JPH09330450A - Paper sheet authenticity discriminating device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a paper sheet authenticity discriminating device efficiently and precisely discriminating precisely forged paper money, etc. SOLUTION: After a microcode image pickup part 10 reads a microcode printed in the prescribed fine area of a paper money to obtain an input image and Fourier transformation processing part 11 prepares a Fourier-transformed image obtained by Fourier-transforming the microcode part of this input image, a discrimination processing part 12 compares the number of frequency components with the microcode frequency included in this Fourier transformed image and a threshold value stored in a storing part 13 to discriminate the paper money.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an authenticity discrimination device for optically reading paper sheets such as banknotes and discriminating the authenticity of the paper sheets, and particularly to a microcode area or the like printed on the banknotes. The present invention relates to a genuine / counterfeit discriminating apparatus that efficiently discriminates genuine / counterfeit of a paper sheet by utilizing a minute area.

[0002]

2. Description of the Related Art With the recent advancement of copying technology, counterfeit banknotes for illegally copying various paper sheets such as banknotes and checks have become a big social problem. Therefore, in order to prevent the circulation of such counterfeit banknotes, it is often the case that the authenticity of the paper sheet is determined by applying various image processing techniques to the image image obtained by optically reading the paper sheet.

For example, the European patent EP 069163.
The 2A1 publication discloses an apparatus and method for discriminating a forged ticket created by using color copy, offset lithographic printing, or the like by applying a Fourier transform to an input image.

That is, in this conventional technique, attention is paid to the fact that adjacent dots or lines existing in a real bill do not exist in a counterfeit bill, and such a difference is realized by Fourier transform.

As described above, in the conventional authenticity determination technology represented by this conventional technology, the image processing technology is applied to the input image itself obtained by optically reading the paper sheet to determine whether the paper sheet is authentic. Is determined.

[0006]

However, with the recent development of counterfeiting technology, it has become possible to create counterfeit banknotes with extremely high precision, so that basic image processing technology is simply applied to the entire input image. It is the actual situation that the authenticity of the paper sheet cannot be accurately determined only by applying the.

For example, according to the above-mentioned prior art, the presence or absence of adjacent dots or lines is detected by using the Fourier transform. The high frequency components of the image subjected to this Fourier transform are the lines and boundary lines of the image. The information on the portion where the density changes abruptly is included, and the low frequency component includes the information on the portion about the rough outline of the image.

For this reason, when a counterfeit banknote in which line segments and outlines of the entire banknote are sharpened to some extent appears, it becomes difficult to discriminate even if such a high-frequency component is used, and rather, quantization when optically reading the banknote. The effects of errors and noise will increase.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to solve the above problems and provide a genuine / counterfeit discriminating device for a paper sheet capable of discriminating highly accurate counterfeit bills and the like efficiently and accurately.

[0010]

In order to achieve the above object, a first invention is a sheet for optically reading a sheet and determining the authenticity of the sheet based on the read input image. Class authenticity determination device, a reading unit that optically reads a predetermined minute area of the paper sheet, and a Fourier transform that performs a discrete Fourier transform on the image read by the reading unit to create a discrete Fourier transform image. Based on a predetermined frequency corresponding to the minute region included in the Fourier transform image created by the transforming unit and the Fourier transforming unit,
And a discrimination processing unit for discriminating the authenticity of the paper sheet.

According to a second aspect of the present invention, the discrimination processing means has a predetermined threshold determined in advance based on the number of frequency components in a predetermined frequency range included in each Fourier transform image of a plurality of paper sheets. Storage means for storing a threshold value, counting means for counting the number of frequency components in the predetermined frequency range included in the Fourier transform image corresponding to the input image created by the Fourier transform means, and the counting means It is characterized by further comprising: a discriminating means for discriminating the authenticity of the paper sheet by comparing the counted frequency component with a threshold value stored in the storage means.

A third aspect of the invention is to count the number of frequency components having a predetermined amplitude or more in the predetermined frequency range for each Fourier transform image of a plurality of paper sheets, and calculate the count value and the paper sheets. Further comprising a determining means for determining a predetermined threshold value based on a histogram showing a correspondence relationship with the number of classes,
The storage means stores the predetermined threshold value determined by the determination means.

According to a fourth aspect of the present invention, the discrimination processing means calculates a mean value of amplitudes of the Fourier transform image created by the Fourier transform means at the predetermined frequency, and the calculating means calculates. It is characterized by further comprising a discriminating means for discriminating the authenticity of the paper sheet by comparing the average value of the amplitudes with a predetermined threshold value.

According to a fifth aspect of the present invention, the discrimination processing means includes an average value of amplitudes at a first frequency corresponding to the micro area of the Fourier transform image created by the Fourier transform means, and an area other than the micro area. An average value calculating means for calculating an average value of amplitudes at a second frequency corresponding to the region, and an average value of amplitudes corresponding to the second frequency calculated by the average value calculating means and the first frequency. Amplitude ratio calculating means for calculating the amplitude ratio with the average value of the corresponding amplitudes,
It is characterized by comprising a discriminating means for discriminating the authenticity of the paper sheet by comparing the amplitude ratio calculated by the amplitude ratio calculating means with a predetermined threshold value.

[0015]

Embodiments according to the present invention will be described below with reference to the drawings. In addition, in this embodiment, a case will be described in which the authenticity of a bill is determined based on a microcode printed on a part of the bill.

FIG. 1 is a diagram showing the construction of a bill discriminating apparatus used in the first embodiment.

The banknote discriminating apparatus 1 shown in FIG. 1 reads a microcode printed on a predetermined minute area of a banknote to obtain an input image, and performs a Fourier transform on the microcode portion of the input image to obtain a Fourier transform image. To create. The microcode has a fixed period, and the frequency peculiar to the microcode is obtained by Fourier transform. The authenticity of the paper sheet is determined based on a predetermined frequency (hereinafter referred to as "microcode frequency") corresponding to the microcode included in the Fourier transform image.

Specifically, a predetermined area of a plurality of banknotes is Fourier-transformed in advance, the number of frequency components having a predetermined amplitude at the microcode frequency is counted from each Fourier-transformed image, and the count value and the corresponding value are counted. A histogram showing the correspondence with the number of paper sheets is created. Then, the threshold value for authenticity determination is specified based on the created histogram.

After that, when the Fourier transform image corresponding to the bill to be discriminated is obtained, the number of frequency components having a predetermined amplitude at the microcode frequency contained in this Fourier transform image is counted, and the count value is calculated. The authenticity of the bill is determined by comparing it with a threshold value.

It should be noted that the process of creating the threshold value for authenticity determination and the bill discriminating process based on the threshold value do not necessarily have to be performed as a series of processes. The threshold value is learned by using, and the banknote discriminating device in which the learned threshold value is written in the ROM is used to quickly discriminate the bill.

As shown in FIG. 1, this bill validator 1
Is a microcode image pickup unit 10 that picks up a microcode printed on a predetermined position of a bill, a Fourier transform processing unit 11 that cuts out a microcode region of an input image and performs preprocessing and Fourier transform, and a genuine / counterfeit of a bill. And a storage unit 13 that stores a threshold value.

The microcode image pickup unit 10 turns on the LED at the time when the conveyed bill is detected by a predetermined pass sensor, and the printed portion of the microcode of the bill is a two-dimensional CCD.
Read with a sensor. Then, the image read by the two-dimensional CCD sensor is output to the Fourier transform processing unit 11. A detailed description of the microcode image pickup unit 10 will be given later.

The Fourier transform processing unit 11 performs shading correction, thinning processing, and microcode region cutting on the input image received from the microcode imaging unit 10, and then performs two-dimensional discrete Fourier transform on the cut region. Transform (two-dimentional discrete Fourier trans
formation processing), and the Fourier transform image obtained by these series of processing is output to the discrimination processing unit 12.

Here, shading correction means CCD
It is intended to correct the variation of light amount due to the characteristics of the pixels and the variation of each pixel of the sensor, and the blank data is Wij,
If the black paper data when the LED is turned off is Bij and the acquired data is Dij, the corrected data is Rij, which is obtained from the calculation formula of Rij = (Dij-Bij) / (Wij-Bij). .

Further, the thinning process is a process for making the line thickness of the microcode uniform in order to facilitate the comparison by the Fourier transform, and various thinning techniques can be applied. .

For example, when thinning processing is performed using a mask operator having a 2 × 2 first-order differential, the coefficients at the upper left portion, upper right portion, lower left portion, and lower right portion of this mask operator are c1 and c2, respectively. , C3 and c4, and the corresponding pixel data are d1, d2, d3 and d4 respectively, Is calculated, and this Z value is used as the pixel value of the upper left pixel.
Note that SQRT (X) indicates the square root of X.

Further, the cutting process of the microcode area is performed as follows.
This is for removing an unnecessary portion from the microcode portion of the bill.

For example, in the case of a 10,000-yen bill in Japan, microcodes are printed on the upper left and right upper portions of the bill and the central portion of the rear face, so the microcode image pickup unit 10 picks up images of these portions, The Fourier transform processing unit 11 cuts out the microcode area. Further, in the case of the 1,000-yen bill, since the microcode is printed on the lower right part of the portrait and the upper right part and the lower left part of the back surface, the microcode is cut out from this area.

The two-dimensional discrete Fourier transform is an extension of the one-dimensional discrete Fourier transform in which the filter operation in the spectral domain is performed in the Fourier spectral domain.
By applying to the image data which is the dimensional information, the frequency component in the horizontal direction (x direction) and the vertical direction (y direction) and the amplitude component at the frequency are calculated.

Then, the two-dimensional discrete Fourier transform is
Similar to the one-dimensional case, the low-frequency component has the characteristics of a rough image, and the high-frequency component has the characteristic that it contains information on the abrupt density change portion. In addition, the fast Fourier transform (FFT: Fa
st Fourier Transformation) is available.

In this embodiment, the two-dimensional discrete Fourier transform is realized by first performing the Fourier transform in the horizontal direction and then performing the Fourier transform in the vertical direction.

The discrimination processing unit 12 counts the number of frequency components having a predetermined amplitude in the microcode frequency included in the acquired Fourier transform image, compares the counted number of pixels with a threshold value, and verifies the authenticity of the bill. To determine.

The threshold is calculated in advance from a plurality of banknotes before the banknote to be discriminated is input.

Specifically, each of these banknotes is Fourier-transformed, the number of frequency components having a predetermined amplitude at the microcode frequency is counted from each Fourier-transformed image to create a histogram, and based on the created histogram Identify the threshold.

The storage unit 13 is a storage unit for storing the threshold value specified based on the histogram, and the stored threshold value is accessed by the discrimination unit 12c described later.

Next, a specific configuration of the discrimination processing section 12 will be described.

As shown in FIG. 1, this discrimination processing unit 13
Is composed of a histogram creating unit 12a, a counting unit 12b, and a discriminating unit 12c.

The histogram creating section 12a is a processing section for creating a histogram from Fourier transform images corresponding to a plurality of banknotes in order to specify a threshold value for authenticity determination in advance.

Specifically, the number of frequency components having a predetermined amplitude at the microcode frequency is counted from each Fourier transform image, and a histogram showing the correspondence relationship between the counted value and the number of paper sheets corresponding thereto. To create.

Therefore, the histogram creating section 12a
The histogram created by can be represented by a graph in which the horizontal axis represents the number of frequency components having a predetermined amplitude at the microcode frequency and the vertical axis represents the Fourier transform image having this count value, that is, the number of banknotes.

The histogram creating section 12a
Identifies a threshold for discrimination based on the created histogram, and outputs the identified threshold to the discriminator 12c. The determination unit 12c that receives this threshold value
The threshold is stored in the storage unit 13.

The counting section 12b is a processing section for counting the number of frequency components having a predetermined amplitude at the microcode frequency from the Fourier transform image corresponding to the bill to be discriminated, and outputs the counting result to the discriminating section.

The discriminating unit 12c stores the count value received from the counting unit 12b, that is, the number of frequency components having a predetermined amplitude at the microcode frequency included in the Fourier transform image corresponding to the bill to be discriminated, and the storage unit 13. It is a processing unit that determines the authenticity of the bill by comparing it with the threshold value stored in.

Specifically, when the count value is equal to or greater than the threshold value, the bill to be determined is determined to be genuine, and when the count value is less than the threshold value, The bill is determined to be counterfeit.

The reason why the threshold value is specified by using the histogram in this way is that in the case of a genuine banknote, the microcode printed on the banknote affects the microcode in the Fourier transform image. In the case of counterfeit banknotes, a frequency component with a certain amplitude appears in the frequency.
This is because the microcode on the counterfeit banknote is crushed and a frequency component having a predetermined amplitude at the microcode frequency does not appear in the Fourier transform image.

That is, when observing this histogram, since genuine banknotes are mainly distributed in the high band part and counterfeit banknotes are mainly distributed in the low band part, the threshold value is set by utilizing such distribution characteristics. .

By using the bill discriminating apparatus 1 having the above-mentioned configuration, it is possible to discriminate the bill from the distribution based on the frequency component having a predetermined amplitude at the microcode frequency contained in the Fourier transform image. .

Next, the microcode image pickup section 10 will be specifically described.

FIG. 2 is a diagram showing an external configuration of the microcode image pickup section 10 shown in FIG.

As shown in FIG. 2, the microcode image pickup unit 10 uses a bill insertion slot 10a for inserting a bill to be discriminated and a two-dimensional CCD sensor 10c for the microcode area of the inserted bill. It is composed of an image pickup unit main body 10b that is optically read by an image input board 10d that uses the read data as image data.

Then, when the bill 20 as the authenticity discrimination target is inserted into the bill insertion slot 10a, the bill is conveyed to the image pickup unit main body 10b, and its microcode portion is two-dimensional C.
It is read by the CD sensor 10c. Then, the read data is output to the image input board 10d via the cable, and the image input board 10d makes the two-dimensional image data.

FIG. 3 is a diagram showing the transportation and imaging timing of the imaging section body 10b.

As shown in FIG. 3A, three pairs of path sensors p1, p2, and p3 are arranged on the banknote conveyance path, and a two-dimensional CC is provided between the path sensors p2 and p3.
The D sensor 10c is arranged. A rotary encoder 30 is provided in the vicinity of each pass sensor to measure the time when the pass sensor detects the bill, that is, the distance from the tip of the bill to the position of a predetermined microcode.

When the pass sensor p1 detects a bill being conveyed, the motor is turned on, and when the pass sensor p2 detects a bill being conveyed, clock counting is started, and after a predetermined count, the LED is turned on and the image is picked up. Run. If the pass sensor p3 detects a bill being conveyed, the motor is turned off.

Specifically, as shown in FIG. 3 (b), when the bill being conveyed passes the pass sensor p2, the pass sensor p2 detects light blocking by the bill and starts counting the mechanical clock. When the count value of the mechanical clock reaches a predetermined value, the two-dimensional CCD sensor 10 is turned on within about 10 microseconds after the LED is turned on.
The imaging by c is performed. The count value at which the LED starts to be turned on is preset based on the position on the banknote where the microcode is printed.

FIG. 4 is a diagram showing a procedure for carrying and imaging the imaging section main body 10b.

As shown in FIG. 4, when a bill to be conveyed shields the pass sensor p1 from light (step 401), the motor is turned on (step 402), and the bill further shields from the pass sensor p2 (step 403). Counting is started (step 404).

If the mechanical clock continues to be counted up to a preset value (step 40)
5-406), after turning on the LED (step 40
7) Image the microcode portion printed on the bill (step 408).

Then, the banknote to be conveyed is the pass sensor p3.
If is shielded from light (step 409), the motor is turned off to end the image pickup process (step 410).

FIG. 5 is a diagram showing an example of the positions of microcodes captured by the two-dimensional CCD sensor.

As shown in FIG. 5A, in the case of US dollar bills, microcodes are printed on the upper left portion and the upper right portion of the portrait for each denomination.

Specifically, the microcode printed on the upper left part of the portrait is 96 mm from the right end of the banknote and 33 mm from the upper end of the banknote, and the microcode printed on the upper right part of the portrait is the banknote. 60 mm from the right end of the banknote and 33 mm from the top of the bill.

As shown in FIG. 5B, in the case of Deutsche Mark, a microcode is printed on the right side of the portrait, and the printing position of the microcode is slightly different for each denomination.

Specifically, for 10 mark, 20 mark and 100 mark, a microcode is printed 22 mm from the right end of the bill and 33 mm from the lower end of the bill, and for 50 mark, 22 m from the right end of the bill.
At m, a microcode is printed at a position 62 mm from the lower end of the bill. Further, in the case of 200 mark, the microcode is printed at a position 25 mm from the right end of the bill and 33 mm from the lower end of the bill.

As shown in FIG. 5 (c), in a Japanese 10,000-yen bill, microcodes are printed on the upper right and upper left portions of the bill and the lower central portion of the back face.

Specifically, on the front side, microcodes are printed 14 mm from the upper end of the banknote and 10 mm from the left end of the banknote, and 14 mm from the upper end of the banknote and 10 mm from the right end of the banknote, and on the back side, A microcode is printed at a position 5 mm from the lower end of the bill and 80 mm from the right end of the bill.

As shown in FIG. 5 (d), in the Japanese 1,000-yen bill, microcodes are printed on the lower right portion of the front surface of the bill, and on the upper right portion and lower left portion of the back surface.

Specifically, a microcode is printed on the front surface at a position 5 mm from the lower end of the banknote and at a position 25 mm from the right end of the banknote, and on the back surface at a position 45 mm from the lower end of the banknote and 45 mm from the left end of the banknote. Then, the microcode is printed at a position 14 mm from the upper end of the bill and 79 mm from the right end of the bill.

As described above, since the position of the microcode printed on the bill differs depending on the type of bill, the microcode image pickup unit 1 can be used depending on the bill to be judged as authenticity.
It is necessary to set the mounting position of 0 and the count value of the mechanical clock in advance.

Next, the hardware configuration of the bill discriminating apparatus 1 will be described.

FIG. 6 is a diagram showing a hardware configuration of the bill discriminating apparatus 1 shown in FIG.

As shown in FIG. 6, this bill discriminating apparatus 1
Includes a CPU 60, a memory 61, a display unit 62, a drive circuit 64b, a CCD control circuit 65b, waveform shaping circuits 66b and 67b, and a drive unit 68, respectively.
3 is connected.

First, the CPU 60 controls the entire circuit,
A central processing unit that performs Fourier transform, histogram creation processing, discrimination processing, and the like, and specifically corresponds to the Fourier transform processing unit 11 and the discrimination processing unit 12 illustrated in FIG. 1.

The respective parts located on the left side of the bus 63 are
It is a component of the microcode imaging unit 10 shown in FIG.

That is, the drive circuit 64b is a circuit for driving the LED 64a after a bill passes through a predetermined pass sensor, and the CCD control circuit 65b is based on the timing pulse output from the timing control circuit 65f.
This is a circuit that controls the sensor 65a. Specifically, the data converted by the A / D converter 65c is sequentially stored in the memory 65d under the control of the control circuit 65e.

The waveform shaping circuits 66b and 67b are
Path sensor 66a and rotary encoder 6 respectively
7a is a circuit that shapes the waveform, and the drive unit 68 drives the banknote transport system and the like.

Next, a discrimination processing procedure and a threshold generation processing procedure performed by the bill discriminating apparatus 1 shown in FIG. 1 will be described.

FIG. 7 is a flowchart showing the discrimination processing procedure of the bill discriminating apparatus 1 shown in FIG.

As shown in FIG. 7, this bill transporting device 1
When an input image is obtained by imaging by the microcode imaging unit 10 (step 701), the Fourier transform processing unit 11 performs shading correction, thinning processing, and microcode region cutout as preprocessing (step 702). This pretreatment can be omitted.

Then, the Fourier transform processing unit 11 performs a two-dimensional fast Fourier transform (FFT) on the cut out region.
To create a Fourier transform image (step 70
3) At the microcode frequency, a genuine bill has a large amplitude and a counterfeit bill has a small amplitude.
a counts the number of frequency components having a predetermined amplitude at the microcode frequency included in this Fourier transform image (step 704).

Then, the discriminating unit 12c compares the count result with the threshold value stored in the storage unit 13,
The authenticity of the bill is determined (step 705), the determination result is output (step 706), and the process ends.

The threshold value stored in the storage unit 13 is previously learned according to a threshold value creating procedure described later.

Next, the procedure of the threshold value creating process performed by the bill discriminating apparatus 1 shown in FIG. 1 will be described.

FIG. 8 is a flow chart showing the procedure of threshold value generation processing of the bill discriminating apparatus 1 shown in FIG.

In the case of creating this threshold value, a plurality of banknotes of the same type are prepared in advance, and the banknotes of the banknotes are picked up by the microcode image pickup section 10 (step 801).

When the input image is obtained by this image pickup, the Fourier transform processing unit 11 performs shading correction, thinning processing, and microcode region cutting as preprocessing (step 802). This pretreatment can be omitted.

Then, the Fourier transform processing unit 11 performs a two-dimensional fast Fourier transform (FFT) on the cut out region.
To create a Fourier transform image (step 80
3), the histogram creation unit 12a counts the number of frequency components having a predetermined amplitude at the microcode frequency included in the Fourier transform image (step 804).

Then, a histogram is created by incrementing the location corresponding to the number of frequency components of this Fourier transform image (step 805), and if there is another banknote, the process proceeds to step 801.

Then, if the histogram creation processing is completed for all the sample banknotes (step 80).
6) Then, a threshold for discrimination is specified based on the distribution of this histogram (step 807).

Next, the Fourier transform processing unit 11 shown in FIG.
The reason for using is explained.

FIG. 9A is a diagram showing an example of the microcode, in which the microcode is "MICROM".
It shows a case of 10 characters "ICRO".

Here, the microcode printed on the bill has a width of about 100 μm to 250 μm, and C
Since the cell size of the CD sensor is about 7 × 10 μm, one character of the microcode in the case of equal size is 1
It is represented by 0 to 25 pixels.

FIG. 9B shows a microcode waveform at the time when the l-th line of the microcode shown in FIG. 9A is read. However, the scanning direction of the CCD sensor is t, and the density at that time is x (t).

As described above, the microcode has a characteristic of showing a waveform having a constant period T. However, this period T is determined depending on the magnification of the lens and the CCD cell size.

FIG. 9C is a diagram showing a power spectrum when the microcode shown in FIG. 9B is Fourier-transformed. However, ω shown in the figure represents a microcode frequency, and has a relationship of ωn = n × 2π / T.

Specifically, when n = 1 and T = 10, ω1 = 2π / 10 = 0.628. For example, when the magnification is 1 ×, that is, n = 10, ω10 = 62.8. Therefore, the frequency component at ω 10 = 62.8 is more remarkable than the frequency components at other frequencies.

Thus, the clearer the microcode can be read, the sharper the peak can be obtained at the microcode frequency ω n.

That is, by referring to the power spectrum of the microcode, it is possible to discriminate between a counterfeit banknote in which the microcode portion is crushed and a genuine banknote in which the microcode can be clearly read. In the form, the Fourier transform is premised on the processing.

Since the microcode is printed on the banknote in various directions such as the vertical direction, the horizontal direction, and the oblique direction, in the present embodiment, the two-dimensional Fourier transform is used to make it dependent on the direction. It is possible to discriminate based on the repetition cycle without performing.

Next, the histogram creation unit 12 shown in FIG.
The process of creating a will be described.

FIG. 10A is a diagram showing the frequency domain when the two-dimensional Fourier transform is performed. Note that here, the microcode frequency from the center 0 of the Fourier transform image is ω, and the amplitude at that time is Aωi (i = 1 to m). Note that m represents the number of data at the microcode frequency ω.

As shown in the figure, this two-dimensional Fourier transform image has a frequency region consisting of a two-dimensional space in the horizontal direction and the vertical direction. The higher the frequency, the higher the frequency.

In this embodiment, the two-dimensional frequency region is divided into a low frequency region below a predetermined reference frequency and a high frequency region above the reference frequency.

Since the low frequency region contains a lot of noise components and most of the microcode components in the high frequency region, only the high frequency region is focused and the low frequency region is excluded. .

Then, for example, when the micro frequency corresponds to the points 90 and 91, each frequency component of the Fourier transform image counts the number of frequency components having the points 90 and 91.

However, when the amplitudes at the points 90 and 91 are below the predetermined level, it is unlikely that the pixel is a microcode portion. Therefore, when the amplitude is below the predetermined amplitude, it is not counted. .

FIG. 10B is a diagram showing a histogram in which the horizontal axis represents the number of frequency components having a predetermined amplitude at the microcode frequency and the vertical axis represents the Fourier transform image having this number of frequency components. Is.

For example, when the number of frequency components that satisfy the conditions at points 90 and 91 in FIG. 10A for a certain Fourier transform image is n, the n portion of the horizontal axis is incremented by one in the vertical direction. To do.

When this processing is performed on each Fourier transform image, the histogram shown in FIG. 10B is obtained.

Here, in the case of a Fourier transform image corresponding to a real bill, there are many frequency components having a predetermined amplitude at the microcode frequency, and in the case of a Fourier transform image corresponding to a counterfeit bill, There are not many frequency components having a predetermined amplitude at the microcode frequency.

Therefore, when such a histogram is created, the distribution in the case where the microcode exists and the distribution in the case where the microcode does not exist are divided into two, and the boundary portion can be selected as the threshold value.

Next, a histogram when the present embodiment is actually applied to German mark notes will be described. However, the counterfeit banknotes used as samples are from high-precision copying machines.

FIG. 11 is a diagram showing an example of a histogram when the present embodiment is applied to German mark notes.

FIG. 11 (a) is a histogram created for each Fourier transform image obtained by reading the microcode portion of a true bill and a copy bill of 50 mark or 200 mark and subjecting the input image to Fourier transform. The microcode frequency corresponding to the microcode is 64 Hz here.

As shown in the figure, specifically, an amplitude of 40 at a microcode frequency of 63 to 70 Hz including an error.
The frequency components satisfying the above conditions are distributed in the area of 60 or less data for the copy ticket and in the area of 60 or more data for the genuine ticket.

Therefore, the threshold value in such a case is 60, and when the number of frequency components having the microcode frequency of 64 Hz is 60 or more in the Fourier transform image corresponding to the bill to be discriminated, The bills can be regarded as genuine bills, and if there are less than 60 bills, they can be regarded as copy bills.

FIG. 11 (b) is a histogram created for each Fourier transform image obtained by reading the microcode portion of a genuine note and a copy note of 10 mark, 20 mark or 100 mark and subjecting the input image to Fourier transform. .

As shown in the figure, regarding the copy ticket,
It is distributed in the area where the number of data is 90 or less.
It is distributed in the above area.

Therefore, the threshold value in such a case is set to 90, and if the number of frequency components having the microcode frequency of 64 Hz in the Fourier transform image corresponding to the bill to be discriminated is 90 or more, the bill is billed. Considered genuine, 9
If it is less than 0, it can be regarded as a copy ticket.

As described above, in the first embodiment, the microcode image pickup section 10 reads the microcode printed on the predetermined minute area of the bill to obtain the input image, and the Fourier transform processing section 11 After creating a Fourier transform image obtained by performing a Fourier transform on the microcode portion of this input image, the discrimination processing unit 12 stores the number of frequency components having a predetermined amplitude at the microcode frequency included in this Fourier transform image and the storage unit 13. Since the bill is discriminated by comparing it with the threshold value stored in, it is possible to discriminate highly accurate counterfeit bills efficiently and accurately.

Further, the histogram creating section 12a counts the number of frequency components having a predetermined amplitude at the microcode frequency from each Fourier transform image corresponding to a plurality of banknotes, and counts the number of counted pixels and each Fourier transform image. Since the threshold for authenticity determination is specified based on the histogram showing the correspondence with the distribution, the threshold for bill identification can be easily set.

The first embodiment has been described above.

By the way, in the first embodiment, the authenticity of a bill is determined by using the frequency component of the Fourier transform image. However, since the Fourier transform image also includes the amplitude component, this amplitude component Can also be used for authenticating bills.

Therefore, a second embodiment for judging the authenticity of the bill using the amplitude component of the Fourier transform image will be described next.

FIG. 12 is a diagram showing the structure of the bill validator 2 used in the second embodiment.

The bill validator 2 shown in FIG. 12 obtains the average value of the amplitudes of the microcode frequencies included in the Fourier transform image (hereinafter referred to as "average amplitude"), and sets this average amplitude as a predetermined threshold value. This is a device for comparing and determining the authenticity of bills.

As shown in FIG. 12, this bill validator 2
Includes a microcode imaging unit 10, a Fourier transform processing unit 11, a discrimination processing unit 120, and a storage unit 13.
Note that the microcode imaging unit 10, the Fourier transform processing unit 11, and the storage unit 13 have the same configurations as those shown in FIG. 1, and thus detailed description thereof will be omitted.

The discrimination processing unit 120 averages the amplitude components of the microcode frequencies included in the Fourier transform image,
The average amplitude calculating unit 1 is a processing unit that compares the average amplitude with a preset threshold value to determine the authenticity of a bill.
20a and a discriminating unit 120b.

The average amplitude calculator 120a sequentially checks the amplitude components of the microcode frequencies contained in the Fourier transform image, and calculates the average amplitude by dividing this amplitude component by the number of data. The reason for performing such averaging is
This is to reduce the influence of image tilt.

Specifically, if there are m pieces of data having the microcode frequency ωi of the Fourier transform image and their amplitudes are Aωij (j = 1, 2, ..., M), the average amplitude Avr (Aωi ) Is It is calculated by the formula.

However, this average amplitude Avr (Aωi)
Has the characteristic that it changes depending on the color of the bill.

For example, as shown in FIG.
When the density of the input image is lower than that of 3 (b), the microcode information is missing in the thin portion.
The amplitude value of the Fourier transform is affected by the amount of chipping.

Therefore, in consideration of the color of the bill, the storage unit 1
It is necessary to specify the threshold value to be stored in 3.

The discriminator 120b includes an average amplitude calculator 120.
It is a processing unit that compares the average amplitude calculated by a with a threshold value stored in the storage unit 13 to determine the authenticity of the bill.

By using the bill discriminating apparatus 2 having the above structure, it is possible to discriminate the authenticity of bills by utilizing not only the microcode frequency of the Fourier transform image but also the average amplitude at the microcode frequency.

Next, the processing procedure of the bill discriminating apparatus 2 shown in FIG. 12 will be described. However, regarding the threshold,
It is assumed to be set in advance.

FIG. 14 is a flow chart showing the processing procedure of the bill discriminating apparatus 2 shown in FIG.

As shown in FIG. 14, this bill transporting device 2
When the input image is obtained by the microcode imaging unit 10 (step 1401), the Fourier transform processing unit 11
Performs shading correction, thinning processing, and microcode area cutout as preprocessing (step 140).
2). This pretreatment can be omitted.

Then, the Fourier transform processing unit 11 performs a two-dimensional fast Fourier transform (FFT) on the cut out region.
To generate a Fourier transform image (step 140
3), the average amplitude calculator 120a extracts the amplitude component of the microcode frequency included in this Fourier transform image (step 1404).

After averaging the amplitude components to calculate the average amplitude (step 1405), the discriminating unit 120
b compares the average amplitude with the threshold value in the storage unit 13 to determine the authenticity of the bill (step 1406), outputs the determination result (step 1407), and ends the process.

As described above, in the second embodiment, the microcode image pickup section 10 reads the microcode printed on a predetermined minute area of the bill to obtain the input image, and the Fourier transform processing section 11 After the Fourier transform image obtained by performing the Fourier transform on the microcode portion of the input image is created, the discrimination processing unit 120 determines the average amplitude of the microcode frequencies included in the Fourier transform image and the threshold value stored in the storage unit 13. Since the bills are discriminated by comparing the above, it is possible to discriminate highly accurate counterfeit bills efficiently and accurately.

The second embodiment has been described above.

Next, a third embodiment for discriminating the authenticity of a bill by using the ratio between the amplitude component of the microcode frequency and the amplitude component of a predetermined frequency other than the microcode frequency will be described.

FIG. 15 is a diagram showing the construction of the bill discriminating apparatus 3 used in the third embodiment.

The bill discriminating apparatus 3 shown in FIG. 15 obtains the ratio between the average amplitude of the microcode frequencies included in the Fourier transform image and the average amplitude of the predetermined frequencies other than the microcode frequency, and determines this amplitude ratio. It is a device that determines the authenticity of a bill by comparing it with a threshold value.

As shown in FIG. 15, this bill validator 3
Includes a microcode image pickup unit 10, a Fourier transform processing unit 11, a discrimination processing unit 150, and a storage unit 13.
Note that the microcode imaging unit 10, the Fourier transform processing unit 11, and the storage unit 13 have the same configurations as those in FIG. 1, and thus detailed description thereof will be omitted.

The discrimination processing section 150 obtains the amplitude ratio between the average amplitude of the microcode frequencies included in the Fourier transform image and the average amplitude other than the microcode frequency, and compares this amplitude ratio with a preset threshold value. Is a processing unit that determines the authenticity of a bill, and the average amplitude calculation unit 150a
And an amplitude ratio calculator 150b and a discriminator 150c.

The average amplitude calculating section 150a is a processing section for calculating the average amplitude of the microcode frequencies included in the Fourier transform image and the average amplitude of predetermined frequencies other than the microcode frequency.

The amplitude ratio calculation unit 150b is a processing unit that divides the average amplitude of predetermined frequencies other than the microcode frequency by the average amplitude of the microcode frequencies included in the Fourier transform image to calculate the amplitude ratio.

Specifically, the average amplitude of the microcode frequency ω1 of the Fourier transform image is Avr (Aω1),
The average amplitude of frequencies ω2 other than the microcode frequency is A
Assuming that vr (Aω2), the amplitude ratio r is calculated by the calculation formula of r = Avr (Aω2) / Avr (Aω1).

The discriminator 150c includes an amplitude ratio calculator 150b.
Is a processing unit that compares the calculated amplitude ratio with the threshold value stored in the storage unit 13 to determine the authenticity of the bill.

By using the bill discriminating apparatus 3 having the above structure, it is possible to discriminate the authenticity of bills by utilizing not only the microcode frequency of the Fourier transform image but also the amplitude ratio at the microcode frequency.

The reason for adopting such an amplitude ratio in the present embodiment is that the amplitude of the microcode frequency and the amplitude of a predetermined frequency other than the microcode frequency are different in magnitude. When the histogram of the amplitude ratio is created, the distributions of the amplitude ratios of the genuine note and the copy note are separated.

FIG. 16 is a diagram showing a histogram when this embodiment is applied to US dollar bills. Here, the microcode frequency is 68 Hz, and the frequencies other than the microcode are 10 Hz. Amplitude is ω1 = 6
8 and ω2 = 10, they are A68 and A10, respectively.
The amplitude ratio is r = Avr (A10) / Avr (A68).
As shown in FIG. 16, referring to this distribution of the amplitude ratio, the amplitude ratio of the copy ticket becomes large and the amplitude ratio of the genuine paper becomes small.

From this, by setting a threshold value between both distributions and storing the threshold value in the storage unit 13, it becomes possible to determine the authenticity of the bill.

Next, the processing procedure of the bill discriminating apparatus 3 shown in FIG. 15 will be described. However, the threshold for discrimination is
It is assumed to be set in advance.

FIG. 17 is a flow chart showing the processing procedure of the bill discriminating apparatus 3 shown in FIG.

As shown in FIG. 17, this bill transport device 3
When an input image is obtained by imaging by the microcode imaging unit 10 (step 1701), the Fourier transform processing unit 11 performs shading correction, thinning processing, and microcode region cutout as preprocessing (step 1702).

Then, the Fourier transform processing unit 11 performs a two-dimensional fast Fourier transform (FFT) on the cut out region.
To generate a Fourier transform image (step 170
3), the average amplitude calculator 150a calculates the average amplitude of the microcode frequencies included in this Fourier transform image and the average amplitude of the frequencies other than the microcode frequency (step 1704).

Then, the amplitude ratio calculation unit 150b calculates the amplitude ratio of both average amplitudes (step 1705), and the discrimination unit 15
0c compares the amplitude ratio with a threshold value stored in the storage unit 13 to determine the authenticity of the bill (step 17).
06), and outputs the determination result (step 170).
7), the process ends.

As described above, in the third embodiment, the microcode image pickup section 10 reads the microcode printed on the predetermined minute area of the bill to obtain the input image, and the Fourier transform processing section 11 After creating a Fourier transform image obtained by performing a Fourier transform on the microcode portion of the input image, the discrimination processing unit 150 determines the amplitude ratio between the microcode frequency included in the Fourier transform image and a predetermined frequency other than the microcode frequency. Since the bills are discriminated by comparing the obtained amplitude ratio with the threshold value stored in the storage unit 13, it is possible to discriminate highly accurate counterfeit bills efficiently and accurately.

The third embodiment has been described above.

Here, the first and second embodiments are used to determine the authenticity of a bill using the frequency component of the Fourier transform image, and the second embodiment is used to determine the authenticity of the bill using the amplitude component of the Fourier transform image. Although the third embodiment is shown separately,
The present invention is not limited to this, and may be configured by combining the respective embodiments.

Further, in the above-mentioned first to third embodiments,
The case of discriminating the authenticity of the bill based on the microcode printed on the bill has been explained, but it is applied when discriminating the authenticity of the bill based on minute areas such as a mesh pattern part attached to various paper sheets. It is also possible to do so.

[0165]

As described in detail above, according to the first aspect of the present invention, a predetermined minute area of a paper sheet is optically read, and a discrete Fourier transform is performed on the read image to obtain a discrete Fourier transform image. Since the authenticity of the paper sheet is determined based on the predetermined frequency corresponding to the minute area included in the generated Fourier transform image, the following effects can be obtained.

1) It is possible to discriminate highly accurate counterfeit banknotes efficiently and accurately.

2) It is possible to discriminate the authenticity of bills at high speed.

Further, in the second invention, a predetermined threshold value determined in advance based on the number of frequency components in a predetermined frequency range included in each Fourier transform image of a plurality of sheets is stored in the storage means. If a Fourier transform image corresponding to the input image is obtained and stored, the number of frequency components in a predetermined frequency range included in this Fourier transform image is counted, and the counted frequency component is stored in the storage means. Since the authenticity of the paper sheet is determined by comparing with the threshold value, the following effects can be obtained.

1) It is possible to make a genuine / counterfeit judgment of a bill based on only the frequency distribution corresponding to the minute area without considering the amplitude component which is easily influenced by the color of the bill.

2) It becomes possible to easily specify the threshold value required for authenticity discrimination.

Further, in the third invention, the number of frequency components having a predetermined amplitude or more in a predetermined frequency range is counted for each Fourier transform image of a plurality of paper sheets, and the count value and the paper sheets are counted. Since the determination means for determining the predetermined threshold value is provided based on the histogram showing the correspondence with the number of sheets, it is possible to predetermine an appropriate threshold value for use in authenticity determination and store it in the storage means. It will be possible.

Further, in the fourth invention, the average value of the amplitudes of the Fourier transform image at a predetermined frequency is calculated, and the calculated average value of the amplitudes is compared with a predetermined threshold value to determine the authenticity of the paper sheet. Since it is configured to discriminate, it is possible to make an accurate authenticity determination based on both the frequency corresponding to the minute area and its amplitude component.

Further, in the fifth invention, the average value of the amplitudes at the first frequency corresponding to the minute area of the Fourier transform image and the average value of the amplitudes at the second frequency corresponding to the areas other than the minute area are set. And the calculated second
The amplitude ratio between the average value of the amplitudes corresponding to the frequency and the average value of the amplitudes corresponding to the first frequency is calculated, and the calculated amplitude ratio is compared with a predetermined threshold value to determine the authenticity of the paper sheet. Since the determination is made, it is possible to make an accurate authenticity determination based on both the frequency corresponding to the minute area and the frequency corresponding to the area other than the minute area.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a bill discriminating apparatus used in the first embodiment.

FIG. 2 is a diagram showing an external configuration of a microcode image pickup unit shown in FIG.

FIG. 3 is a diagram showing the transportation and imaging timing of the imaging unit main body shown in FIG.

FIG. 4 is a diagram showing a procedure for carrying and imaging the imaging unit main body shown in FIG.

FIG. 5 is a diagram showing an example of positions of microcodes captured by a two-dimensional CCD sensor.

FIG. 6 is a diagram showing a hardware configuration of the bill discriminating apparatus shown in FIG. 1.

FIG. 7 is a flowchart showing a discrimination processing procedure of the bill discriminating apparatus shown in FIG. 1.

FIG. 8 is a flowchart showing a threshold value generation processing procedure of the bill discriminating apparatus shown in FIG. 1.

9 is an explanatory diagram of a Fourier transform unit shown in FIG.

FIG. 10 is an explanatory diagram of a histogram creation unit shown in FIG.

FIG. 11 is a diagram showing an example of a histogram when the first embodiment is applied to German mark bills.

FIG. 12 is a diagram showing a configuration of a bill validator used in the second embodiment.

FIG. 13 is a diagram showing an example of shading of an input image due to a difference in color of a bill.

14 is a flowchart showing a processing procedure of the bill discriminating apparatus shown in FIG.

FIG. 15 is a diagram showing a configuration of a bill discriminating apparatus used in the third embodiment.

FIG. 16 is a diagram showing a histogram when the third embodiment is applied to US dollar bills.

17 is a flowchart showing a processing procedure of the bill discriminating apparatus shown in FIG.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Banknote discrimination device, 10 ... Microcode imaging part, 1
1 ... Fourier transform processing unit, 12 ... Discrimination processing unit, 12a
… Histogram creation unit, 12b… Counting unit, 12c…
Discrimination unit, 13 ... Storage unit, 10a ... Banknote insertion slot, 10
b ... Imaging unit main body, 10c ... Two-dimensional CCD sensor, 10
d ... Image input board, 30 ... Rotary encoder, 6
0 ... CPU, 61 ... Memory, 62 ... Display unit, 63
... bus, 64a ... LED, 64b ... driving circuit, 65
a ... CCD sensor, 65b ... CCD control circuit, 65c
... A / D converter, 65d ... Memory, 65e ... Control circuit, 65f ... Timing control circuit, 66a ... Path sensor, 66b, 67b ... Waveform shaping circuit, 67a ... Rotary encoder, 68 ... Drive unit, 2 ... Banknote identifying device, 120 ... discrimination processing unit, 120a ... average amplitude calculation unit, 120b ... discrimination unit, 3 ... bill recognition device, 150
... discrimination processing unit, 150a ... average amplitude calculation unit, 150b
... Amplitude ratio calculation unit, 150c ... Discrimination unit

Claims (5)

[Claims] 1. A paper sheet authenticity determination device for optically reading a paper sheet and determining the authenticity of the paper sheet based on the read input image, wherein a predetermined minute region of the paper sheet is provided. Included in the Fourier transform image created by the Fourier transform unit; a reading unit that optically reads An authenticity determination device for a paper sheet, comprising: a determination processing unit for determining the authenticity of the paper sheet based on a predetermined frequency corresponding to the minute area. 2. The memory for storing the predetermined threshold value determined in advance based on the number of frequency components in a predetermined frequency range included in each Fourier transform image of a plurality of sheets of paper. Means, counting means for counting the number of frequency components in the predetermined frequency range included in the Fourier transform image corresponding to the input image created by the Fourier transform means, and the frequency component counted by the counting means 2. An authenticity determining device for a paper sheet according to claim 1, further comprising a determining means for comparing the threshold value stored in the storage means with the authenticity of the paper sheet. 3. The number of frequency components having a predetermined amplitude or more in the predetermined frequency range is counted for each Fourier transform image of a plurality of paper sheets, and the count value corresponds to the number of paper sheets. The determination means for determining a predetermined threshold value based on a histogram showing the relationship is further provided, and the storage means stores the predetermined threshold value determined by the determination means. Authenticity determination device for paper sheets. 4. The discrimination processing means calculates a mean value of amplitudes of the Fourier transform image created by the Fourier transform means at the predetermined frequency, and a mean value of amplitudes calculated by the calculating means. The authenticity determination device for a paper sheet according to claim 1, further comprising a determination unit that determines the authenticity of the paper sheet by comparing with a predetermined threshold value. 5. The discrimination processing means includes an average value of amplitudes at a first frequency corresponding to the micro area of the Fourier transform image created by the Fourier transform means, and a second value corresponding to an area other than the micro area. Average value calculating means for calculating the average value of the amplitude at each frequency, the average value of the amplitude corresponding to the second frequency calculated by the average value calculating means, and the average value of the amplitude corresponding to the first frequency. An amplitude ratio calculating means for calculating an amplitude ratio of the above and a judging means for comparing the amplitude ratio calculated by the amplitude ratio calculating means with a predetermined threshold value to judge the authenticity of the paper sheet. The authenticity discriminating device for paper sheets according to claim 1.