JPH0644285B2 - Collation system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an image support carrier that carries reference position information that facilitates the matching of registered images and input images.

In Japan, seals are used to identify the person in almost every transaction and procedure. For example, in order to withdraw a deposit from a bank, it is sufficient to confirm the validity of the passbook and seal, and no further confirmation is required to confirm the identity. Seals are also used for loan lending certificates and are often used for business transactions. In addition, a stamp is required to issue a check. For this reason, it would be extremely effective if a simple verification system for a seal stamp by a computer was developed. However, it has been extremely difficult for a personal seal to be collated by a computer because the imprint image is not constant. However, since the image of the seal imprint is reproduced from the same seal stamp, it is logically possible to perform collation by a pattern matching method that identifies ordinary printed characters.

FIG. 1 shows some typical imprint images formed by the same conventional seal stamp. FIG. 7A shows an image of a seal imprint pressed in the normal direction, and an image in this state will usually be registered in the computer. The figure (b) shows the image of the seal imprint when tilted to the left, and the figure (c) shows the image of the seal imprint when tilted to the right. This kind of inclination usually occurs frequently, and if it is left as it is, the pattern matching processing with the image shown in FIG. In particular, since many stamps in practical use have a round outline, it is not possible to determine the inclination of the image of the seal image from the outline. In addition, in practical stamps, it is often difficult to determine the inclination of the image of the seal imprint from the character shape because the character form is often unique in order to distinguish the seal stamp of another person from the seal stamp of the self. . This is all the more true for important seals that require verification. Therefore, the method left behind is extremely complicated and long in that the contours of the images of both imprints are aligned and then the pattern of the image of one imprint is rotated 360 ° to detect the peak of pattern matching. It will take processing time. Moreover, according to such a method, one of the pseudo peaks generated in the middle of comparison may be mistakenly recognized as a true peak, and an incorrect collation may be established. Therefore, it is difficult to obtain high reliability of collation. Is.

An object of the present invention is to provide a collation system capable of collating images in a completely different manner even with one image support carrier.

Hereinafter, examples of the present invention will be described in detail.

FIG. 2A is a block diagram showing the configuration of the image matching system to which the present invention is applied. In the figure, 1 is a transaction document having an image 2 of a seal imprint of an image carrier of an embodiment, 3 is an image reader for reading the image 2 of the seal imprint by a solid-state image sensor CCD, and 4 is image information processing. A central processing unit (CPU), 5 is a disk device for storing a registered image of the imprint image 2, and 6 is a keyboard for inputting, for example, a bank account number or a personal identification code.

An example of the image matching process performed using the above configuration is to read an image 2 of a seal imprint of a seal having a reference mark in advance by an image reader 3 and store the seal imprint image in a registration file of a disk device 5 to register the image. The image registration for forming the image and another seal imprint image 2 formed from the same seal on another occasion are read by the image reader 3 and the seal imprint image is used as an input image, and the CPU 4 compares the registered image with the input image. In this case, after recognizing the positions of both reference marks, one of the imprint images is rotated and moved to perform matching processing of the images in a state where the positions of both reference marks are matched, thereby making a quick and accurate determination of matching. Consists of image matching.

FIG. 2B is a block diagram of the functions substantially executed by the CPU 4. The image read by the image reader 3 is displaced so that the position of the reference mark becomes a predetermined position. Is normalized by the normalizing means 4 ₁ . Known coordinate conversion processing is used for this normalization processing. 6 is
It is a password information input means for further adding a specific displacement if necessary, and an arbitrary displacement can be added using this displacement information as a parameter. 6 is not limited to a keyboard, but may be, for example, an ID card reader. When registering a reading obtained image is registered in Yotsute disk device 5 the normalized image by normalizing means 4 ₁ to registration means 4 _2. Matches image registered in the normalizing means 4 ₁ normalized image and disk device 5 in the time entered for the reading obtained image is matched in the collation unit 4 _3, matching the collation result between the two images The presence or absence of

FIG. 3 shows some images of seal imprints having a reference mark integrally as one embodiment of the present invention. Although this will be described later, the present invention is not limited to the seal stamp, and naturally includes an ID card having an ID mark and the like. That of FIG. (A) has two reference marks 2 ₁ projecting inwardly of the circular profile, have a two reference marks 2 ₂ protruding to the outside of the contour as in FIG (b) and, FIG ones (c) has two reference marks 2 ₃ forming a cut on the contour, the reference mark 2 ₄ things two consisting point symbols inside the contour the (d) of FIG Have. Since the reference mark described above is configured not to form the character portion of the seal stamp, the identification by the computer is easy.
The number of reference marks may be one, but there is an advantage in using a plurality of reference marks. This is because, in general, even if one of the fiducial marks fades because the stamp is not pressed uniformly and cannot be recognized, it is possible to perform quick verification with the remaining fiducial marks. Further, usually, if two reference marks are detected, the inclination of the image of the seal imprint can be grasped accurately, so that the matching process of the image becomes easy. Further advantages will be apparent from the description given below. In this way, the number and shape of the reference marks are not limited to those shown in the figure, and may be any as long as they can be identified as indicating the reference position. If the two reference marks have different shapes, the orientation of the image becomes clearer, which is more convenient for normalizing the displacement of the image to be described later. Figure 3 (e)
Ones having two reference marks 2 ₁ projecting inwardly of the rectangular outline. In this way, if the contour is a square, it is possible to know the inclination of the imprint to some extent from the shape of only the contour.
From this point of view, the part of the arrow where the contour of the imprint intersects with the internal character is regarded as the reference, as shown in FIG. It would be possible to iterate and obtain matching matches. However, in order to obtain a highly reliable collation result with constant accuracy and processing speed for general purposes, it is preferable to provide the reference marks as shown in FIGS.

The image of the above-mentioned seal imprint is read in advance by the image reader 3 and registered in the disk device 5. The image may be registered in the form of the read image as it is, but according to a preferable aspect of the collating device, the CPU 4 first identifies the reference marks from the image of the read imprint, and then the reference marks are detected. The entire imprint image is rotated and moved so as to have a predetermined arrangement. In this way, it is convenient both for collating the imprint images and for managing the registered images.

FIG. 5 is a flow chart showing the processing procedure for image registration of the image of the seal imprint, and FIGS. 4 (a) to (c) and (d) to (f) follow the procedure shown in FIG. It is an explanatory view showing a state where an image of a seal imprint is registered as an image. Less than,
The image registration process will be described in detail with reference to both drawings. In step 101, the image reader 3 reads, for example, the image 2 of the imprint formed on the document for registration [FIG. 4 (a)]. The image reader 3 has a built-in solid-state image sensor CC
The image 2 of the imprint is read at D and converted into an electrical analog signal. The analog signal is further amplified, compared with a predetermined threshold value, and converted into a logical binary signal consisting of ON / OFF. Alternatively, an A / D converter may be used to convert to a quantized code that can express light and shade. If this is done, the image of the seal imprint will be converted more faithfully, so that even the subtle characteristics of the stamp used, for example, can be accurately captured.
In any case, the digitalized image data is reproduced in its original form and stored in the memory of the CPU 4 as the imprint image shown in FIG. 4 (b). At step 102, the CPU 4 identifies the _two reference marks 22 from the stored imprint image. As described above, the two reference marks are formed in a unique form so that they can be distinguished from the character part of the seal stamp, so that the CPU 4 can easily identify them. The processing of the following steps 103 to 106 is processing in which the CPU 4 normalizes the inclination of the imprint image. In other words, setting the segment k connecting the step 103 two reference marks 2 ₂ was identified in a straight line, step 10
In step 4, the midpoint 0 of the two reference marks is obtained on this line segment k, in step 105 a line segment 1 is set on this midpoint 0 so as to be orthogonal to the line segment k, and in step 106 this midpoint is set. The entire imprint image is rotated in the arrow direction until the line segment k becomes horizontal with 0 as the center of rotation. The rotation process of the imprint image may be a well-known process based on the coordinate conversion process generally used in the process such as graphic display. The rotation process can be obtained by a single coordinate conversion calculation from the relative position with respect to the horizontally assumed line segment k. The direction of rotation is not limited to the direction of the arrow, but if
If the fiducial marks are of different shape, they will always be rotated, for example so that one mark is on the left and the other on the right.
Finally, in step 107, the center point 0 of the imprint image whose inclination is normalized is aligned with the center of the registered file 5 ₁ having a predetermined size as shown in FIG. 4 (c). be registered. It should be noted that each line segment is conceptual and is not actually registered as a pattern.

Next, the image of the seal imprint shown in FIG. 4 (d) is originally obtained from the same seal stamp as that shown in FIG. 4 (a), only the attachment position of the reference mark of the seal stamp is different. Is. When this is registered in the same manner as described above, the registration image is registered as shown in FIG. 7F after the state of the seal imprint image shown in FIG. Imprint the matching process of the image will be described in detail later, but here by a comparison of FIG. (C) the registration file 5 ₁ Fig. (F) of the registered files 5 ₂ so as to align the square two imprint images It is clear that no matching agreement can be obtained.

In this way, the image information of the present invention can be registered in a completely different form even between seals that are similar to each other. As described above, the only difference between the imprint images 2 shown in FIGS.
Only the position where ₂ is provided. Moreover since the image registration process of seal impression is based only on the position of the reference mark 2 ₂ actually where it has been pressed in the image of the seal forming the supporting carrier documents, also affiliated to whether pressed any inclination Instead, the two registered images maintain a remarkable difference by always maintaining the positional relationship as shown in FIGS. Therefore, if you do in this way, even if the seal stamps that are common and indistinguishable from the seal stamps used by other people, each one just adds a reference mark to its own place The image registration is done.

Next, the collation process of the seal imprint image is performed, for example, by processing the image of the seal imprint formed on another occasion by the same seal stamp used at the time of registration in the same procedure as that performed in the registration process. .

FIG. 6 is a flow chart showing a processing procedure for image collation of the image of the seal imprint, and FIGS. 4 (g) to (i) show a state in which the image of the seal imprint is collated according to the procedure of FIG. It is explanatory drawing shown with an image. Hereinafter, the image matching process will be described in detail with reference to both drawings. In step 210, the image 2 of the imprint shown in FIG. 4 (g) is read by the image reader 3. This image 2 of the seal imprint is formed on another document at another occasion by the same seal stamp as that shown in FIG. The read image data is reproduced in its original form by the CPU 4, and is stored in the memory as a seal imprint image shown in FIG. 4 (h). Step 2 following
The processing of 02 to 206 is performed by the step 102 of FIG.
The process is similar to that of steps 106 to 106, and a line segment k ', a midpoint 0', and a line segment l'that is orthogonal to the line segment k'on the midpoint 0'are provided so that each step corresponds. After that, the entire imprint image is rotated in the direction of the arrow with the center point 0'as the center of rotation until the line segment k'is horizontal. Finally segment k 'and Figure 4 in step 207 segment k and drawing on registration file _{5 2} shown in (f) (i), the line segment l and drawing the figure (f) (i) The mating processing of both imprint images is performed in a state where the line segment l'of
Here, it is obvious that matching of the collation can be obtained by comparing both the imprint images of (f) and (i). The reason _{for 2} suddenly registered file 5 to the other party of the match in the above description has been selected, for example, account number by the input connexion the registration file 5 ₂ from the keyboard 6 in the case of pulling out the bank deposits have been retrieved It is a thing. In this way, the positioning for verification is performed only based on the position of the reference mark as in the case of registration, so as long as it is an image of a seal imprint formed from the same seal, it is not stamped anywhere on the transaction document. No matter what the tilt, and at what angle it is pushed, you always get a quick, accurate and reliable matching match.

As described above, for example, it is possible to register completely different images for a similar seal stamp, and it is always possible to perform a reliable matching match for the same seal stamp. Therefore, the present invention can be applied to a highly reliable seal stamp collation system that brings relief to each transaction. And, such an effect is that the image information source on the image support carrier is provided with the reference mark, the image which is variously displaced based on the reference mark is allowed to be registered, and the reference mark is always accurate. This is only achieved by the configuration that enables accurate matching.

Furthermore, the present inventor proposes another novel image matching method on an image supporting carrier on the above-mentioned technique. One preferred embodiment thereof includes input of a personal identification code from the keyboard 6, which enables registration of a plurality of different imprint images from the same seal stamp in advance, and thereafter, at a different opportunity, the seal stamps obtained from the same seal stamp can be registered. According to the image and the input of the personal identification code, it is possible to perform a separate and independent verification. In this way, the user can set various transactions independently using the same seal stamp, and the security of this transaction can be separately guaranteed by inputting the personal identification code.

FIG. 8 is a flow chart showing the processing procedure for registering the image of the seal imprint image, and FIGS. 7A to 7C are explanatory diagrams showing the image of the state in which the image of the seal imprint is registered. Hereinafter, the registration process will be described in detail with reference to both drawings. First, in step 301, the CPU 4 waits for the input of the personal identification code from the keyboard 6. The personal identification code may be, for example, the date of birth of the user, or may be a combination of character and symbol codes. The point is that anything can be used as long as the CPU 4 can perform logical processing. In step 302, the CPU 4 obtains the rotation angle α of the registered image for specifying the image registration of itself from the input secret code. In step 303, the 7th image is read by the image reader 3.
Image 2 of the imprint formed from the image support carrier shown in FIG.
Is read, and the imprint image is reproduced in the memory of the CPU 4 in the as-read form as shown in FIG. 7 (b). In step 304, the CPU 4 identifies _two fiducial marks 22 from the imprint image. The following processing of steps 305 to 308 is processing in which the CPU 4 normalizes the inclination of the imprint image with the angle specified by the input of the personal identification code. In other words, setting the segment k connecting the step 305 two reference marks 2 ₂ was identified in a straight line, in step 306 obtains a middle point 0 of the two reference marks on the line segment k, in step 307 the midpoint A line segment m that intersects the line segment k at an angle α is set on O, and in step 308, the line segment m is set with this midpoint 0 as the center of rotation.
Rotate the entire imprint image in the direction of the arrow until is vertical. Finally, in step 309, to register with aligning the midpoint 0 of the imprint image in the center of the registration file 5 ₃ having a predetermined size as shown in FIG. 7 (c). Now, the image 2 of the seal imprint shown in FIG. 7 (d) is obtained from the same seal stamp as described above. For example, the person submits with the intention of registering individually to set up another transaction with the same seal. The image 2 of this imprint is also registered according to the processing procedure of FIG. However, in this case, only the input of the personal identification code that identifies the self registration is different. Therefore, in this case, β is obtained as the rotation angle of the imprint image, for example, by the processing of step 302 [FIG. 7 (e)].
Then, the entire imprint image is rotated in the direction of the arrow until the line segment m'is vertical. However, it is assumed that the rotation angle β is rotated from the line segment m ′ to the right. Step 309
In imprint As shown in FIG. 7 (f) the image is registered in the registration file 5 _4. It is obvious that no match can be obtained by comparing both the imprint images of FIGS. 13C and 13F. That is, although the seal stamp is originally the same, the seal imprint image is registered in a completely different form as shown in the figure.

Now, the image matching process is similar to this. The flow chart is shown in FIG. 9, and the image aspect is shown in FIGS. 7 (g) to (i). In step 401, input the PIN code, and in step 402
Then, β is obtained as the rotation angle of the image from the secret code. Here, the image 2 of the seal imprint shown in FIG. 7 (g) is shown in FIG. 7 (a).
And the same seal stamp as shown in (d). In step 403, the image reader 3 reads this, and the imprint image is reproduced in the memory of the CPU 4 in the as-read form as shown in FIG. 7 (h). In step 404, the CPU 4 identifies the fiducial mark. The subsequent processing of steps 405 to 408 is similar to the processing of steps 305 to 308 shown in FIG. 8, and the line segment k'and the midpoint 0'are set so that each step corresponds to
A line segment m'that intersects with the line segment k'at an angle β is set on the midpoint 0 ', and then the line image m'is made vertical until the line segment m'is vertical with the midpoint 0'as the center of rotation. The whole is rotated in the direction of the arrow. Finally, in step 409, FIG.
The matching process of both imprint images is performed in a state in which the line segment m on the registered file 54 shown in FIG. ₄ and the line segment m ′ in FIG. still,
Why register file 5 ₄ is for comparison, for example because the account numbers match. Here, it is obvious that matching of the collation can be obtained by comparing both the imprint images of (f) and (i). This is because the personal identification codes input at the time of registration of the image of the seal imprint shown in FIG. 9D and at the time of collation of the image of the seal imprint shown in FIG. The rotation angles were both calculated as β, and matching agreement was obtained. Even if the account numbers are mistaken during the collation and the imprint images of both (c) and (i) are compared, it is obvious that the collation cannot be matched. This is because the passwords of both imprints are different and the rotation angles of both imprint images are different, such as α and β.

As described above, by using the image supporting carrier of the present invention, a plurality of different types of important transactions can be set with the same seal. When making bank deposits, it is common to use a separate seal for each account opening. Of course, each household has its own account, but in the end there are many similar seals with the same surname, and it becomes difficult to know which one corresponds to which account. On the other hand, the matching of seals is strictly performed in banks.
Therefore, if nothing is done, the collation work will have to rely on the eyes of a skilled clerk. According to the present invention, it is possible to open a plurality of accounts with one single seal. Nevertheless, the privacy of each individual is sufficiently protected. This is because the security code guarantees the security of such transactions.
According to the present invention, a matching match for one's own account is always obtained, but it is not possible to obtain a matching match for another account. It is realized only by the remarkable difference between the registered imprint images based on the reference mark and the displacement given by the secret code. Only when the difference between the registered images is remarkable is it possible to perform quick and reliable automatic collation by the computer.

The above description is based on an example in which the image supporting carrier of the present invention is specified as a seal. However, the image-supporting carrier of the present invention is not limited to a seal stamp, and can be applied to all carriers supporting an identification mark used for identification. An example of this type of application is an ID card used for facility management, entrance control, and the like. In other words, it is natural from the above detailed description that even if the card is provided to this type of card, the setting position of the card in the reader can be freely set and the same card can be given a plurality of uses. .

As described above, according to the present invention, for example, even if the same seal stamps having different mark positions are used, the collation can be distinguished, and in the case of one seal stamp, a reliable collation can be performed no matter how it is pressed. A match is obtained. Therefore, there is an effect that collation of many similar seals having the same surname can be performed safely and surely.

Further, according to the present invention, when the personal identification information inputted is different even in one seal, the collation can be distinguished, and when the personal identification information is the same, a reliable collation matching can be obtained. Further, according to the present invention, one seal can perform different lighting for multiple transactions, and even if one's personal identification code coincides with another person's personal identification, if the mark position of the personal seal is different, Different verification will be performed. Furthermore, according to the present invention, since the inputted personal identification information is changed to the direction angle, it is only necessary to provide the collation system with a certain conversion algorithm, and it is not necessary to register the personal identification information itself.
On the contrary, there is an effect that the secret information cannot be easily known from the direction angle of the result.

The present invention includes various aspects other than the above-mentioned embodiments, the scope of which is specified only by the claims.

[Brief description of drawings]

1 (a) to 1 (c) are views showing images of some typical seal imprints formed by the same conventional seal stamp, and FIG. 2 (A) is an example of image matching to which the present invention is applied. 2 is a block diagram showing the configuration of the system, FIG. 2 (B) is a block diagram showing the functions executed by the CPU, and FIGS. 3 (a) to 3 (f) are embodiments in which a reference mark is integrated according to the present invention. Showing several images of the seal imprint of FIG. 4, FIGS. 4 (a) to (i) showing images of the displacement of the image of the seal imprint to be processed, and FIG. 5 showing a processing procedure of the seal image registration. Flow chart, FIG. 6 is a flow chart showing a processing procedure of seal imprint image collation, FIGS. 7 (a) to (i) are diagrams showing displacement of an image of a seal imprint, and FIG. 8 is a processing procedure of seal imprint image registration. FIG. 9 is a flow chart showing a processing procedure of seal image matching. Is Chiyato. Here, 1 ... Transaction documents, 2 ... Image of imprint, 3 ... Image reader, 4 ... Central processing unit, 5 ... Disk device, 6 ... Keyboard.

Claims (1)

[Claims] 1. An image supporting carrier having a plurality of reference marks on an image supporting surface, an information input circuit for inputting personal identification information, a setting circuit for setting a direction angle according to the input personal identification information, and reference image information in advance. A reference memory storing the image information, an image memory storing the image information obtained by reading the image of the image supporting surface with a scanner, and recognizing the image of the reference mark in the image information read from the image memory and recognizing the mark position. And a mark position detection circuit that specifies the mark position detection circuit, and the image information is rotated about a predetermined point on a straight line connecting the mark positions specified by the mark position detection circuit until the straight line makes a predetermined direction and the set direction angle. An arithmetic circuit, and a collation device including a collation circuit that collates the image information after the rotation with the reference image information in a state where the predetermined point and the predetermined point of the reference image information overlap each other. A collation system characterized in that