JP2006338331A - Device and method for registering slip - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To determine the authenticity of a slip even when the surface of the slip is smooth, to eliminate the need of strictly adjusting an illumination direction and the position/posture of the slip when detecting characteristics of the slip and to prevent a person familiar with a authenticity determination method from forging the slip. <P>SOLUTION: This slip registering device for authenticity determination has an identifier recognizing means for recognizing an identifier printed beforehand on the surface of a slip, a mark detecting means for detecting a mark printed beforehand on the surface of the slip, a feature extracting means for extracting a feature amount of the slip from a transparent image of the slip, and a registering means for registering the feature amount as a standard feature amount to be used for slip identification. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a paper piece registration apparatus and a paper piece registration method for authenticating paper pieces.

  In recent years, in order to prevent forgery of documents and protect copyrights, a technique for printing an invisible and non-tamperable pattern on paper has been developed. For example, Non-Patent Document 1 describes a technique for embedding information so as not to be visually recognized by cutting a part of a character. In Non-Patent Document 2, when a gray image is hard-copied by binarization display, the signature image is scanned and different pattern cells are recorded in the output image and the verification image in pixels corresponding to the signature portion. Describes a technique for secretly recording signatures.

  On the other hand, depending on the purpose of the document, the authenticity of the paper itself may be asked. For example, stamps, banknotes, transportation and box office tickets have social value on the piece of paper itself. There is no value in the copy. In some cases, copying the same content itself is prohibited. When such a document is used, it is necessary to identify the identity of the piece of paper. Non-Patent Document 3 and Non-Patent Document 4 are methods for specifying a paper piece from the characteristics of the paper piece itself. In this method, unevenness on the surface of the paper is detected by a reflected image, this feature is expressed numerically, and the authenticity of the paper piece is determined by comparing it with a feature printed in advance on the paper piece with a barcode. At the time of registration, a black dot representing an area where the unevenness is observed is printed. At the time of authenticity determination, this black spot is first detected, and then the feature is detected according to the detection result.

  Further, in Patent Document 1, identification information (tag) of a piece of paper is adopted from a large number of patterns such as dots that are irregularly distributed on the piece of paper, and a pattern signal detected from an identification object is used as a tag. A means for verifying by comparing with a pattern signal detected in advance as a characteristic of the above is disclosed.

United States Patent 4,785,290 Katsuaki Hayashi, Yoshihiro Ueda, Masashi Iwata, Haruhiko Kimura "A Proposal of Information Hiding to Hardcopy Using Text Processing" Journal of Information Processing Society of Japan Vol.44 No.11 (2003) Kazuhiro Oka, Yasuhiro Nakamura, Matsui, Koshio "Embedding signatures in hardcopy images using the density pattern method" IEICE Transactions Vol.J79-D2 No.9 pp.1624-1626 (1996) Eric Metois, Paul M. Yarin, Noah Salzman, and Joshua R. Smith, 擢 iberFingerprint Identification (Proceedings of the Third Workshop on Automatic Identification, Tarrytown, NY, March 2002, pp. 147-154) Joshua R. Smith, mperceptible Sensory Channels, (IEEE Computer, Vol. 37, No. 6, pp. 84-85, June 2004) Neil Gershenfeld, he Physics of Information Technology pp. 183-186, Cambridge University Press, 2000

  However, the techniques disclosed in Non-Patent Document 1 and Non-Patent Document 2 are techniques for embedding information in the printed content, and are not related to the printed paper itself. When copying on different paper with a high-resolution copying machine, the embedded information is copied in the same way. For this reason, it cannot be determined whether the paper is genuine or a copy.

  Further, the conventional techniques disclosed in Non-Patent Document 3 to Non-Patent Document 4 have three problems. The first problem is that the unevenness of the surface of the high-quality paper is very small, and it is difficult to optically detect the unevenness from the reflected image. In particular, when the coating is applied to make the surface smooth, it is almost impossible to detect irregularities. For this reason, the application target of this prior art is limited to a piece of paper having a relatively low quality and a rough surface.

  The second problem is that the reflected image changes depending on the direction of illumination. In order to capture the unevenness of the surface of the paper with the brightness of the reflected image, the illumination light needs to be incident obliquely rather than perpendicular to the surface of the paper piece. The brightness and darkness obtained in this way varies depending on the angle of illumination. For this reason, it is necessary to adjust illumination to an appropriate angle for each piece of paper.

  The third problem is that it is possible to forge anything that knows the feature extraction method and barcode printing method of the present technology. That is, forgery cannot be detected when black dots are printed in accordance with the present technology, features are extracted from the area, and the result is printed with a barcode. For this reason, unless the feature extraction method and the barcode printing method are concealed, the authenticity determination itself is impossible.

  Furthermore, since the prior art disclosed in Patent Document 1 employs identification information that causes irregular changes that cannot actually be replicated, it is difficult to forge even if a determination method is known. For this reason, there is a drawback in that strict alignment is required for determination in order to observe a limited number of points.

  A first problem to be solved by the present invention is to provide a paper piece registration apparatus and a paper piece registration method that can determine the authenticity of a paper piece even when the surface is smooth.

  A second problem to be solved by the present invention is to provide a paper piece registration apparatus and a paper piece registration method that do not require strict adjustment of the illumination direction and the position and orientation of the paper when detecting the characteristics of the paper piece. .

  A third problem to be solved by the present invention is to provide a paper piece registration apparatus and a paper piece registration method that are impossible for counterfeit even for a person who knows the authenticity judgment method and that can be easily judged.

  In order to solve the above-described problem, a paper piece registration apparatus according to the present invention includes an identifier recognition unit that recognizes an identifier printed in advance on a surface of a paper piece, a mark detection unit that detects a mark printed in advance on the surface of the paper piece, The image processing apparatus includes a feature extraction unit that extracts a feature amount of the paper piece from the transmission image, and a registration unit that registers the feature amount as a standard feature amount used for identifying the piece of paper.

  According to the present invention, a feature that identifies a piece of paper is detected using a transmission image instead of unevenness on the surface of the piece of paper. When a transmission image is used, a two-dimensional image based on the internal structure of the paper can be obtained even when the surface is smooth. This image contains sufficient features to identify the piece of paper. A transmission image is obtained by illuminating a piece of paper perpendicularly from the back.

  In the present invention, the feature quantity of a piece of paper is collectively managed together with the identifier of the piece of paper in a database on a remote server. When authenticating, the identifier printed on the piece of paper is recognized, and the standard feature amount corresponding to the identifier is compared with the feature amount newly extracted from the piece of paper. According to this method, anything not registered in the database is regarded as a freight.

  According to the present invention, it is possible to determine the authenticity of a paper piece having a smooth surface by using a transmission image. Further, it is not necessary to adjust the illumination direction when detecting the feature of the piece of paper. Further, by managing the feature amount and the feature extraction area in a database, it is impossible for a person who knows the authenticity determination method to forge, and the determination operation is easy.

Hereinafter, the present invention will be described in detail with reference to examples.
[First embodiment]
As a first embodiment of the present invention, a paper piece authenticity determination system using a network will be described with reference to FIGS.
First, FIG. 1 shows a configuration example of a paper piece registration / identification device (sometimes simply referred to as a paper piece identification device). This figure is a diagram showing a basic configuration example of the paper piece identification device used in each embodiment of the present invention. This apparatus is configured as a system including at least one computer and various input / output terminal devices connected to the computer via an interface. By causing a computer to execute a plurality of programs each describing a predetermined processing procedure, this system is caused to function as a paper piece registration / identification device. That is, the paper piece registration / identification device 100 is configured by connecting the transmission image photographing means 101, the reflected image photographing means 102, the computing means 103, the communication means 104, and the storage means 105 via a bus. The transmission image photographing means 102 is used to detect the feature of the piece of paper. The principle of the transmission image photographing means 101 will be described later with reference to FIG. 4A. The reflected image photographing means 102 is used for recognizing marks and identifiers printed on a piece of paper. Here, the identifier is a unique integer value assigned to each piece of paper. The mark is printed on a piece of paper in order to designate a region from which the feature amount of the piece of paper is extracted.

  As an example, the computing means 103, the communication means 104, and the storage means 105 are configured using a computer equipped with a predetermined program. Further, the transmission image photographing means 101 and the reflection image photographing means 102 constitute a part of an input / output terminal device for the computer, and A / D conversion is performed on the data of the piece of paper image obtained by these photographing means and the data is transmitted to the computer. It has a function.

  That is, the calculation means 103 is a general-purpose MPU (Micro Processing Unit), and identifier recognition processing 322 and 313, mark detection processing 324 and 314, feature extraction processing 325 and 315, distance calculation processing 318, and perturbation processing described in FIG. 316, used to execute the determination process 319. The communication means 104 is a connection interface to a general-purpose LAN such as IEEE802.11b. Data is exchanged with a server that manages a standard feature quantity database 209, which will be described later, via the communication stage means 104.

  The storage means 105 is a non-volatile memory such as Flash EPROM (Erasable Programmable Read Only Memory). The storage means 105 stores a program 106 that describes the processing procedures of the identifier recognition 106, mark detection 107, perturbation 108, feature extraction processing 109, distance calculation 110, and determination processing 111. Also, a processing procedure for use in registration processing to be described later and a processing procedure for use in authenticity determination processing are described in the program 106 so that they can be switched and used. The feature quantity 112 and the identifier 113 are also stored in the storage unit 105.

  The paper piece registration / identification device 100 performs a feature recognition process, a mark detection process, a feature extraction process, a distance calculation process, a perturbation process, and a determination process with a general-purpose MPU, thereby enabling the feature of the paper piece in the area specified by the mark. The quantity is output together with the identifier recognition result via the communication means.

  Note that the feature quantity 112 and the identifier 113 may be managed by the storage unit 105 of the computer main body (or server) constituting the paper piece registration / identification device 100, but are located at a remote location with respect to the paper piece registration / identification device 100. It is desirable to collectively manage with a database on another server connected via a network.

  FIG. 2 shows a configuration example of a paper piece authenticity determination system using such a network. This system is configured as an authenticity determination system 200 including a plurality of registration / identification devices including a plurality of computers and various terminal devices connected via a network. That is, the present system 200 includes a registration / identification processing server (or terminal device) having a printing device 202, a first registration / identification device 203, and a second registration / identification device 204 for a piece of paper 201, and a feature database management server. These servers (or terminal devices) are connected by a network 205. Each of the first registration / identification device 203 and the second registration / identification device 204 has a configuration corresponding to the registration / identification device 100 shown in the embodiment of FIG. However, the feature quantity database 209 of each registration / identification apparatus is different in that it is stored in a common feature quantity database management server 206 that is remote from the registration / identification apparatus main body and connected by a network. Needless to say, a server (or a terminal device) may be individually configured for each process of registration and identification.

  According to the present system 200, the feature amount database management server 206 executes identifier issue processing and identifier / mark printing processing using the printing apparatus 202. Further, the registration processing to the standard feature database is also executed by the feature database management server 206. Other registration processing is executed by the first registration / identification device 203.

  The authenticity determination process is executed by the second registration / identification device 204. However, the standard feature value calling process in the authenticity determination process is executed by the feature value database management server 206.

  FIG. 3 shows a more specific configuration and flow of each processing program of the first embodiment. In this example, the process is divided into three stages: printing, registration, and authentication.

  The printing process 301 is a process for printing a mark and an identifier necessary for registration processing and identification processing on a piece of paper 303 on the piece of paper. The registration process 307 is a process for extracting the feature amount of the paper piece 320 and registering it in the standard feature amount database 209. Here, the feature amount is one or more numerical values obtained from a transmission image of a piece of paper. The feature amount extracted at the time of registration is called a standard feature amount because it is used as a standard feature amount at the time of authenticity determination. Here, it is assumed that the mark 320 and the identifier are printed on the paper piece 320 in advance by the printing process 301. The authenticity determination process 302 is a process for determining the authenticity of the paper piece 310 based on the feature amount. Here, only a piece of paper registered in the standard feature database 209 in advance is determined to be genuine. If it is registered in advance, the authentication processing outputs TRUE, and if not registered, FALSE is output.

Hereinafter, each process of printing, registration, and authenticity determination will be described in detail.
In the printing process 301, first, the identifier issue process 305 refers to the standard feature database 209 and issues an unused integer value as a new identifier as a new identifier. Next, the identifier and the mark are printed on a piece of paper (304). A normal printer is used for printing. The identifier is printed on a piece of paper with a barcode or number. Details of the mark printing format will be described with reference to FIG.

  In the registration process 307, first, a reflected image of the paper piece 320 is captured (321). Subsequently, a transmission image is captured (323). The identifier printed on the piece of paper is recognized from the obtained reflection image or transmission image (322). Here, the existing barcode recognition technology or character recognition technology is used. Next, a mark is detected from the reflected image or transmitted image (324). Further, a feature amount is extracted from the obtained transmission image (325). At this time, the feature amount is extracted from the area designated by the mark. Next, the obtained feature quantity is registered with the identifier in the standard feature quantity database 209 (326).

  In the authenticity determination process 302, first, a reflected image of the piece of paper 310 is captured (311). Subsequently, a transmission image is captured (312). The identifier printed on the piece of paper is recognized from the obtained reflection image or transmission image (313). Next, referring to the obtained identifier, the standard feature quantity corresponding to the identifier is called from the feature quantity database 209 (317). Next, a mark is detected from the reflected image or transmitted image (314). The feature extraction region coordinates estimated from the mark position may include errors that occur during the printing and detection process. Therefore, the feature extraction area coordinates are shifted little by little in the perturbation processing 316, the feature quantity is extracted from the shifted area (315), and the degree of difference (distance value) between the obtained feature quantity and the standard feature quantity is calculated ( 318). One or more ways of shifting are stored in advance as perturbation parameters. Here, the Euclidean distance is used as a distance scale. Finally, the authenticity determination result is output in the determination process 319. In the determination process 319, if the obtained distance value is equal to or less than a preset threshold value θ under any of the perturbation conditions, the paper piece 310 is regarded as genuine and outputs TRUE. If the distance value is larger than the threshold value θ under any perturbation condition set in advance, the paper piece 310 is regarded as a bowl and FALSE is output.

  The imaging principle of the transmission image is shown in FIG. 4A. The light source 401 illuminates the paper piece 402 from the back. As described in Non-Patent Document 5, paper fibers are originally transparent, but incident light is scattered by a fine structure in a piece of paper. According to the inventor's observation, the way of light transmission at this time reflects the fine structure in the paper piece and subtle changes in the thickness of the paper piece, and the resulting transmission image is unique to each piece of paper. Become a thing. This image is formed on the two-dimensional image sensor 404 by the lens 403. With such a device configuration, a transmission image of a piece of paper can be taken.

  FIG. 4B shows an example of a transmission image of a piece of paper. This is a high-quality paper imaged at 600 DPI using a commercially available transmissive scanner. The shading is emphasized during printing.

  The processing procedure of the registration process is shown in FIG. First, a transmission image of a piece of paper to be registered is captured (501), and then a reflection image is captured (502). The identifier printed on the piece of paper is recognized from the obtained reflection image or transmission image (503). Next, a mark is detected from the reflected image or transmitted image (504). Next, a feature amount is extracted from the transmission image (505). Finally, the obtained feature quantity is registered in the standard feature quantity database 209 as a pair with the identifier (506).

  FIG. 6 shows the processing procedure of the authenticity determination process. First, a reflected image of the piece of paper 310 to be authenticated is taken (601). Subsequently, a transmission image is captured (602). The identifier printed on the piece of paper is recognized from the obtained reflection image or transmission image (603). Next, referring to the obtained identifier, the standard feature quantity corresponding to the identifier is called from the standard feature quantity database 209 (604). Next, a mark is detected from the reflected image or transmitted image (605). Next, the following processing is repeated in the loop 606 for the number of perturbation parameters. First, the mark detection result is corrected with a perturbation parameter (609). Next, a feature amount is extracted from the feature amount extraction region obtained by correction (610). Next, a distance value between the obtained feature quantity and the standard feature quantity is calculated (611). Next, if the distance value is equal to or smaller than the threshold θ (612), the determination result is set to TRUE (613), the process jumps to Label1 (608), and the process ends. If the loop ends, the determination result is set to FALSE (607), and the process ends.

  Note that the feature extraction processing is used in the registration processing 307 and the authenticity determination processing 302 in FIG. 3, but these are all the same processing.

FIG. 7 shows a detailed configuration example of this feature extraction processing.
In the feature extraction processing 701, first, based on the specified feature extraction region coordinates, a partial image is cut out from the transmission image and normalized (702). At this time, the image is projected and transformed in accordance with a rectangle (standard rectangle) having a predetermined size. Next, a low pass filter (LPF) is applied to the image to reduce noise (703). Next, histogram normalization is performed on the obtained image (704). This is a process of obtaining a histogram of pixel values and converting the pixel values so that the histogram becomes flat.

  By this processing, it is possible to correct a difference in characteristics between the transmission image capturing 323 at the time of registration and the transmission image capturing 312 at the time of authenticity determination. Finally, numerical features are extracted from the image obtained as a result of histogram normalization (704) (705). As the numerical feature, for example, the pixel value of each pixel is used. In this case, if the output of image cropping / normalization (702) is a rectangular RGB color image with horizontal w pixels and vertical h pixels, 3 × w × h feature quantities are obtained. As another feature example, a pre-designated filter output at a pre-designated coordinate is used. If necessary, a plurality of designated coordinates and / or filter types are preliminarily extracted, and a plurality of feature amounts are extracted.

  An outline of the image cutout / normalization process (702) is shown in FIG. In this embodiment, the feature extraction area is designated by the coordinates of the four corners in the transmission image. In this case, as shown in FIG. 8A, the feature extraction area is not necessarily a standard rectangle. It may also be tilted. Therefore, the pixel value in the feature extraction area is copied into a standard rectangle using projective transformation. At this time, the coordinates (P1, P2, P3, P4) of the four corners of the feature extraction area shown in (A) are the coordinates (P'1, P'2, P'2, P4) of the standard rectangle as shown in (B). Set projective transformation parameters to match P'3, P'4).

  An outline of the perturbation process 316 of FIG. 3 is shown in FIG. In this embodiment, n perturbation parameters δi (1 ≦ I ≦ n) are stored in advance. Each perturbation parameter δi consists of eight variables (δxi1, δyi1, δxi2, δyi2, δxi3, δyi3, δxi4, δyi4). Using each variable, the feature extraction region is changed as shown in FIG. 9A to FIG. 9B.

  In the loop 606 of the perturbation process 316, the feature extraction region is changed using n perturbation parameters one by one (609), and feature extraction (610) and distance calculation (611) are executed.

  FIG. 10 is an example of an experimental result in which δxi1 = δxi2 = δxi3 = δxi4 = δxi, δyi1 = δyi2 = δyi3 = δyi4 = δyi, δxI and δyi are varied, and the output value of the distance calculation processing 318 is measured. As can be seen from FIG. 10, with the appropriate perturbation parameters (in this example at the center of the figure), the distance value is significantly smaller, while in other cases (ie when the location is shifted), the distance value Is uniformly large. In addition, experiments have shown that the distance value obtained from the paper piece of the garment increases similarly. For this reason, using an appropriate threshold value, it is possible to distinguish between a real article and a bowl.

  FIG. 11 schematically shows an example of a mark in the present embodiment. In this example, three black dots 1102, 1103 and 1104 are printed as marks on a piece of paper 1101 together with a bar code 1105 representing an identifier as shown in FIG. This mark designates an area as shown by the hatched portion 1106 in FIG. The hatched portion 1106 is not printed on a piece of paper, but is used in the drawing to schematically show the feature extraction region.

  The three points on the mark are regarded as representing P1, P4, and P3 in FIG. The coordinates of P2, which is the remaining point, are estimated from P1, P4, and P3 based on the assumption that the feature extraction region is a parallelogram.

  According to the present embodiment, a feature that identifies a piece of paper is detected using a transmission image instead of the unevenness on the surface of the piece of paper. When a transmission image is used, a two-dimensional image based on the internal structure of the paper can be obtained even when the surface is smooth. This image contains sufficient features to identify the piece of paper. A transmission image is obtained by illuminating a piece of paper perpendicularly from the back. There is no change in the two-dimensional image depending on the illumination direction as in the method using the reflected light. For this reason, if a rotation / position correction is performed by image processing after obtaining a transmission image, it is possible to extract features necessary for authenticity determination. That is, the authenticity of a piece of paper with a smooth surface can be determined by using the transmission image. Further, when detecting the feature of the piece of paper, it is not necessary to adjust the illumination direction.

  In the present invention, the feature quantity of a piece of paper is collectively managed together with the identifier of the piece of paper in a database on a remote server. When authenticating, the identifier printed on the piece of paper is recognized, and the standard feature amount corresponding to the identifier is compared with the feature amount newly extracted from the piece of paper. According to this method, anything not registered in the database is regarded as a freight. For this reason, even a person who knows the authenticity determination method of this method cannot forge. That is, by managing the feature amount and the feature extraction area in a remote database away from the apparatus main body, it becomes impossible for a person who knows the authenticity determination method to forge, so that the reliability of the system can be improved.

  In addition, since a mark for alignment is detected using a two-dimensional sensor such as a two-dimensional image sensor and the characteristics of the transmitted image are detected accordingly, strict alignment is not necessary, and determination operation is easy. Can be done.

  Furthermore, according to the present embodiment, processing speed can be improved by performing distributed processing of registration / identification by a plurality of registration / identification devices connected via a network. Therefore, it is possible to construct a system suitable for authenticity determination processing for a large amount of paper pieces. Further, since standard feature amounts corresponding to a plurality of registration / identification devices are managed by a common feature amount database management server, the reliability of the system can be improved.

[Second Example]
A second embodiment of the present invention will be described with reference to FIGS. First, the configuration of a paper piece authenticity determination system 1200 of this embodiment is shown in FIG. This system includes a registration device (server or terminal) 1201, an authenticity determination device 1202 (server or terminal), and a feature database management server 1204, which are connected via a network 1203. The paper piece registration / identification device 1200 according to the present embodiment collectively manages the coordinates of the feature extraction area in a database, and the feature quantity database management server 1204 includes a standard feature quantity database 1205 and a feature extraction area coordinate database 1206. ing. The identifier issue and standard feature registration processing in the registration process are executed by the feature database management server 1204. The identifier / mark printing, transmission image capturing, and feature extraction are executed by the registration device 1201. The standard feature call processing in the authenticity determination process is executed by the feature database management server 1204. Processing other than that is executed by the authenticity determination device 1202.

  FIG. 13 shows the configuration and processing flow of the second embodiment. In the registration process 1301, first, a new identifier is issued with reference to the standard feature quantity database 1311 (1304), and the identifier is printed on a piece of paper 1302 (1303). Next, a transmission image of the paper piece is captured (1305). Next, a feature extraction region is set at random in the feature extraction region coordinate setting processing 1306, and a feature amount is extracted from the transmission image (1307). The obtained feature quantity is registered in the standard feature quantity database 1311 together with the identifier (1308). Further, the feature extraction region coordinates are registered in the feature extraction region coordinate database 1310 as a pair with the identifier (1309).

  In the authenticity determination processing 1312, first, a reflected image of the paper piece 1313 is captured (1314), and further a transmitted image is captured (1317). Using the obtained reflection image or transmission image, the identifier printed on the paper piece 1313 is recognized (1315). Next, feature extraction region coordinates corresponding to the identifier are called from the feature extraction region coordinate database 1310 (1318). The called feature extraction area coordinates are perturbed (1320), and feature quantities are extracted from the transmission image (1321). On the other hand, the standard feature quantity corresponding to the identifier is called from the standard feature quantity database (1316). A distance value (1321) between the feature value (1319) obtained from the paper piece 1313 and the standard feature value is calculated. In the determination processing 1322, if the obtained distance value is equal to or less than a preset threshold value θ under any of the perturbation conditions, the paper piece 1313 is regarded as genuine and outputs TRUE. If the distance value is larger than the threshold value θ under any perturbation condition set in advance, the paper piece 1313 is regarded as a bowl and FALSE is output.

  According to the present embodiment, it is possible to determine the authenticity of a paper piece having a smooth surface by using a transmission image. Further, when detecting the feature of the piece of paper, it is not necessary to adjust the illumination direction. Formal management of feature amounts and feature extraction areas in a remote database remote from the main body of the device makes it impossible for a person who knows the authenticity determination method to forge. In addition, since the coordinates of the feature extraction area itself are collectively managed in the database, the safety is further improved. Furthermore, since a mark for alignment is detected using a two-dimensional sensor and the feature of the transmitted image is detected in accordance with this, a precise alignment is not necessary.

[Third embodiment]
A third embodiment of the present invention is shown in FIG. In this embodiment, the present invention is applied to identification of an identity of a paper piece, for example, a plurality of kinds of official paper pieces such as stamps, banknotes and tickets. In this embodiment, a process of specifying which of a plurality of types of paper pieces registered in advance corresponds to a paper piece to be identified. The hardware configuration of the paper piece authenticity determination system of this embodiment is the same as that of the embodiment of FIG. 2, for example, and it is assumed that a program for identity identification processing is added to the storage means.

  In the registration process 1401, first, a mark is printed on a piece of paper 1403 (1404), and a transmission image is taken (1406). A feature quantity is extracted from the obtained transmission image (1407), and is registered in the standard feature quantity database 1409 together with the identifier obtained in the identifier issuing process (1405) (1408). The registration process 1401 is executed for various paper pieces, and a plurality of paper pieces are registered in the standard feature amount database 1409. In the paper piece identification processing 1402, first, a reflected image is taken from the paper piece 1410 (1411), and further a transmission image is taken (1412). Next, a mark is detected from the obtained reflection image or transmission image (1414), perturbed (1416), and a feature quantity is extracted from the transmission image (1415). A distance value between the obtained feature quantity and the feature quantity in the standard feature quantity database 1409 is calculated (1418), and a combination of a perturbation parameter and an identifier having the smallest distance is obtained (1419). The identifier of the result is used as the paper piece identification result. As a result, from the plurality of types of paper pieces registered in the standard feature quantity database 1409, the one most similar to the paper piece 1410 can be found.

[Fourth embodiment]
FIG. 15 shows the flow of processing of the paper piece registration / identification apparatus according to the fourth embodiment of the present invention. There is no paper piece registration / identification device of this embodiment corresponding to the standard feature database 209 of the first embodiment. Instead, in this embodiment, the standard feature amount of the paper piece is printed on the paper piece by a barcode or the like. Other configurations are the same as those of the first embodiment.

  When authenticating, the standard feature is recognized from a piece of paper and then compared with the feature obtained from the transmission image. In the registration process 1501, first, a transmission image of a piece of paper 1502 is captured (1503), feature amounts are extracted (1504), and feature amounts and marks are printed (1505). If necessary, standard feature values are encrypted and printed.

  In the authenticity determination processing 1506, first, a reflected image of the paper piece 1507 is captured (1508), and further, a transmission image is captured using a two-dimensional sensor (1150). A standard feature amount printed on a piece of paper is recognized from the obtained reflection image or transmission image (1509). Next, a mark on a piece of paper is detected (1511), the obtained feature extraction region is perturbed (1513), a feature quantity is extracted (1512), and a distance value from the standard feature quantity is calculated (1514). The feature extraction region coordinates estimated from the mark position may include errors that occur during the printing and detection process. Therefore, the feature extraction area coordinates are shifted little by little by perturbation processing (1513), feature quantities are extracted from the shifted areas (1512), and the degree of difference (distance value) between the obtained feature quantities and the standard feature quantities is calculated. (1514). In the judgment process (1515), the authenticity judgment result is output. In the determination process, TRUE is output if the distance value is equal to or smaller than the threshold value θ, and FALSE is output if there is no perturbation parameter whose distance value is equal to or smaller than θ.

  According to the present embodiment, it is possible to determine the authenticity of a paper piece having a smooth surface by using a transmission image. Further, when detecting the feature of the piece of paper, it is not necessary to adjust the illumination direction. By managing the feature amount and feature extraction area in a database, it becomes impossible for a person who knows the authenticity determination method to forge. In addition, since the mark is detected using a two-dimensional sensor and the characteristics of the transmission image are detected according to the mark, strict alignment is not necessary. Further, since the standard feature quantity of the paper piece is printed on the paper piece, the standard feature quantity database is not required.

[Fifth embodiment]
In each of the above embodiments, the mark and the identifier or the feature amount are printed separately on a piece of paper. As a fifth embodiment of the present invention, they can be printed with the same barcode. In the example of FIG. 16A, the bar code 1602 printed on the piece of paper 1601 has the coordinates of the feature extraction area together with the identifier or feature amount. In this case, the coordinate value is described in a relative position with respect to the barcode. Thus, if the barcode is read, a feature extraction region as shown by the hatched portion 1603 in FIG. 16B can be obtained.

  The standard feature quantity database and feature extraction area coordinate database in each embodiment of the present invention described above do not necessarily have to be provided in a database management server that is physically remote from the registration / identification apparatus main body. General operators, including those who know the authenticity judgment method, apply various restrictions to the authority to access the standard feature database and feature extraction area coordinate database, so that the feature database management server is physically registered with the registration / identification device. Specifically, it may be configured to be installed at a close position.

It is a figure which shows the basic structural example of the paper piece identification device used for each Example of this invention. It is a figure which shows the structural example of the authenticity determination system which becomes the 1st Example of this invention. It is a figure which shows the structure of the process of the 1st Example of this invention. It is a figure which shows the imaging principle of the transmission image in a 1st Example. It is a figure which shows the example of the permeation | transmission image in a 1st Example. It is a figure which shows the procedure of the registration process in a 1st Example. It is a figure which shows the procedure of the authenticity determination process in a 1st Example. It is a figure which shows the structure of the feature extraction process in a 1st Example. It is a schematic diagram of the image cutout / normalization process in the first embodiment. It is a schematic diagram of the perturbation process in a 1st Example. It is a figure which shows the relationship between the distance value and the amount of perturbations in a 1st Example. It is a figure which shows the example of the mark in a 1st Example. It is a figure which shows the system configuration example of the 2nd Example of this invention. It is a figure which shows the structure of the process of the 2nd Example of this invention. It is a figure which shows the structure of the process of the 3rd Example of this invention. It is a figure which shows the structure of the process of the 4th Example of this invention. It is a figure which shows the structure of the process of the 5th Example of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Paper piece registration / identification apparatus, 101 ... Transmission image photographing means, 102 ... Reflection image photographing means, 103 ... Calculation means, 104 ... Communication means, 105 ... Storage means, 200 ... Authentication judgment system, 201 ... Paper piece, 202 ... Printing Device 203 ... first registration / identification device 204 ... second registration / identification device 205 ... network 206 ... feature amount database management server 209 ... standard feature amount database 301 ... registration processing 302 ... authentication determination Processing, 303 ... Paper piece, 307 ... Print processing, 310 ... Paper piece, 320 ... Paper piece, 401 ... Light source, 402 ... Paper piece, 403 ... Lens, 404 ... Two-dimensional image sensor, 1101 ... Paper piece, 1102, 1103, 1104 ... Mark, 1105: Bar code, 1106: Feature extraction area.

Claims (10)

  Identifier recognition means for recognizing an identifier printed in advance on the surface of the paper piece; mark detection means for detecting a mark printed in advance on the surface of the paper piece; and feature extraction means for extracting a feature quantity of the paper piece from the transmission image of the paper piece. And a registration means for registering the feature quantity as a standard feature quantity used for identifying the piece of paper.   2. The paper piece registration apparatus according to claim 1, further comprising a standard feature database that registers the standard feature and the identifier as a pair.   2. The feature extraction area setting means for randomly setting a feature extraction area, a feature extraction means for extracting a standard feature amount in the feature extraction area of the paper piece from a transmission image of the paper piece, and the feature extraction area. A paper piece registration apparatus comprising standard feature quantity registration means for registering the coordinates in the feature extraction area coordinate database and registering the standard feature quantity of the paper piece in the standard feature quantity database.   The paper piece registration apparatus according to claim 1, wherein the registration unit has a function of printing a standard feature amount of the paper piece on the paper piece.   The paper piece registration apparatus according to claim 1, wherein the feature amount of the paper piece is a feature amount obtained by numerically extracting the feature of the paper piece from the transmission image of the paper piece.   3. The server according to claim 2, further comprising a remote server connected to the paper piece registration / identification device via a network, wherein the standard feature database is installed in the server. Paper piece registration device.   7. The paper piece registration apparatus according to claim 6, wherein a print function for printing a mark and an identifier necessary for identification processing on the paper piece is installed in the server.   The paper piece registration apparatus according to claim 6, wherein the server has a function of issuing the identifier, a function of registering the coordinates of the feature extraction area, and a function of registering the standard feature amount. An identifier recognition means for recognizing an identifier; a mark detection means for detecting a mark printed on the surface of the paper piece; a feature extraction means for extracting a feature quantity of the paper piece from the transmission image of the paper piece; A paper piece registration method in a paper piece registration / identification device registered in the standard feature quantity database as a standard feature quantity used for
Taking a transmission image of the piece of paper,
Recognizing an identifier printed in advance on the piece of paper from the obtained transmission image,
From the transmission image, a mark printed in advance on the piece of paper is detected,
Extracting a standard feature amount of the piece of paper from an area designated by the mark in the transmission image;
A piece of paper registration method, wherein the extracted standard feature is paired with an identifier and registered in a standard feature database. The paper piece registration / identification device according to claim 9, wherein the paper piece registration / identification device includes a server in a remote location connected via a network, and the standard feature amount database is installed in the server,
By the server
Issuing the identifier and printing the identifier along with the mark on the piece of paper;
A paper piece registration method characterized in that a feature of the paper piece is numerically extracted from the transmission image of the paper piece and is registered as the standard feature quantity of the paper piece in the standard feature quantity database.

Images