Patent Application
20080151324
This application is based on and claims priority under 35 USC 119 from Japanese Patent Application No. 2006-344675, filed Dec. 21, 2006.
1. Technical Field
The present invention relates to a surface-reading apparatus, a subject verification apparatus and a storage medium at which a subject verification program is stored.
2. Related Art
There are cases in which the surface of a subject is read and a characteristic of a surface condition of the subject that has been read is utilized in carrying out, for example, inspection or the like of an article. In recent years, surface matching technology has attracted attention as a technology which can enable assurances of originality of printed materials and prevention of leaks of information.
A surface matching technology is, for example, if subjects are papers, a technology for registering an image of a fiber structure of a portion of a paper at which an original has been printed, the image serving as a characteristic particular to that paper, and when a document is to be matched, comparing an image of a fiber structure of the document with the particular characteristic of the paper to judge whether or not that document is the original.
Because distributions of fibers in paper are random, it is thought that the probability of there being two papers in which the states of entanglement of fibers are the same is extremely small. Therefore, surface matching technologies are considered extremely good as a method for judging authenticity of paper documents.
However, depending on types of subjects, reading the characteristics of surface conditions may be difficult. For example, with a high-quality paper such as a coated paper, because the surface is smooth, it is difficult to read a characteristic of a surface condition as is, and application of a surface matching technology may be difficult.
According to an aspect of the invention, there is provided a surface-reading apparatus that includes a subject-flexing mechanism and a surface-reading component. The subject-flexing mechanism causes a subject to flex in one of a convex form and a concave form. The surface-reading component reads a characteristic of a surface condition of the subject that has been flexed by the subject-flexing mechanism.
Exemplary embodiments of the present invention will be described in detail based on the following figures, wherein:
Herebelow, examples of embodiments of the present invention will be described in detail with reference to the drawings.
—Structure—
A document verification apparatus 100 is an example of a subject verification apparatus of an exemplary embodiment of the present invention, and is used for verification of authenticity of documents (paper documents such as securities certificates, various kinds of title deeds, contracts, insurance certificates, residency registrations, birth certificates, warranties, travel tickets, bank notes, confidential papers and the like, and ID cards and the like) which are examples of subjects for exemplary embodiments of the present invention. Herein, it is sufficient for the subject to be a subject of which a surface can be read. As well as documents which are formed of papers such as high-quality paper, ordinary paper and the like, materials which can be flexed are included, such as various types of film, metal foils, thin metal plates and planographic printing plates. Electronic papers are also included as subjects.
As shown in
In the document verification apparatus 100, instead of the flexing roller 6, a flexing guide 16 may be provided, as shown in
Below, structure of the document verification apparatus 100 will be described in further detail.
As shown in
Here, when the document P is flexed, a radius of curvature in the range of about 15 to 30 mm is preferable. Therefore, a radius of the flexing roller 6 in the range of about 15 to 35 mm is preferable. However, this is just an example; the radius of curvature will differ in accordance with the subject.
The light source 10 is fixed at a position from which light hits a curved portion of the document P when the flexing roller 6 is raised and the document P is flexed, as shown in
An LED, halogen lamp, fluorescent light, xenon lamp or the like can be employed as the light source 10.
As shown in
The judgment computer 20 is an example of a computer for executing a subject verification program of the present invention, and is equipped with a CPU 21 and a memory 28, as shown in
At the CPU 21, when a signal indicating that the document P has been detected is inputted from the optical sensor 14, the control circuit 24 outputs control signals to the flexing roller-raising/lowering mechanism 7 and the light source 10, causing the flexing roller 6 to be raised toward the document P and at the same time lighting up the light source 10. Then, the control circuit 24 inputs a measurement commencement instruction to the signal processing circuit 22.
The signal processing circuit 22 receives the measurement commencement instruction from the control circuit 24, receives image signals including a non-reproducible random pattern which has been read by the reading section 12, performs the predetermined signal processing, such as amplification and the like, and then outputs results to the characteristic value extraction section 26. That is, image data that the reading section 12 has read from the document P is inputted to the characteristic value extraction section 26. Herein, the random pattern may utilize a fiber distribution which represents a dispersion state of fibers in the document P, disordered portions, which are portions at which printing on the document P is disordered, thickness variations of the document P, and so forth.
The characteristic value extraction section 26 performs extraction of the characteristic of the printed image from the inputted image data. The characteristic extraction is performed by, for example, the following procedure.
The reading results from the reading section 12 are divided up into meshes of a suitable size (a number of meshes d equals a height M×a width N) and quantized, a density of each mesh is represented by a density level q and sampled, and the reading results are converted to a mosaic-form image. From the image which has been quantized and sampled thus, if the density of a j-th mesh is xj, this pattern is represented by a characteristic vector x=(x1, x2, xd)·t (t being a transposition vector). A density of the corresponding image region is provided by each element of the characteristic vector. The pattern that is obtained is represented as a single point in a characteristic space that is spanned by the characteristic vector.
When the original of a document P is imaged, the characteristic value extraction section 26 finds the characteristic vector through the procedure described above, and saves the obtained characteristic vector to the memory 28 together with an identification number of the document P, to serve as characteristic information of the original. A method for associating information representing characteristic vectors with identification numbers is not particularly limited in the embodiments of the present invention but could be implemented by, for example, a table representing correspondences between characteristic vectors and identification numbers, and a portion or more of a data name of the information that a characteristic vector represents could be used in the identification number.
The comparison section 30 compares information representing the characteristic vector inputted from the characteristic value extraction section 26 (referred to as a ‘calculated characteristic vector’) with a characteristic vector that is the characteristic vector of the original which has been registered in the memory 28 (referred to as a ‘registered characteristic vector’), and determines whether or not the document P is the original in accordance with a degree of similarity; that is, the comparison section 30 judges authenticity of the document P. The degree of similarity between the calculated characteristic vector and the registered characteristic vector which is used for this authenticity judgment can be found by calculating a distance between the calculated characteristic vector and the registered characteristic vector (a Euclidean distance, a Mahalanobis distance or the like). The shorter the distance that is found, the more similar the two vectors are shown to be. In the comparison section 30, of the registered characteristic vectors registered in the memory 28, the calculated characteristic vector may be compared with only a registered characteristic vector with a matching identification number (matching), or may be compared with all the registered vectors (identification).
For the present exemplary embodiment, an example of a case in which authenticity of the document P is judged by a distance between the calculated characteristic vector and the registered characteristic vector is described, the authenticity may be judged from an angle between the vectors. Furthermore, other than mosaic processing as described above, it may be possible to directly match images obtained by the image capture element and evaluate degrees of similarity by correlation values, cumulative squared errors and the like.
Further, other than identifying and/or matching the image obtained by the reading section 12 in real space, it may be possible to, for example, transform the obtained image into the frequency domain by a two-dimensional Fourier transform and identify or match the images in Fourier space. In such a case, a pre-registered image and an image of the subject printed article are combined in Fourier space, a correlation strength image is obtained by a reverse Fourier transform, and a degree of similarity of the two images can be evaluated from a peak value of this image. For example, if the size of an amplitude peak matches or exceeds a pre-set threshold, it is judged that the images match, that is, that the printed articles are the same.
Further, other than identifying/matching by an image data level as described above, the identification/matching may be implemented by the level of an extracted characteristic. For example, there are methods of calculating centers of gravity of microscopic points which are arranged in stripes (ink) and using distances between the centers of gravity, or positions thereof or the like, as characteristics. Such a method can describe a characteristic with less data than a data amount that is ordinarily handled for an image data level.
The comparison section 30 inputs a signal representing the authenticity judgment result of the document P, which has been judged by comparison of the calculated characteristic vector with the registered characteristic vector, to the judgment result signal output section 32. The judgment result signal output section 32 may cause the authenticity judgment result to be displayed at a display component, such as an LCD display, or may cause predetermined processing to be commenced at a downstream device or cause predetermined processing to be prevented.
At the judgment computer 20, an operation for judgment of a document P by the described procedure can be implemented by the following sequence of operations in the CPU 21 and the memory 28, in accordance with provision of a medium.
Herein, in the judgment computer 20, the functions of the signal processing circuit 22, the control circuit 24, the characteristic value extraction section 26, the comparison section 30 and the judgment result signal output section 32 provided at the CPU 21 may be implemented by a computer program. An example of this judgment computer 20 is shown in
In the judgment computer 20 which is formed to realize these functions with a computer program, a part or all of the functions of the signal processing circuit 22, the control circuit 24, the characteristic value extraction section 26, the comparison section 30 and the judgment result signal output section 32 can be implemented by a subject verification program 150, which is a computer program. The subject verification program 150 itself, data that is used with the subject verification program 150 and suchlike can be stored at a storage medium which is readable by the computer. As shown in
The subject verification program 150 is saved to the storage medium. Then, the recording medium is mounted and the subject verification program 150 which has been saved thereto is read out by, for example, the reading section 166 or an interface 174 of the judgment computer 20 and stored in an internal memory 162 or a hard disk 170. The subject verification program 150 is executed by a CPU 164 and can realize the functions of the signal processing circuit 22, the control circuit 24, the characteristic value extraction section 26, the comparison section 30 and the judgment result signal output section 32. The judgment computer 20 may also be connected with various other devices via an interface 168, and can be connected with, for example, a display device which displays information, an input device at which a user inputs information, and the like.
Naturally, it may be possible for a portion of the functions to be constituted by hardware or for all to be constituted by hardware. Furthermore, it may be possible to constitute a program which includes an exemplary embodiment of the present invention together with other structures.
—Operation—
Next, operation of the present exemplary embodiment will be described.
In order to judge authenticity of documents P at the document verification apparatus 100, characteristic vectors of the originals of the documents P must be pre-registered beforehand.
Registration of the characteristic vectors is carried out in accordance with the following procedure.
The original of a document P is nipped by the conveyance roller 2 and conveyed through the document verification apparatus 100 in the conveyance direction a. When the document P is nipped by the conveyance roller 4 as shown in
In step S100 of
When the flexing roller 6 has risen and the document P has curved, in step S1104, an identification symbol of the document P is acquired, and in step S106, an image within an observation region is read by the reading section 12.
The identification symbol of the document P may be inputted by an operator from an input component such as, for example, a keyboard or the like, and may be acquired by reading an image of a region that includes the identification symbol from the document P with the reading section 12 and performing OCR (optical character recognition) processing on results of this reading.
For reading of the image, an instruction for reading of the printed image is sent from the control circuit 24 to the reading section 12, an image in the observation region S is read by the reading section 12, and a signal representing results of this reading is received at the signal processing circuit 22. Predetermined signal processing is performed and image data representing the image within the observation region S is obtained. As mentioned earlier, this image data includes a non-reproducible random pattern from a time of printing.
When the image data has been acquired, in step S108, the image data is quantized into pre-specified steps and sampled by the characteristic value extraction section 26, to be converted to a mosaic image. Then, in step S110, a characteristic vector is calculated from the quantized and sampled image data.
When the characteristic vector of the image has been calculated, in step S112, data representing the characteristic vector is associated with the identification symbol that was acquired in step S104 and is saved to the memory 28 together with the identification symbol, and registration processing of the original is completed.
According to the procedure described above, as a characteristic quantity of a document P that has been loaded at the reading section 12, a characteristic vector of an image in the observation region S (i.e., the registered characteristic vector) is registered to the memory 28 in association with that document P.
Here, the document P and the registered characteristic could be associated by encoding data representing the registered characteristic (the registered characteristic vector) or the like and printing the data onto the document P itself. In such a case, the memory 28 may be omitted.
Next, a procedure for judging authenticity of a document P by matching will be described.
A document P that is to be matched is nipped by the conveyance roller 2 and conveyed through the document verification apparatus 100 in the conveyance direction a. When the document P is nipped by the conveyance roller 4 as shown in
In step S120 of
When the flexing roller 6 has risen and the document P has curved, in step S122, an identification symbol of the document P is acquired, and in step S124, an image within the observation region S is read by the reading section 12. Then, image data representing the image in the observation region S, which is obtained from results of the reading, is quantized into pre-specified steps, sampled and converted to a mosaic image in step S126, and the characteristic vector is calculated in step S128. Because the processing of steps S120 to S128 is similar to the registration processing (steps S100 to S108 of
When the characteristic vector has been calculated, in step S130, of all the registered characteristic vectors which have been registered in the memory 28, the registered characteristic vector corresponding to the identification symbol acquired in step S122 is selected and read out by the comparison section 30. In step S132, the characteristic vector calculated in step S1128 and the registered characteristic vector that has been read out are compared by the comparison section 30. If the result of this comparison is that a degree of similarity of the two vectors matches or exceeds a predetermined threshold specified in advance, processing flows from step S134 to step S136 and it is judged that the matching object document P is the ‘original’ (the genuine article), but in other cases, the processing flows from step S134 to step S138 and it is judged that the document P is ‘not original’ (a counterfeit).
More specifically, a distance between the calculated characteristic vector and the registered characteristic vector is found, and if this distance is shorter than the predetermined threshold specified in advance, the judgment is ‘original’, and if the distance is longer than the threshold, the judgment is ‘not original’. For the threshold that is used here, the threshold may be specified with a predetermined tolerance range in expectation of errors in the registered characteristic vector and the calculated characteristic vector (errors in reading by the reading section 12, errors in quantization and sampling, etc.). In other words, a size of the threshold may be suitably selected in accordance with the need to carry out authenticity judgments strictly or generously.
Furthermore, because malfunctions such as various operational errors, mispositioning and the like may occur at times of matching, a final judgment may be obtained by judgment results of a number of repetitions, and re-tries may be allowed until it has been judged from comparison results that the printed article is not the original a predetermined number of times.
Meanwhile, in a case in which data representing the registered characteristic vector has been encoded and printed onto the surface (or a rear face) of the document P, it may be possible to read this data from the surface (or rear face) of the document P with the reading section 12 (or a dedicated reading component) and use this data for matching.
Finally, in step S140, a signal representing the judgment result of ‘original’ or ‘not original’ is outputted from the judgment result signal output section 32, and the matching processing ends.
For the matching processing described above, a case in which a registered characteristic vector corresponding with an identification symbol is selected and the calculated characteristic vector and the registered characteristic vector are compared one-to-one has been described. However, if a number of sets of registration data in the memory 28 is small, rather than utilizing an identification symbol, a calculated characteristic vector may be compared with all the registered characteristic vectors.
A procedure for a case of comparing a calculated characteristic vector with all registered characteristic vectors is shown in
As shown in
When the flexing roller 6 has risen and the document P has curved, an image within the observation region S is read by the reading section 12 in step S152, quantization and sampling are performed in step S154, and calculation of the characteristic vector is performed in step S156.
In step S158, the calculated characteristic vector that has been calculated is respectively compared by the comparison section 30 with all the registered characteristic vectors that have been registered in the memory 28. Then, in step S160, it is judged whether or not a highest similarity value, which is the highest of degrees of similarity between the registered characteristic vectors and the calculated characteristic vector, is at or above a predetermined threshold which has been specified in advance. If the highest similarity value equals or exceeds the threshold, the processing advances to step S1162 and it is judged that the matching object document P is an ‘original’ (genuine). On the other hand, if the highest similarity value is less than the threshold, the processing advances to step S164 and it is judged that the document P has ‘no correspondence’ (is a counterfeit).
That is, respective distances between the calculated characteristic vector and all the registered characteristic vectors are found and, basically, the registered printed article whose registered characteristic vector has the shortest distance from the calculated characteristic vector is judged to be the original, but if even the shortest distance is a distance further than the pre-specified threshold, then ‘no correspondence’ with the registered printed articles is judged.
For the threshold that is used here, the threshold may be specified with a predetermined tolerance range applied, similarly to the case of matching processing. Furthermore, because malfunctions such as various operational errors, mispositioning and the like may occur at times of matching, a final judgment may be obtained by judgment results of a number of repetitions, and re-tries may be allowed until it has been judged from comparison results that the printed article is not an original a predetermined number of times.
Then, in step S166, a signal representing the judgment result of ‘original’ or ‘no correspondence’ is outputted from the judgment result signal output section 32, and the identification processing ends.
Thus, in the present exemplary embodiment of the invention, a random pattern of a non-reproducible image of a document P is used for identification matching processing, a characteristic of a random pattern of a non-reproducible image at a previously legitimized document P (original) has been pre-registered, and authenticity (original/not original) of the verification object document P is determined by comparison with a characteristic according to the non-reproducible random pattern of the verification object document P.
As can be seen from
Relationships between the size of the threshold value specified at the comparison section 30 and probabilities of misjudgments of authenticity of documents P are shown in
As is shown in
In contrast, when the document P is flexed to a radius of curvature of, for example, 25 mm, as shown in
Besides papers with smooth surfaces such as coated papers, characteristics of surface conditions of, for example, ordinary papers may be read in the state of being flexed by the flexing roller 6. Similarly to a paper with a smooth surface such as a coated paper, protrusions and indentations of the surface at the apex are physically emphasized, and angles at which light from the light source 10 meets the surface are not uniform.
In a document verification apparatus 102 of a second exemplary embodiment of the present invention, the flexing guide 16, of the document verification apparatus of the mode shown in
A. Thus, the document verification apparatus 102 is an example in which curvature of the curved portion of the document P can be altered.
As shown in
As shown in
When the flexing guide 16 is at a lowered position, the nut portion 9E, the arm member 9A and the flexing guide 16 are at the positions shown by solid lines in
The inclination of the flexing guide 16 is altered by the actuator 9G in order to alter curvature of the document P in the state in which the document P is flexed by the flexing guide 16. As shown by the solid lines in
In the document verification apparatus 102, a subject-flexing mechanism is formed by the flexing guide 16 and the flexing guide-raising and lowering mechanism 9.
Except in the respects described above, structures and operations of the document verification apparatus 102 are similar to the document verification apparatus relating to the first exemplary embodiment of the present invention.
The document verification apparatus 102 features the following characteristic in addition to the characteristics that the document verification apparatus 100 features. Specifically, even with the same subject, there are many different patterns of the surface with different curvatures. Therefore, even at the same portion of the same subject, a plurality of patterns can be obtained by varying the curvature. Hence, if a subject is not judged to be an original unless all of plural patterns match, a misjudgment such that a counterfeit is judged genuine can be prevented even in a case in which one pattern has been counterfeited.
As shown in
Except in the respects described above, the document verification apparatus 104 features structures similar to the document verification apparatus of the first exemplary embodiment of the present invention. In regard to operations, in step S106 of
As shown in
Except in the respects described above, the document verification apparatus 106 features structures similar to the document verification apparatus 102 of the second exemplary embodiment of the present invention, including the flexing guide-raising and lowering mechanism.
In the document verification apparatus 106, when the flexing guide 36 is raised toward the document P as shown in
The document verification apparatus 106 features operations similar to a document verification apparatus of the first exemplary embodiment of the present invention, except in the respects described above.
—Structure—
A document verification apparatus 108 relating to the fifth exemplary embodiment of the present invention is an example of a document verification apparatus which causes the document P to flex in a direction perpendicular to the conveyance direction of the document P. As shown in
—Operation—
Next, operation of the document verification apparatus 108 will be described. As shown in
—Structure—
An example of a document verification apparatus for verifying authenticity of booklet-type documents P, such as passports and the like, will be described below.
As shown in
Concave surfaces 61 are formed at a portion of the platen glass 60 at which the document P is placed. Correspondingly, protruding surfaces 63 with shapes corresponding to the concave surfaces 61 are formed at positions of the platen cover 62 that correspond to the concave surfaces 61. The platen glass 60 and the platen cover 62 constitute a subject-flexing mechanism of the present invention.
—Operation—
Verification of a document P is performed by the document verification apparatus 110 with the following procedure.
First, the document P is placed on the concave surfaces 61 of the platen glass 60 and the platen cover 62 is closed. When the platen cover 62 is closed, the document is pressed against the concave surfaces 61 of the platen glass 60 by the protruding surfaces 63 of the platen cover 62. Thus, the document P is flexed into convex shapes toward the light source 64.
When the platen cover 62 has been closed, the light source 64 lights and an obtained image is read by the image capture element 66. Procedures of reading an original of a document P and of matching with the original are as described for the first exemplary embodiment of the present invention.
Hereabove, for the exemplary embodiments of the present invention, examples in which subject-reading apparatuses of the exemplary embodiments of the present invention are used for verification of documents have been described. However, subject-reading apparatuses of exemplary embodiments of the present invention may be employed for process management, online product inspection and the like in a papermaking plant, a film fabrication plant or the like, or for online product inspection at a planographic printing plate fabrication line. In such a case, the subject-reading apparatus may read a characteristic particular to subjects, and therefore the subjects need not necessarily feature random patterns as described above.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2006-344675 | Dec 2006 | JP | national |
