JPH08221632A - Genuineness judging method for printed pattern - Google Patents

Abstract

PURPOSE: To improve the genuineness judging performance by dividing the entire surface of inputted printed matter into plural areas and judging whether or not each area is similar to the area of a reference category. CONSTITUTION: A printed matter which has been inputted is conveyed by a recognition and conveyance unit 1 and a CCD image sensor 3 outputs an analog signal of transmitted light of the printed matter. An A/D converter 4 converts the analog signal into a digital signal, which is written in a buffer memory 5. A preprocessing part 6 cuts only a necessary area out of the printed matter. A denomination discrimination part 7 specifies one reference category. A zone division part 8-1 divide the entire surface of the printed matter into plural areas. A similarity calculation part 8-2 calculates cosine between vectors including density data on pixels of the printed matter as elements and vectors as density data elements of the reference categories for every area as similarities. An addition part 803 judges whether or not a similarity exceeds a prescribed value and adds the number of zones whose similarities exceed the prescribed value. A comparison and decision part 804 compares the number of similar areas with a prescribed value to judge the genuineness of the printed pattern.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for discriminating a printing pattern such as bills and securities, and more particularly to a method for discriminating its authenticity.

[0002]

2. Description of the Related Art Conventionally, as a method of discriminating a pattern of a printed matter, it is treated as a multidimensional space vector having density data of the printed matter as an element (hereinafter referred to as a discriminated vector),
A method is known in which the degree of similarity with a standard pattern vector (the pattern vector of a real printed matter) is calculated to obtain a relative matching rate. This is based on the cosine of the standard pattern vector and the discriminated vector (if they are similar, this value is close to 1), and all categories (in the case of banknotes, denomination and front (back), normal (reverse)) ), And the one having the maximum value and exceeding the specified value is defined as the category of the object to be discriminated.

[0003]

However, the conventional printing pattern authenticity determination method has the following problems (a) and (b). (A) When the number of dimensions of the feature space in which a printed matter is treated as a multidimensional space vector increases, it becomes difficult to determine the similarity based on the size of the cosine angle (for example, a pattern of a partial area of the printed matter is a real printed matter). , The size of the cosine angle due to the difference is less likely to appear), the separation characteristic of the similarity deteriorates, and high-precision discrimination becomes difficult. (B) Since the patterns on the entire surface of the printed matter are compared, when the similarity value does not satisfy the specified value and is determined to be abnormal, it is impossible to determine which portion is the abnormal area for the printed matter.

[0004]

In order to solve the above-mentioned problems, the method for determining the authenticity of a printed matter pattern according to the first aspect of the present invention includes density data creation processing for creating density data of input printed matter for each pixel. Zone division processing for dividing the entire surface of the printed matter into a plurality of areas is executed. Then, reference data of a reference category for collating the pattern of the input printed matter and density data corresponding to pixels of the printed matter are prepared, and for each of the areas divided by the zone division processing, Based on the density data of the printed matter and the density data of the reference category, a similarity calculation process for calculating the similarity within the area is executed. Further, an addition process of comparing the similarity with a prescribed value to determine whether the area of the printed matter is similar to the area of the reference category, and calculating the number of similar areas; Based on the number of similar areas calculated by the above, the comparison / determination processing for determining the authenticity of the printed matter is executed.

[0005]

According to the first aspect of the present invention, since the authenticity determination method for the printed matter pattern is configured as described above, the entire surface of the printed matter is divided into a plurality of areas by the zone division processing, and the printed matter is divided into each area. Based on the density data and the density data of the reference category, the degree of similarity between the area of the printed matter and the area of the reference category is calculated. Further, the degree of similarity is compared with a prescribed value given in advance to determine whether the area of the printed matter and the area of the reference category are similar. And
The authenticity of the input printed matter is determined based on the number of similar areas. Therefore, the above problem can be solved.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a block diagram of a print pattern authenticity determining apparatus for carrying out a print pattern authenticity determining method according to an embodiment of the present invention. The authenticity determination method of the print pattern of the present embodiment is different from the conventional authenticity determination method in that the entire surface of the printed matter is divided into a plurality of areas and the authenticity is determined for each area. is there. This printing pattern determination device has a recognition transport unit 1 and a control circuit 2. A printed matter on which a pattern is printed is input to the recognition transport unit 1. A CCD sensor 3 is connected to the output side of the recognition and transport unit 1 and the control circuit 2.
In the CCD image sensor 3, CCDs as light receiving elements are arranged in an array above the printed material in an array in a direction perpendicular to the printed material transportation direction based on a timing signal T1 input from the control circuit 2. It outputs a signal of transmitted light or reflected light of a printed matter that passes through. An A / D converter 4 for converting into digital data based on a timing signal T2 input from the control circuit 2 is connected to the output sides of the control circuit 2 and the CCD image sensor 3. A buffer memory 5 is connected to the output sides of the control circuit 2 and the A / D converter 4, and a preprocessing section 6 is connected to the output side of the buffer memory 5.
The denomination determining unit 7 is connected to the output side of the preprocessing unit 6, and the authenticity determining unit 8 is connected to the output side of the denomination determining unit 7. The authenticity determination unit 8 includes a zone division unit 8-1,
Similarity calculation unit 8-2, addition unit 8-3, and comparison determination unit 8
-4. The zone division section 8-1 is connected to the output side of the denomination determination section 7, and the zone division section 8-
The similarity calculation unit 8-2 is connected to the output side of 1.
An adder 8-3 is connected to the output side of the similarity calculator 8-2, and a comparison determination unit 8-4 is connected to the output side of the adder 8-3. The determination result is output from the comparison determination unit 8-4.

A method for determining the authenticity of a print pattern according to an embodiment of the present invention will be described below. [Density Data Creation Process] A printed matter on which a print pattern is printed is inserted into the recognition conveyance unit 1, and the printed matter is conveyed by the rollers provided in the recognition conveyance unit 1. The printed matter conveyed by the recognition conveyance unit 1 is irradiated to a position above or below the printed matter by a plurality of LEDs (not shown) arranged in a direction perpendicular to the conveying direction. The printed matter is illuminated by the LED and outputs reflected light or transmitted light according to the shade of the printed pattern of the printed matter. This reflected light or transmitted light is condensed by a condenser lens (not shown) and received by the CCD image sensor 3. In the CCD image sensor 3, a plurality of CCDs arranged in a line at right angles to the transport direction receive the light condensed by the condenser lens and generate an analog signal. This analog signal is output for each scan line based on the timing signal T1 output from the control circuit 2. In the A / D converter 4, the CCD image sensor 3 is output from the control circuit 2 during one scan line.
Signal T having pulse signals for the number of effective pixels of
Based on 2, the analog signal on one scan line output from the CCD image sensor 3 is sampled and converted into a digital signal.

This digital signal corrects the difference in the light emission amount due to the position of the LED (the LED arranged facing the center of the printed matter emits most light to the printed matter) and the variation in the LED, CD on one scan line
A digital signal for one line in the conveyance direction of the printed matter is sent to the buffer memory 5 by a timing signal T3 via a sequential shading correction circuit (not shown) for making the amount of received light of D uniform. At this time, in the buffer memory 5,
In order to accurately extract the edges of the printed matter, a region of the printed matter that is larger by a certain size (hereinafter referred to as the background portion) is also output as density data. Buffer memory 5
The entered value is processed into a required form by the preprocessing unit 6. The pre-processing unit 6, for example, removes a background portion from the printed matter and accurately cuts out only a necessary portion, a position correction cutout circuit, an integration / average value for reducing data variation, And a normalization circuit and the like. First, the position detection circuit accurately cuts out only a necessary portion of the area other than the background portion of the printed matter. Then, the position correction cutout circuit corrects the cutout portion from the deviation from the normal position when inserting the printed matter. Next, the integration / averaging circuit performs integration / averaging on the position-corrected portion to reduce data variation. Finally, the normalization circuit normalizes the density data by performing multiplication and addition / subtraction in order to suppress variations in the density of the printed matter, and outputs the density data for each pixel to the denomination discrimination unit 7.

[Kin Type Judgment Processing] The denomination judging section 7 judges the kind, front / back, and direction of the printed matter. First, the density data of pixels output from the preprocessing unit 6 is binarized. Then, regarding the reference category of the printed matter to be discriminated (for example, for banknotes, for each denomination of 1,000-yen, 5,000-yen, and 10,000-yen notes, 12 types in total for front and back, and direction), Similarly, the binarized data of the density data pre-processed is stored in the ROM, and the binary data of each category is stored.
The cosine of the multidimensional vector in which the binarized data is arranged by the number of pixels and the multidimensional vector in which the binarized data of the printed matter is arranged by the number of pixels is obtained, and the authenticity of the printed matter is discriminated from the category having the largest value. A category (hereinafter, this category is referred to as a specified category).

Next, the authenticity determination section 8 determines the authenticity of the printed matter. First Embodiment FIG. 2 is a flowchart showing the processing of the authenticity determination unit 8 according to the first embodiment of the present invention. Hereinafter, the processing content of the authenticity determination unit of the printed matter according to the first embodiment of the present invention will be described with reference to the drawings. [Zone division processing] In step S1, the zone division unit 8-1 divides the entire area of the printed matter into pre-divided regions (hereinafter referred to as zones) into the denomination categories specified by the denomination determination unit 7. ), And the following processing is performed. Here, each of the divided areas is referred to as zone 1, zone 2, ..., Zone N. FIG. 3 is a diagram showing an example of zone division of a bill. Here, the entire area of the banknote is divided into four in the horizontal direction and four in the vertical direction, for a total of 16
The figure shows the case of division into individual zones. When the total number of pixels of the printed matter is P and the number of divisions is N, the number of pixels in each divided zone is P / N. Also, a ROM (not shown)
As reference data for each reference category to be collated, density data for each pixel in each zone is prepared as reference data. Reference data RE (i) (j) (i = 1) of the category specified by the denomination determination unit 7 from the ROM
~ N, j = 1 to P / N) are read. The data of the judgment object for each zone of the printed matter output from the pre-processing unit 7 is TE
(I) (j) (i = 1 to N, j = 1 to P / N).
The magnitude (norm) RL (i) of the vector having the zone-specific reference data as an element is obtained by the following equation (1).

[0011]

[Equation 1] [Similarity calculation processing] The similarity calculation unit 8-2 performs the following steps S2 to S10. In step S2, each variable j = 1, i = 1, k = 1, CN = 0, RE =
Initialize to 0. Here, j is a variable indicating the number of pixels included in each zone, i is a variable indicating the number of zones, k is a variable indicating the number of rejected zones, CN is a variable indicating the number of normal zones, and RE is An array that stores the numbers of the rejected zones. In step S3, i> N
It is determined whether or not true and false are determined for all the zones. When the authenticity is not determined for all the zones (i ≦ N), the process proceeds to step S4, and when the authenticity is determined for all the zones, the process proceeds to step S18. In step S4, a product-sum variable S of the data of the object to be discriminated and the reference data for each zone and a variable TL for obtaining the sum of squares of the data of the object to be discriminated are initialized. S is a variable that represents the inner product of the vector of the data of the object to be discriminated for each zone and the vector of the reference data, and TL is a variable that represents the norm of the vector of the data of the object to be discriminated.
In step S5, it is determined whether j> P / N. When the processing in steps S6 and S7 is completed (j> P / N) for all the pixels in each zone i, the process proceeds to step S10, and if there are unprocessed pixels (j ≦ P / N).
N), and proceeds to step S6.

In step S6, the product of the data TE (i) (j) of the object to be discriminated for the j-th pixel in the zone i and the reference data RE (i) (j) corresponding to that pixel is calculated, and the product thereof is calculated. The product is added to the contents of variable S. In step S7, the data TE (i) (j) of the object to be discriminated is squared, and the result and the contents of TL are added. Step S8
In, j is incremented by 1 and the process proceeds to step S9. In step S9, the process returns to step S5. Step S10
In S, S / (TL ^1/2 × RL (i)) is obtained and is substituted into M. The variable M is a P / N having data TE (i) (j) (j = 1 to P / N) of the object to be discriminated and reference data RE (i) (j) for the zone i, respectively.
Cosine between dimensional vectors, data TE (i) (j) of the object to be discriminated in zone i and reference data RE (i) (j)
Is the similarity with. M becomes 0 ≦ M ≦ 1, and the zone i
For, if the data TE (i) (j) of the object to be discriminated is similar to the reference data RE (i) (j), it will be close to 1.

[Addition Processing] The addition unit 8-3 performs the following steps S11 to S17. In step S11, it is determined whether M is equal to or larger than a specified value α (for example, 0.8). If M is equal to or greater than the specified value α, step S12
If M is less than the specified value α, the process proceeds to step S13. In step S12, the data TE (i) (j) of the object to be discriminated in the zone i is the reference data RE (i).
Assuming that it is similar to (j), the similar number C of the zones
Increase NT by 1. In step S13, assuming that the data TE (i) (j) of the object to be discriminated in the zone i is not similar to the reference data RE (i) (j), the rejecting zone number i is substituted into RE (k). . In step S14, the number k of rejected zones is incremented by 1. In step S15, i is incremented by 1 to indicate the next zone. In step S16, initialization is performed with j = 1. In step S17, step S
Return to 3.

[Comparison / Discrimination Processing] In the comparison / discrimination unit 8-4,
The following steps S18 to S20 are performed. In step S18, the number CN of similar zones is the specified value β.
(For example, when N = 16, β = 15, this value varies depending on the criterion of the fitness judgment). If CN is equal to or greater than the specified value β, the process proceeds to step S19, and if smaller than the specified value β, the process proceeds to step S20. Step S
In 19, the number CN of similar zones is equal to or greater than the specified value β, so it is determined that the printed matter is a genuine note. In step S20, since the number CN of similar zones is smaller than the specified value β, a signal for rejecting the printed matter is sent to the higher-level device. At this time, the number RE (k) of the rejected zone is output together with the rejected signal. As described above, in the first embodiment, the following advantages (1),
There is (2). (1) The entire area of the printed matter is divided into a plurality of areas (zones), and similar comparison is performed for each zone to determine the authenticity of the entire area. Therefore, the entire bill is regarded as one area. Higher recognition ability can be expected as compared with the conventional method of determining by the above. (2) An abnormal area different from the reference data can be extracted by storing information such as its zone number (similarity if necessary).

Second Embodiment FIG. 4 is a flow chart showing the processing of the true / false determination unit of the second embodiment of the present invention. The authenticity determination unit of the second embodiment differs from the authenticity determination unit of the first embodiment in that each zone i
Regarding the reference data RE (i) (j) and the data TE (i) (j) of the object to be discriminated in the zone are defined as the similarity. Hereinafter, the authenticity determination method for the printed matter pattern according to the second embodiment of the present invention will be described with reference to the drawings. First, the density data creation process is performed as in the first embodiment. Then, the following processing is performed. [Zone Division Processing] In step S31, the zone division unit 8-1 divides the entire area of the printed matter into zones of the denomination specified by the denomination determination unit 7, and performs the following processing. . When the total number of pixels of the printed matter is P and the number of divisions is N, the number of pixels in each divided zone is P.
/ N. Further, in the ROM (not shown), the density data of each pixel in each zone of each category and the average value RE (i) of the zone-specific reference data are prepared as reference data. From this ROM, reference data RE (i) (j) (i = 1 to N, j) of the category specified by the denomination determination unit 7
= 1 to P / N), and the average value RE of the zone-specific reference data
Read (i). Then, the reference data RE (i)
The comparison reference data RE ′ (i) (j) is obtained from (j) by the equation (2), and the size (norm) RL (i) of the comparison reference data is obtained by the equation (3). RE ′ (i) (j) = RE (i) (j) −RE (i) (2)

[Equation 2] [Similarity Calculation Processing] The similarity calculation unit 8-2 performs the following steps S32 to S51. In step S32, each variable is initialized to j = 1 and i = 1. here,
j is a variable indicating the number of pixels included in each zone, and i is a variable indicating the number of zones. In step S33, i>
Determine if it is N, for all zones,
It is determined whether the average value of TE (i) (j) of the object to be discriminated is obtained. When the average value has not been obtained for all zones (i ≦ N), the process proceeds to step S34, and when the average value has been obtained for all zones (i> N), the process proceeds to step S43. In step S34, the variable MN for obtaining the sum of the data of the objects to be discriminated for each zone is initialized (0).
To do. In step S35, for the zone i, it is determined whether the sum for the zone i is obtained by j> P
It is determined by / N. If the sum is required for the zone i, the process proceeds to step S39. If the sum is not required, the process proceeds to step S36.

In step S36, the contents of MN are set to T.
Add E (i) (j). In step S38,
It returns to step S35. In step S39, the average value MN / of the discriminated data TE (i) (j) of zone i
Substitute (P / N) into TM '(i). Step S40
At i, i is incremented by 1 to indicate the next zone number. In step S41, j = 1 is initialized. In step S42, the process returns to step S33. In step S43, i = 1, j = 1, k = 1, CN =
0, RE = 0 is initialized. In step S44,
It is determined whether i> N, and it is determined whether or not the data TE (i) (j) of the object to be discriminated is true or false for all the zones. If true / false is determined for all the zones, the process proceeds to step S59, and if there is an undetermined zone, the process proceeds to step S45. Step S4
5, the difference TE ′ (i) (j) between the data of the object to be discriminated for each zone and the average value and the comparison reference data RE ′ (i)
A variable TL for obtaining the sum of squares of the product sum variable S and TE ′ (i) (j) with (j) is initialized to 0.

In step S48, the contents of the variable TL and the square of the data TE '(i) (j) of the object to be discriminated are added. In step S49, j is incremented by 1. In step S50, the process returns to step S46. In step S51, S / (TL ^1/2 × RL (i)) is obtained and is substituted into M. The variable M is a vector between a vector having data TE ′ (i) (j) (j = 1 to P / N) as an element and a vector having reference data RE ′ (i) (j) as an element for the zone i. Cosine of zone i
Is a correlation coefficient indicating the degree of similarity between the data TE (i) (j) of the object to be judged and the reference data RE (i) (j). M is −1 ≦ M ≦ 1, and is close to 1 if the data of the object to be discriminated is similar to the reference data. If M is -1, the negative correlation is 100%, which means that the density value of the pattern is reversed with respect to the average value.

[Addition Processing] The addition unit 8-3 performs the following steps S52 to S58. In step S52, it is determined whether M is equal to or larger than a specified value α (for example, 0.8). If M is equal to or greater than the specified value α, step S53
If M is less than the specified value α, the process proceeds to step S54. In step S53, the data TE (i) (j) of the object to be discriminated in the zone i is the reference data RE (i).
Assuming that it is similar to (j), the similar number C of the zones
Increase NT by 1. In step S54, assuming that the data TE (i) (j) of the object to be discriminated in the zone i is not similar to the reference data RE (i) (j), the zone number i to be rejected is substituted into RE (k). . In step S55, the number k of rejected zones is incremented by 1. In step S56, i is incremented by 1 to indicate the next zone. In step S57, initialization is performed with j = 1. In Step S58, Step S
Return to 44.

[Comparison / Discrimination Processing] In the comparison / discrimination unit 8-4,
The following steps S59 to S61 are performed. In step S59, it is determined whether the number CN of similar zones is equal to or greater than the specified value β. If CN is the specified value β or more,
If it is smaller than the specified value β, the process proceeds to step S61. In step S60, since the number CN of similar zones is equal to or greater than the specified value β, it is determined that the printed matter is a genuine note. In step S61, since the number CN of similar zones is smaller than the specified value β, a signal for rejecting the printed matter is sent to the host device. This time,
The number RE (k) of the rejected zone is output together with the rejected signal. As described above, the second embodiment has the following advantages in addition to the first embodiment. Since the similarity for each zone takes a value of -1 or more and 1 or less, the range of discrimination is expanded twice as compared with the first embodiment, and higher-precision discrimination can be expected. Therefore, it can be expected that the authenticity determining ability of each zone is improved.

The present invention is not limited to the above embodiment, and various modifications can be made. The following are examples of such modifications. (1) The present invention can be used for an automatic cash deposit / withdrawal machine (ATM) mainly used in financial institutions, and a bill discriminating method in a money changer. Further, it can be applied as a discrimination method in various automatic discriminating machines for printed matter in which the standard category of discriminant can be set in advance in a limited manner. (2) In this embodiment, the denomination discrimination unit 7 discriminates the denomination category, but the denomination discrimination unit 7 is not used, and the authenticity discrimination unit 8 discriminates the authenticity of all categories. You may (3) The authenticity of each zone is determined by whether the sum of the differences between the data TE (i) (j) of the object to be discriminated and the reference data RE (i) (j) is within the specified range. You may (4) Whether the Mahalanobis distance for obtaining the sum of squares of the difference between the data TE (i) (j) of the non-discriminator and the reference data RE (i) (j) exceeds the specified value is used to determine the authenticity of each zone. It may be determined.

(5) The denomination determination unit 7 may use multivalued data output from the preprocessing unit 6 instead of the binarized data. (6) Generally, since the density value of a pattern of a printed matter such as a bill has a statistical variation for each zone or denomination, the prescribed value α is changed for each reference category or zone in consideration of this variation. You may do it. (7) In the process of step S7 in FIG. 2, the magnitude of the vector of the data TE (i) (j) of the object to be discriminated is constantly normalized (for example, the norm is set to 1) for each zone. Therefore, it can be omitted. Also, the process of step S48 in FIG. 4 can be omitted by normalizing the magnitude of the vector of TE '(i) (j) constant for each zone. (8) RL (i) in equations (1) and (3) may be stored in a ROM or the like, or may be calculated appropriately.

[0022]

As described in detail above, according to the first to fourth inventions, the entire surface of the input printed matter is divided into a plurality of areas, and each area is similar to the area of the reference category. Since it is discriminated whether or not it is present, it is possible to improve the recognition ability for discriminating between true and false.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a print pattern authenticity determination device for performing a print pattern authenticity determination method according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a process of a true / false determination unit according to the first embodiment of the present invention.

FIG. 3 is a diagram showing an example of zone division of a printed matter.

FIG. 4 is a flowchart showing a process of a true / false determination unit showing a second embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Recognition transport unit 2 Control circuit 3 CCD image sensor 4 A / D converter 5 Buffer memory 6 Preprocessing unit 7 Money type determination unit 8 True / False determination unit 8-1 Zone division unit 8-2 Similarity calculation unit 8-3 Adder 8-4 Comparison judgment unit

Claims (4)

[Claims] 1. A density data creation process for creating density data of an input printed matter for each pixel, a zone division process for dividing the entire surface of the printed matter into a plurality of areas, and a pattern of the inputted printed matter. Density data corresponding to pixels of the printed matter are prepared as reference data of the reference category, and for each of the areas divided by the zone division processing, the density data of the printed matter in the area and the density of the reference category. Based on the data, a similarity calculation process for calculating the similarity in the area, and comparing the similarity with a specified value to determine whether the area of the printed matter and the area of the reference category are similar or dissimilar. Based on the number of similar or dissimilar regions calculated by the addition process that determines and calculates the number of similar or dissimilar regions, A comparison determination process of determining the authenticity of the printed matter authenticity determination method of the printed pattern, characterized in that it has to be executed. 2. The similarity calculation process includes a vector having elements of density data of pixels of the printed matter in each area and a vector having elements of the density data corresponding to the density data of the printed matter of the reference category. The authenticity determination method of the print pattern according to claim 1, wherein a cosine of or a value based on the cosine is calculated as the similarity. 3. The similarity calculation process is performed by the correlation coefficient of the density data of pixels of the printed matter in each area and the density data corresponding to the density data of the printed matter of the reference category in the area or the phase. The authenticity determination method for a print pattern according to claim 1, wherein a value based on a relation number is calculated as the similarity. 4. The authenticity determination method for a print pattern according to claim 1, 2 or 3, wherein the specified value is changed according to a statistical variation in the print pattern of the printed matter.