1 is a system example in which a scanner 1, an authentication server 2, and a database server 3 are connected by a communication network 4. Unlike the example of FIG. 1, the example of FIG. 2 is a system example in which the database server 3 is not provided, the scanner 1 and the authentication server 2 are included, and the data table 13 is provided in the authentication server 2.

  In the example of FIG. 1, the functions of the scanner 1, the authentication server 2, and the database server 3 are separated. However, as hardware, the scanner 1 may have the functions of the authentication server 2 and the database server 3. . Further, the authentication server 2 and the database server 3 may be a single server and have two functions. In the example of FIG. 2, the functions of the scanner 1 and the authentication server 2 are separated. However, as the hardware, the scanner 1 may have the function of the authentication server 2.

  In the following description, the example shown in FIG.

  The scanner 1 includes at least an image capturing unit 5 that captures registration and authentication seals as seal image data, and a transmission unit 6.

  The authentication server 2 includes an image acquisition unit 7 that acquires the imprint image data at the time of registration and authentication from the scanner 1, an input unit 8 that inputs identification data for registration and authentication related to the imprint image data, An authentication unit 9 that performs authentication by comparing a binary one-dimensional template (to be described later) with binary one-dimensional information for authentication, a registration template to be created at the time of registration, and binary one-dimensional information to be created at the time of authentication are created. The information generation part 10, the display part 11, and the transmission part 12 are provided at least.

  The input unit 8 plays a role of inputting identification data related to the registration and authentication imprint images, and for example, a user interface can be used. As the user interface 8, a keyboard, a mouse, a touch panel, or the like can be used. Hereinafter, an example in which a keyboard is used as an input unit will be described.

  The information generation unit 10 is a registration information generation unit during registration and an authentication information generation unit during authentication.

  The database server 3 includes at least a data table 13 that stores registration information generated by the information generation unit.

  Examples of the communication line network 4 include communication lines such as LAN, WAN, Internet, and dedicated lines.

  The seal impression authentication system according to the present invention includes a registration process and an authentication process. Hereinafter, each of the registration process and the authentication process will be described in detail.

(1) Registration Process An example of the registration process will be described below with reference to FIG.

  First, a registration screen is displayed on the display unit 11 (S1). The display unit 11 displays a registration screen (not shown) for input of identification data for specifying the seal of the seal to be registered and the registrant.

  Next, identification data is input from the keyboard 8 (S2). The input identification data is sent to the registration information generation unit. The identification data may be personal information that does not overlap for each user. For example, ID numbers such as bank account numbers, license numbers, social insurance numbers, and taxpayer numbers are preferable. If necessary, user attributes such as name, date of birth, company department, and title may be entered.

  Next, the scanner 1 activates the image capturing unit 5 of the scanner and captures a seal image (S3). Imprinting is performed by checking the imprinted characters up and down. At this time, the horizontal and vertical directions are not necessarily aligned. The registration imprint image data captured in S3 is transmitted from the transmission unit 6 of the scanner 1 to the authentication server 2 and is transmitted to the image acquisition unit 7 of the authentication server 2. The image acquisition unit 7 acquires the sent registration imprint image data, sends the registration imprint image data to the registration information generation unit 10, and the registration information generation unit 10 acquires the registration imprint image data and identification data. (S4).

  The registration information generation unit 10 performs image processing on the acquired registration imprint image data so as to be binarized imprint image data using a settable / changeable threshold (S5). For example, if the imprint image data sent from the scanner 1 is fetched with 256 density gradations, the pixel value is set to the threshold value 195 from the gradation, and the image processing is performed with white above the threshold and black below the threshold. FIG. 4A shows an example of a 256-tone impression image, and FIG. 4B shows an example of a monochrome image.

  The seal stamp processed by the image processing is used as the registration stamp image data.

  Next, the registration information generation unit 10 performs a binary one-dimensional template creation process (S6).

  Hereinafter, the system flow of the binary one-dimensional template creation process will be described with reference to FIG.

  First, the registration imprint image data is divided (S20). When dividing, it is necessary to find the center of the seal. For example, in the case of a circular imprint, the center and diameter of the outer periphery of the imprint are created as parameters whose diameter can be changed by the registration information generation unit 10 and are obtained by correlation with the outer periphery of the imprint. In the case of a square mark, the registration information generation unit 10 creates horizontal and vertical lines so that the vertical and horizontal lengths of the square mark exactly match, and the center is determined from the maximum vertical and horizontal length of the square mark. Ask. The center obtained in this way is also the center for the division process. FIG. 6 shows an example of dividing the seal image data of a circle. Reference numeral 400 denotes a line obtained by dividing the seal impression, and reference numeral 401 denotes a block divided by the line 400. In a preferred embodiment of the present invention, it is preferable to use 36 blocks of 6 horizontal blocks and 6 vertical blocks, but the present invention is not limited to this. Also, numbers are assigned to the blocks, and the scanning method is 1, 2, 3,... 6 from the upper left to the right. The last lower right block may be assigned the number 36. The method of assigning this number is not limited to this.

  Next, discrete cosine transform (DCT transform) is performed for each block among the divided blocks (S21).

When an image of one block is input imprint image information (digital image information) of M × N pixels, when an image signal (pixel value) at a horizontal position m and a vertical position n is f (m, n), The DCT coefficient at the horizontal position u and the vertical direction v on the frequency axis is given by F (u, v).
The two-dimensional DCT is defined by the following equation. However, the signal f (m, n) is m = 0... M-1, n = 0.

  Here, C (u) and C (v) are normalization constants as shown in the following equation.

  Since F (u, v) is a real number, a separable function of the amplitude part and the sign part thereof (Separable function)

  Can be described as However,

It is.

  The DCT coefficient is obtained by the above DCT conversion (S22). From the DCT coefficient F (u, v), the DCT amplitude and the DCT code (plus and minus) can be obtained.

  The DCT amplitude represents the power at each component (frequency) position. In other words, the DCT amplitude represents the magnitude of the component included in the original signal. The DCT amplitude is information related to the energy of the original signal. It can be said that it bears.

  Even with the same DCT amplitude, if the DCT codes are different, the signals are clearly different. The DCT code has information that is a skeleton of the signal when the signal is configured. That is, the original signal can be identified by the positive / negative arrangement of the DCT code, and the contour line image can be obtained only from the positive / negative DCT code. This becomes clear when looking at an image of sign-only synthesis (also referred to as SOS), which is the inverse transformation of only the code indicating the characteristic properties of the DCT code.

  Therefore, the DCT code reveals differences in typefaces due to the outline of the imprinted image in each block, the thickness of the imprinted character, the slope of the splash, the degree of splash, and the like. Generally, the difference in the difference appears in the high frequency area when DCT conversion is performed.

  Next, a predetermined number of addresses of the DCT coefficient are obtained from the one having a larger absolute value of the DCT coefficient, and a DCT code corresponding to the address is extracted (S23). The predetermined number depends on the number of pixels, but in the case of one block 32 × 32 pixels that can be preferably applied in the present invention, the predetermined number is determined to be about 10% (for example, about 100 points) from the larger absolute value of the DCT coefficient. Number (address number). Processing for obtaining a predetermined number of addresses is feature point extraction processing. FIG. 7 shows a distribution diagram of feature points and DCT codes of the feature points. FIG. 7 shows a distribution map of DCT codes of feature points in 8 × 8 pixels for convenience of explanation.

  Next, based on the sign of the feature point obtained by the process of S23, a binary one-dimensional template expressed as 1 for plus and 0 for minus is generated (S24). Based on the result, a binary one-dimensional template for one block as shown in FIG. 8 can be created.

  When the binary one-dimensional template is created, the process is completed for one block. This is done for the number of blocks. That is, the processing from S21 to S24 is repeated until the number of blocks is completed. The iterative process is performed to generate binary one-dimensional templates for the number of blocks.

  When the binary one-dimensional templates for the number of blocks are generated as described above, the process returns to the registration process of FIG. 5 and the registrant identification data (for example, ID) input by the input unit and the generated plural A registration template comprising the binary one-dimensional template and the registration imprint image data corresponding to the plurality of binary one-dimensional templates is created (S7). FIG. 9 shows an example of the created registration template. In addition, the binary one-dimensional template of the registration templates may be configured by combining a plurality of binary one-dimensional templates.

  In FIG. 9, 500 is a registrant ID, 501 is a binary one-dimensional template, 502 is a registrant registration imprint image data, and 503 is registered with ID 500 and a binary one-dimensional template 501. This is a registration template composed of book imprint image data 502.

  The authentication server 2 sends the created registration template to the database server 3, and the database server 3 stores the registration template in the data table 14 (S8).

  FIG. 10 shows an example in which a registration template 503 is stored. The format of the data table is not limited to the illustrated format, and personal information such as user attributes may be registered. The user attribute is an attribute of a registrant when registering a seal such as name, reading, date of birth, company name, company department, and title.

  Other examples of the data table are shown in FIGS.

  In FIG. 11, there are a main data table and a sub data table. In the main data table, a sub data table name and imprint image data corresponding to the ID are registered, and the sub data table corresponds to the ID and the block number. The number of pixels of each block and a binary one-dimensional template are registered. Further, personal information such as registrant attributes may be registered as necessary. This is a process of dividing and storing the registration template. As a method for realizing this, the order and information for forming the registration template when the registration information generating unit generates the registration template. The amount of information may be determined, and what information the amount of information is determined in advance. For example, in the registration template, at least an ID 500, a binary one-dimensional template 501, and an imprint image file name 502 are stored. As an example of the amount of information, if the ID is designated as the first 8 bytes, the binary one-dimensional template is designated as 16 bytes following the ID, and the imprint image file name is designated as 64 bytes subsequent to the binary one-dimensional template. The total amount of information is 88 bytes, and the registration template is stored in a field corresponding to the data table with reference to the information amount designated for each piece of information. As a result, even if the data table is distributed, it is possible to store the data table according to the name, the information amount, and the order in which the registration templates are configured.

  Also, the sub-data table name is determined from the typeface of the imprint, the outer shape of the imprint, and the edge of the character portion of the imprint, and the kana is determined in the order of the Japanese syllabary. You may decide whether to save in the table. The typefaces of seal imprints include seal typefaces, Kei typefaces, antique typefaces, other slave typefaces, samurai typefaces, line typefaces, and the like. Further, the sub-table names may be classified according to the type of seal such as a corporate seal or a personal seal. Moreover, you may classify | categorize according to magnitude | sizes, such as a diameter of the outer periphery shape of an imprint. FIG. 12 shows an example in which an imprint typeface is used as a sub-data table name.

  As a reference method to the sub-data table, although not shown in the figure, the ID number etc., for example, the 1000th generation goes to the sub-data table 1, and the 10000th generation goes to the sub-data table 2, so that the ID You may refer to the sub-data table by number, etc., or you may go to the sub-table by referring to the imprint shape such as a square mark, circle mark, or other oval mark. , It may be possible to refer to the sub-data table divided in the order of 50 tones. The way of reading the characters may be to identify the imprinted characters, or may be identified by registration information such as name and reading by the registrant's attribute information.

  Although not shown in the figure, when creating various sub-data tables, when sorting by the number of IDs, etc. in the main data table, sorting by imprint shape, imprint typeface, etc. The shape of the imprint, the typeface of the imprint, and the like may be registered corresponding to the identification data. For example, if you have a sub-data table for an imprint typeface and want to create a sub-data table based on the shape of an imprint, if the shape of the imprint is stored in the main data table, the data based on the shape of the imprint is stored. The registrant data corresponding to the base can be stored in the sub-data table.

(2) Authentication process Next, the authentication process will be described.

  FIG. 13 shows an example of the authentication process.

  An authentication screen is displayed on the display unit 11 (S30). The display unit 11 shows an authentication screen (not shown) having at least an ID input field and a determination button.

  When an ID is entered in the input field with the keyboard 8 (S31) and the enter button is pressed, the scanner 5 is activated and enters the capture state.

  Next, the seal imprinted on the dedicated paper by the scanner 5 is taken in (S32). The seal is captured as a multi-valued image. The captured seal imprint is sent from the transmission unit 6 of the scanner 1 to the image acquisition unit 7 of the authentication server 2 as authentication seal image data. The sent authentication imprint image data is sent from the image acquisition unit 7 to the authentication information generation unit 10, and the authentication information generation unit 10 acquires the authentication imprint image data (S33).

  The authentication server 2 calls a registration template corresponding to an ID (for example, GSD0001) from the data table among the registration templates (S34).

  Of the called registration templates, the registration imprint image data is sent to the authentication information generation unit 10, and the sent registration imprint image data is compared with the acquired authentication imprint image data. Alignment processing is performed so that the imprint image data is matched with the registration main imprint image data (S35). Specifically, a plurality of test binary imprint images are created using the threshold values that can be changed at the time of registration in the same way as the registration of the acquired imprint image data for authentication, and further, the main imprint image data for registration and the test The optimum threshold is determined using the cross-correlation value with the binary imprint image. At this time, the cross-correlation value is obtained by shifting the test binary imprint image in increments of 1 pixel and rotating in increments of 3 degrees to find the maximum correlation value. When the process of S35 is performed with the same seal, the maximum correlation value is 0.9 or more. If a different imprint is used, it will be 0.6 or less. The position where 0.9 or higher is determined as the position of the test binary imprint image. After the determination, the approximate position and the vicinity thereof are searched in increments of 1 pixel, and the rotation is searched in increments of 1 degree. Thus, the cross-correlation value is obtained, and the maximum correlation value is searched. The location where the maximum correlation value is found coincides. Next, the threshold value that can be changed at the time of registration is changed by ± 3 at the matching location, and the threshold value when the cross-correlation value becomes the maximum value with the main imprint image test binary imprint image for registration is set as the optimum threshold value. The authentication imprint image is binarized to create an authentication binary imprint image.

  If the seal is completely different, if the correlation value is less than 0.9 even after image processing, it may be instructed to capture the seal again. That is, it instructs to return to the processing of S32. Further, the authentication process itself may be terminated, or an authentication screen may be displayed to restart the authentication process from the beginning. That is, it is to return to S30.

  Next, in the authentication information generation unit 10, binary one-dimensional information is created (S36).

  Hereinafter, a description will be given of the binary one-dimensional information creation processing flow with reference to FIG.

  First, the authentication information generation unit 10 divides the binary imprint image data for authentication as in the registration process (S50). When divided, the authentication binary imprint image data is preferably divided so as to match the number of pixels (for example, one block 32 × 32 pixels) obtained by dividing the main imprint image for registration as in the registration process.

  The divided binary imprint image data for authentication is subjected to DCT conversion for each block by the same method as the registration process (S51). Next, a DCT coefficient is calculated (S52).

  Next, DCT codes corresponding to the DCT coefficients up to a predetermined number from the upper absolute value of the DCT coefficients are obtained (S53). This process is the same as the registration process, and a predetermined number of definitions are the same as the registration process. The process of S53 is a process for extracting the feature points of the seal impression image shown in one block.

  Based on the sign of the feature point, binary one-dimensional information represented by “0” and “1” is generated (S54). Since this process is performed in one block of the divided image data, the above process is performed for the number of blocks. This is performed for the number of blocks, and binary one-dimensional information for the number of blocks is generated. That is, S51 to S54 are repeated for the number of blocks.

  The binary one-dimensional information generated by the authentication information generation unit 10 is sent to the authentication unit 9. When sent, the binary one-dimensional information generated for each block is temporarily stored in a memory such as RAM until it is created, and sent to the authentication unit when processing for the number of blocks is completed. It may be sent to the authentication unit when it is generated for each block, and is not particularly limited.

  Upon obtaining the binary one-dimensional information, the authenticating unit 9 returns to the process of FIG. 13 and compares the binary one-dimensional template and the binary one-dimensional information among the called registration templates for each block (S37). ). For example, it is assumed that the block in the third row and the fourth column has binary one-dimensional information called ABCF for a binary one-dimensional template called ABCD. The hexadecimal data is compared. When converted to a binary number, the binary one-dimensional template of the block in the third row and the fourth column is 1010101111001101, and the binary one-dimensional information of the block in the third row and the fourth column is information of 1010101111001111. Contrast with each bit of the binary number. Since 15 points out of 16 points are matched, the matching rate is 93.75%. When the coincidence rate is obtained, the comparison with one block is completed. This is performed for the number of blocks, and the coincidence rate is calculated for each block.

  The calculated match rates of all the blocks and the average value of the match rates are sent to the display unit (S38).

  If the average value of the matching rate is 93% or more and each matching rate is 93% or more, authentication is permitted. If the average value of the match rate is less than 93%, the authentication is rejected. In either case of permission or rejection, the result is sent to the display unit.

  When the result that the average value of the coincidence rate is equal to or greater than the threshold value and there is a block with the coincidence rate of less than 93%, the background color of the corresponding block of the imprint image data is changed to such a degree that the imprint is visible. Further, a comment such as “re-examination required” is sent to the display unit (S39). At this time, the commented portion may be reconfirmed by implied or the like, but is not limited thereto.

  Here, the average value of the match rate of each block and the match rate for the number of blocks is 93%, but this is a threshold value of the match rate, and this value can be changed.

  The display unit displays the sent result (S40). An example of the displayed result is shown in FIG. FIG. 15 displays the binary imprint image data for authentication, the average coincidence rate, the coincidence rate of blocks that were less than the threshold, and comments.

  As mentioned above, although preferable embodiment of this invention was described, embodiment of this invention is not limited above, The following other preferable aspects can also be employ | adopted.

  In the process of S22, the position information of feature points obtained about 100 points may be stored with i as the horizontal and j as the vertical. The stored position information can be used without performing the feature point extraction process of the test seal impression in the authentication process described later.

  In the process of S24, the number of imprint pixels of the imprint image of the block corresponding to the binary one-dimensional template may be stored. That is, the binary one-dimensional template may be configured by the number of pixels and binary one-dimensional information.

  During the processing of S7, binary one-dimensional templates among the registration templates may be aggregated for the number of blocks, and may be created as one, or two or more binary one-dimensional templates may be synthesized to create a plurality. May be. Moreover, the combination method in the case of synthesizing two or more binary one-dimensional templates is arbitrary and is not particularly limited. Also, when collecting binary one-dimensional templates for the number of blocks and combining them into one, the scanning direction starts from the upper left block to the right in order from the upper left block, and when the first row ends, the scanning starts from the block on the left side of the next row. Turn to the right and scan until all blocks have been scanned. The scanning direction is not limited to this, and the block may be scanned obliquely or may be scanned from the upper right to the left.

  In the process of S34, instead of calling the registration template, the user may go to the sub-data table corresponding to the ID and call necessary information. Necessary information may be referred to each time, or information corresponding to the ID of the sub data table may be first extracted and stored in the authentication server.

  After the image processing is performed in the process of S50, before the process of S51, the number of pixels having an imprint in each block, that is, the number of pixels having the imprint of each block in the binarized imprint image is stored. You may keep it.

  In the process of S37, when the number of pixels is stored in the process of S50, not only the matching rate of each block but also the number of pixels is calculated as a difference between each block at the time of registration and at the time of authentication. At this time, if at least one of the number of pixels of the imprint image at the time of registration in each block and the number of pixels of the imprint image at the time of authentication is less than 50, it may be excluded from the authentication target. Good. It is also possible to perform authentication by obtaining the total reliability by fuzzy inference using the match rate and the difference.

  Hereinafter, fuzzy inference will be described.

  In the case of seal stamp authentication, it is generally considered that only the calculation of the coincidence rate with a template is sufficient, but depending on the shape of the seal impression in the block, the template alone may not be sufficient. Fuzzy reasoning is introduced for the purpose of preventing the matching rate of templates from becoming unstable due to blurring of imprints.

  In introducing fuzzy inference, the number of imprinted pixels of the corresponding block is used in addition to the template matching rate. A large pixel difference between the registration imprint for registration and the binary imprint for authentication means that there is a high possibility of different imprints, and if the pixel difference is small, the possibility of the same imprint is high. . This property is the same property as the template matching rate, but does not necessarily match. The template check is a check of feature points from the frequency plane, and the pixel difference check is a check from the image plane itself, and has the property of compensating for the missing portions. The fuzzy reasoning used is a direct method, and the parameters used are the template matching rate and the pixel difference. FIG. 16A shows a membership function for “template matching rate / large”, and FIG. 16B shows a membership function for “template matching rate / small”. 17 (a) shows a membership function for “pixel difference / large”. FIG. 17B shows the membership function for “pixel difference / small”.

  The rule parameter is a pixel difference between the template matching rate, the registration imprint image for registration, and the binary imprint image for authentication. The following rules are applied to the blocks for which the coincidence rate and pixel difference are obtained, the reliability of each block is obtained, and the final conclusion is obtained as a weighted average thereof.

  Rule 1: If the matching rate is large and the pixel difference is large, the reliability cannot be said.

  Rule 2: If the match rate is large and the pixel difference is small, the reliability is high.

  Rule 3: If the matching rate is small and the pixel difference is large, the reliability is low.

  Rule 4: If the matching rate is small and the pixel difference is small, the reliability cannot be said.

  These are generally called if and then rules. For example, in rule 1, the first half “if the match rate is large and the pixel difference is large” is called if, and the second half “the reliability cannot be said” is called “then”. This applies to a function in which the matching rate is large in FIG. 16A and a function in which the matching rate is small in FIG. 16B. FIG. 17A applies to a function when the pixel difference is large, and FIG. 17B applies to a function when the pixel difference is small. In the first half, the reliability area is determined by AND operation. When the reliability area is determined by the four rules, the result is ORed to obtain the center of gravity.

  For example, it is assumed that there is a block with a matching rate of 69% and a pixel difference of 54. FIG. 18 shows an example using fuzzy inference using the block matching rate and the pixel difference as parameters.

First, in each rule, the above parameters are applied to determine the reliability area.
The center of gravity when the determined reliability area is output by OR operation is the reliability of the block. When this center of gravity is obtained, the reliability is 0.2352. Thereby, the reliability of one block is determined.

  The reliability of all the blocks is obtained by performing the above method on other blocks, and finally the total reliability is determined by a weighted average. The following formula shows the formula for obtaining the total reliability.

  If S is the overall reliability, M (n) is the match rate of the nth block, and C (n) is the weight for the nth block,

It is. Here, the number 36 is an example of the number of divisions and is not fixed.

  C (n) reduces the influence of the instability of a block having a small pixel difference on the overall reliability.

Then,

Defined by

  The weight of the mathematical formula 8 depends on the number of imprinted pixels in each block. Even if the matching rate is low, the reliability is slightly low if the number of pixels of imprinted blocks is small. In order to obtain an average with the same weight as the reliability of the block having the number of pixels, it is not fair. Therefore, in order to eliminate these unfairness, weights depending on the number of pixels with imprints are attached based on experience. If the number of pixels with imprints is a sufficient block and the matching rate is lower than the threshold value, the block number, the number of pixels, the matching rate, and the wording “re-examination required” together with the numerical value of the total reliability according to Equation 7. Is output and the block is colored. An output example is shown in FIG. The total reliability shown in FIG. 19 is a real number of two decimal places represented between −1 and +1, the block number is a block number assigned when dividing, and the number of pixels is authentication. This is the number of imprint pixels corresponding to the block number of the binary imprint image data, and the coincidence rate is the coincidence rate at the block number. A comment is issued when the block number is less than a threshold value depending on the reliability.

The block diagram which shows an example of the system of this invention The block diagram which shows the other example of the system of this invention The figure which shows the example of a flow of the registration process of this invention (A) is an imprint image captured by the image capturing unit of the scanner, and (b) is a diagram showing a binary imprint image after image processing. The figure which shows the example of a flow of the binary one-dimensional template creation process of this invention The figure which shows the example of a division | segmentation of the seal image of this invention The figure which shows the feature point distribution of the DCT code after carrying out DCT conversion of one divided | segmented block of this invention The figure which shows the example of the binary one-dimensional template of this invention The figure which shows the example of the template for registration of this invention The figure which shows the example of the data table of this invention The figure which shows the example of the data table in the other aspect of this invention. The figure which shows the example of the data table in the other aspect of this invention. The figure which shows the example of a flow of the authentication process of this invention The figure which shows the example of a flow of the binary one-dimensional information creation process of this invention The figure which shows the example of a display of the authentication result of this invention The figure which shows the example of the membership function which concerns on the coincidence rate in the fuzzy reasoning of this invention The figure which shows the example of the membership function which concerns on the pixel difference in the fuzzy reasoning of this invention The figure which shows the example which derives the reliability of one block in the fuzzy inference of this invention The figure which shows the example of a display of the authentication result at the time of using the fuzzy reasoning of this invention

Explanation of symbols

1: Scanner 2: Authentication server 3: Database server 4: Communication network 5: Image capture unit 6: Transmission unit 7: Image acquisition unit 8: Input unit 9: Authentication unit 10: Information generation unit 11: Display unit 12: Transmitter 13: Data table

Claims (10)

A scanner having an image capturing unit for capturing a seal imprint, an authentication server having an input unit for inputting identification data, an information generating unit for generating registration information and authentication information, and an authentication unit for authenticating a seal image,
An information generation unit that generates registration information in the authentication server acquires registration imprint image data captured by the scanner and identification data input by the input unit, and the information generation unit acquires the acquired registration. A process for dividing the seal image data into a plurality of blocks, a process for performing DCT conversion for each block, a process for calculating a DCT coefficient, and the DCT coefficients up to a predetermined number from the upper absolute value of the DCT coefficient A process for obtaining a one-to-one address (address) and a DCT code, and a process for generating a binary one-dimensional template for the number of blocks from the DCT code,
Means for storing the generated binary one-dimensional template corresponding to the number of blocks, the acquired identification data, and the acquired registration imprint image data in association with each other;
An information generation unit for generating authentication information in the authentication server acquires the imprint image data for authentication captured by the scanner and the identification data input by the input unit, and the information generation unit converts the identification data into the identification data. Processing for calling corresponding registration imprint image data and binary one-dimensional template, alignment processing for performing alignment based on the correlation between the called registration imprint image data and authentication imprint image data, and the alignment processing Processing for dividing the obtained authentication imprint image data into a plurality of blocks, processing for DCT conversion for each block, processing for calculating a DCT coefficient, and a predetermined number from the top of the absolute value of the DCT coefficient A process of obtaining an address (address) and a DCT code corresponding to a DCT coefficient on a one-to-one basis, and generating binary one-dimensional information for the number of blocks from the DCT code Run the process,
The authentication unit compares the binary one-dimensional template corresponding to the identification data generated by the information generation unit that generates registration information and the binary one-dimensional information generated by the information generation unit that generates authentication information. Imprint authentication system characterized by authentication.   The imprint authentication system according to claim 1, wherein the binary one-dimensional template and the binary one-dimensional information for each block generated by the information generation unit include the number of pixels of the imprinted pixels in the configuration.   When the number of pixels with imprints in the binary one-dimensional template and / or binary one-dimensional information is equal to or less than the threshold, the authentication unit removes the target from the authentication target and performs the authentication using only the blocks larger than the threshold. The seal stamp authentication system according to claim 2, wherein the system is executed.   3. The authentication unit compares the binary one-dimensional template and the binary one-dimensional information, compares the number of pixels, and further authenticates using fuzzy inference based on the comparison result. Or the imprint authentication system of 3 description.   Means for storing the generated binary one-dimensional template for the number of blocks, the acquired identification data, and the acquired registration imprint image data in correspondence with a data table in the authentication server or the authentication server; The seal stamp authentication system according to claim 1, wherein the seal stamp authentication system is a data table in a provided database server.   6. The data table is configured by associating the generated binary one-dimensional template corresponding to the number of blocks, the acquired identification data, and the acquired registration imprint image data. Seal imprint authentication system.   The data table is configured by associating the generated binary one-dimensional template for the number of blocks, the acquired identification data, the acquired registration imprint image data, and user attributes. The seal stamp authentication system according to claim 5.   The data table includes a main data table and a sub data table, and the main data table includes the identification data, a sub data table name corresponding to the identification data, and registered seal image data. The seal imprint authentication system according to claim 5.   The sub-data table is information for generating the identification data, the block number of each block divided into the plurality of blocks corresponding to the identification data, the number of imprint pixels of each block divided into the plurality of blocks, and the registration information 9. The seal imprint authentication system according to claim 8, comprising a binary one-dimensional template generated by the generation unit.   Means for classifying storage in the sub-data table in the data table (1) a configuration corresponding to an outer peripheral shape of an imprint including at least three outer imprint shapes of a round mark, a square mark, and an oval mark; A configuration in which reading kana is discriminated from the edge of the character part and arranged in the order of the Japanese syllabary, or (3) a configuration corresponding to an imprint typeface that includes at least three imprint typefaces, a seal type, a typeface, and an old type The seal stamp authentication system according to claim 8 or 9, wherein the seal stamp authentication system is any one of the above.