JP3167547B2 - Media authenticity discrimination device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for discriminating the authenticity of various media such as banknotes and securities by reading the image of the media and discriminating the authenticity of the media.

[0002]

2. Description of the Related Art An apparatus for automatically storing and processing securities in circulation such as securities, gift certificates, bills, etc.
Alternatively, the authenticity of these media is determined at an appropriate timing. As a specific method, for example, an image of a medium is converted into an electrical signal as an optical reflection or transmission pattern or a pattern having magnetic characteristics and taken in, and the characteristic image is compared with a reference genuine bill. For this purpose, an apparatus for reading an image on a medium and determining whether or not the read signal level is within a reference density range has been developed (Japanese Patent Application No. 5-78074).

[0003]

However, the conventional medium authenticity discriminating apparatus which operates by the above-described method has the following problems to be solved. For example, authenticity discrimination of banknotes requires high accuracy. However, there is a certain difference in the level of the read signal between the new and the heavily soiled bills, even if the bills are the same. Therefore,
If the reference density range for judging a genuine note is set to be narrow, a slightly dirty genuine note is not judged as a genuine note, and if the reference density range is widened, a fake note may be judged as a genuine note. For this reason, conventionally, an empirical allowable range is determined, and such a standard is set by trial and error. However, it is extremely uncertain that counterfeit notes do not actually exist or may emerge, so it is not possible to set quantitatively objective criteria for this type of standard. There was a problem.

[0004]

The present invention employs the following structure to solve the above problems. The apparatus of the present invention reads an image of a medium to be discriminated, and for each pixel, a level determination unit that determines whether or not the read signal level is within a reference density range. The number of pixels for which the level is determined to be within the reference density range is determined, and if the number of pixels exceeds the threshold value, the medium is determined to be a genuine bill; otherwise, the medium is determined to be a genuine bill. Unit.

[0005] The level determination section assumes that the read signal levels of the corresponding pixels of a group of genuine bills for obtaining statistical data are normally distributed around the average value, and sets the average value at the center. The upper and lower limits are selected symmetrically on both sides to determine the reference concentration range. When the probability that the read signal level of each pixel is accommodated in the reference density range is referred to as an accommodation probability, the authenticity determination unit determines the accommodation probability of a certain number of pixels as two terms.
Distribution, and the probability of accommodating this fixed number of pixels is
Expressed as a function of the number of pixels that should be considered
Is integrated over a range in which the number of pixels takes a predetermined value,
The integrated value sets the confidence level for the required truth judgment.
A threshold for the number of pixels is determined so as to exceed the threshold.

[0006]

This device selects an upper limit and a lower limit symmetrically on both sides with respect to the average value of the read signal level of a genuine bill. When the read signal level is within the reference density range, the level determination unit notifies the truth determination unit of the fact. The authenticity determination unit counts the number of corresponding pixels with a pixel number counter or the like, and compares the counted number with a reference value. Since the signal level of the genuine bill is regarded as a normal distribution, the accommodation probability can be accurately calculated. When a threshold value of the number of pixels to be determined to be within the reference density range is determined for a genuine bill, what percentage of the genuine bill is determined to be genuine can be obtained by calculation. Therefore, the threshold value of the number of pixels can be determined according to the required confidence level for authenticity determination.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the embodiments shown in the drawings. FIG. 1 is a block diagram showing an embodiment of the apparatus of the present invention. This device reads an image of the medium 1 such as a bill, for example, and finally determines whether the bill or the like is genuine or fake, and outputs a determination result 5. For this purpose, the medium 1 is scanned in the direction of arrow 2 to recognize a read signal level for each pixel and perform a certain process, and a level determination unit 3 based on the output of the level determination unit 3. A true / false determining unit 4 for determining the true / false state.

The image reader 1 of the level determination unit 3
Reference numeral 1 denotes an image line sensor or the like which can optically read an image on the medium 1 and convert the image into an electric signal corresponding to each pixel. In addition, image reader 1
The output of 1 is input to the comparator 13 via the A / D converter 12. The comparator 13 is set so that an upper limit value 14 and a lower limit value 15 of the read signal level which are set in advance in a manner described later are input. Further, the authenticity determination unit 4 receives the signal output from the comparator 13 of the level determination unit 3 and, when the comparison result for each pixel is included in the reference density range, a pixel number counter that increments by “1”. 16 are provided. The output of the pixel counter 16 is input to the comparator 17 together with the reference value 18. The comparator 17 determines that the output of the pixel number counter 16 exceeds the reference value 18 and determines that the banknote is genuine. Otherwise, the comparator 17 outputs a determination result 5 that determines that the banknote is counterfeit. I have.

More specifically, the A / D converter 12 is a circuit that converts an analog electric signal output from the image reader 11 into a digital signal and inputs the digital signal to the comparator 13. The comparator 13 receives two digital values of an upper limit value 14 and a lower limit value 15 and compares the digital value with a digital value corresponding to a read signal level output from the A / D converter 12 to obtain an upper limit value 1.
When it is between 4 and the lower limit value 15, that is, in the reference density range,
It is composed of a window comparator that performs operations such as setting the output to a high level. The pixel number counter 16 is a counter that increments by 1 when the comparator 13 performs one comparison and outputs a high-level signal at that time, and is reset immediately before starting a new medium reading. The comparator 17 receives the number of pixels output from the pixel number counter 16 after the level determination unit 3 finishes determining whether or not all the read pixels of the medium 1 are within the reference density range, and sets the reference value 18 set in advance. And a comparator for outputting a high level signal when the reference value 18 is exceeded, and a low level signal otherwise.

FIG. 2 is an explanatory diagram showing the read signal level of a group of genuine bills in order to determine the reference density range set by the level determination section 3. The horizontal axis of this graph is the pixel number, that is, the number from 0 to N assigned to each pixel in that order when the medium is scanned as shown by the arrow 2 in FIG. , Each read signal level is shown on the vertical axis. This level is represented by, for example, 256 stages of 8 bits in the case of a digital signal. Conventionally, for a large number of genuine bills, the read signal of each pixel is obtained, and the upper limit value A and the lower limit value B are empirically determined as shown in FIG. On the other hand, in the apparatus of the present invention, the read signal level obtained from a group of genuine bills is used as a statistical sample, and the reference density range is determined by a certain theory different from the allowable range.

FIG. 3 is an explanatory diagram of a level judgment process according to the present invention. The horizontal axis in the figure indicates the normalized pixel density, and the vertical axis indicates the probability density. The probability density is a ratio of each pixel density to the entire sample. Further, the accommodation probability to be obtained from now on means the probability that the read signal level of each pixel of the genuine bill is accommodated in the reference density range. The upper limit value A and the lower limit value B of the permissible range are empirically obtained. However, when a large number of genuine bills are focused on a certain pixel and the read signal level thereof is determined, as shown in FIG. It is considered that a normal distribution is shown around M. In this example, the pixel of the pixel number K shown in FIG. 2 is shown as the pixel of interest.

FIG. 4 is a diagram for explaining the contents of the accommodation probability.
The normalized pixel density shown on the horizontal axis in FIG. 3 is a value obtained by subtracting the average value μi from the actual density xi of each pixel and dividing by the standard deviation σi, where yi is the density. Therefore, the accommodation probability is represented as a standard normal distribution with an average value of “0” and a standard deviation of “1”. The upper limit value ai and the lower limit value bi are set symmetrically at equal intervals on the left and right with the average value of the highest probability density as the center, here α = 3.

The area of the hatched portion of the graph sandwiched between the upper limit and the lower limit of this graph is the probability that the read signal level of each pixel of the genuine bill is accommodated in the reference density range. Is called the accommodation probability. The accommodation probability is determined by the value of α shown in FIG. 3, where α is a multiplier arbitrarily selected according to the desired accommodation probability.

If it is considered that the density of a group of genuine bills for a certain pixel is normally distributed as described above, and if 95% of the genuine bills are determined to be genuine bills, the accommodation probability becomes 0.95. Is set, and α is obtained by back calculation, whereby the upper limit value and the lower limit value of the density level can be specifically obtained. The accommodation probability is a value obtained by integrating the normal distribution function from -α to α, as shown in the lowermost column of FIG.

The level determination section 3 shown in FIG. 1 symmetrically selects the upper limit value and the lower limit value on both sides about the average value to determine the reference density range. The following operational effects are obtained as compared with the case where the upper and lower limits of the range are used. FIG. 5 is an explanatory diagram of the determination level setting effect according to the present invention. This figure shows the effect that can be expected by providing the upper limit and the lower limit of the allowable range of each pixel density symmetrically from the average after normalization. In this example, the accommodation probability is set to be about 99.7% in both the case of the present invention and the case of the comparative example. However, in the case of the present invention, the upper limit and the lower limit are set asymmetrically empirically in the case of the comparative example, as compared with the case where the upper limit and the lower limit are set symmetrically around the average value. . This relationship is shown in FIG. Therefore, the width of the accommodation section is 6 in the present invention, 6.6 in the comparative example, and so on.
The comparative example is about 10% wider than the present invention.
Accordingly, the probability of falsely determining a counterfeit note as a genuine note increases by the width.

Next, the contents of the judgment processing in the authenticity judgment section shown in FIG. 1 will be described. FIG. 6 is a graph showing the number of pixels on the horizontal axis and the probability density on the vertical axis. This graph is obtained by obtaining the number of pixels in the reference density range for each genuine bill of a group, and plotting the ratio of the number of genuine bills corresponding to the number of pixels in the whole as the probability density. . Since each pixel of the genuine bill is accommodated in the reference density range with a predetermined constant probability, when the accommodation probability of each pixel in the allowable range is uniformly set to Pα, the total number of pixels is N.
Then, the probability that the number of pixels accommodated in the allowable range is i can be represented by a binomial distribution with a double underline in FIG. What can be expressed by this binomial distribution is the probability that the number of pixels contained in the reference density range among the pixels of the entire image becomes a certain number.

Here, when t or more pixels out of the total number N of pixels are included in the above-mentioned reference density range, processing is performed so as to judge this as a genuine note. The threshold value t may be set so as to satisfy a certain level of reliability among genuine bills. For example, as shown in FIG. 6, when the threshold value t is determined, the confidence level P is an integrated value of the expression of the area indicated by hatching in the figure, that is, the equation for the number of pixels from t to N.
Therefore, if the required confidence level for genuine bill judgment is, for example, 90%, the threshold value t can be concretely determined by applying this to the equation for obtaining P shown in FIG. . According to this configuration, the determination process can be performed based on a criterion based on specific numerical values that determine what percentage of genuine bills are genuine bills, thereby enabling objective evaluation.

FIG. 7 is a flowchart illustrating a specific operation of the apparatus of the present invention. First, in step S1, i and c are reset to “0” as parameters. This i corresponds to the pixel number, and c is a parameter indicating the number of pixels included in the reference density range. Step S
In step 2, i is incremented, and in step S3, a read signal for one pixel is obtained. This is Xi
And Then, it is determined in step S4 whether or not the pixel density Xi is between the upper limit value ai and the lower limit value bi. If it is within this range, that is, the reference density range, the process proceeds to step S5, and the number of pixels c is incremented. If not, step S5 is skipped and step S6 is performed.
Move on to In step S6, it is determined whether or not the level determination has been completed for all the pixels. When the level determination is completed for all the pixels, i becomes N.
Then, it is determined whether or not the counted pixel number c exceeds the threshold value t. Note that, in the determination in step S7, it is also determined that the bill is genuine if c is equal to t. That is,
Here, if the condition is satisfied, the process proceeds to step S8,
The target medium is determined to be genuine. On the other hand, if the condition is not satisfied, the process proceeds to step S9, and it is determined that the ticket is a fake ticket.

As described above, according to the present invention, the authenticity discrimination reference is provided so that a genuine bill can be accommodated as little as possible with a predetermined fixed probability. Therefore, it is preferable that the reference density range is set to a minimum range for preventing erroneous determination, and the threshold value t of the number of pixels is ideally set to the maximum value at which a genuine note can be determined. If the accommodation probability and confidence level required in advance are numerically determined as described above, and a threshold value, an upper limit value, and a lower limit value that satisfy the value are obtained by back calculation,
A highly reliable quantitative objective truth discrimination standard can be set.

The present invention is not limited to the above embodiment. In the above embodiment, the image of the medium is read optically and converted into an electric signal, but it may be a signal read magnetically. Further, the present invention can be applied not only to banknotes but also to authenticity determination of various securities and the like.

[0021]

The medium authenticity discriminating apparatus according to the present invention described above reads an image of a medium to be discriminated, and the level judging unit sets the read signal level of the corresponding pixel of the group of authentic bills to be centered on the average value. Assuming that it is normally distributed, the upper limit and lower limit are selected symmetrically on both sides around the average value to determine the reference concentration range, so that it is objectively quantitatively highly reliable A true / false discrimination standard can be selected, and the probability of erroneously discriminating a fake bill as a true bill can be sufficiently suppressed.

Further, when the authenticity determination unit assumes that the accommodation probabilities are uniform and independent for each pixel, the number of pixels to be determined to be within the reference density range and the probability density corresponding to the number of pixels The threshold value for the number of pixels including a substantial portion of the genuine bill is determined from the relationship described above, so that a highly reliable authenticity determination criterion can be similarly determined.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the apparatus of the present invention.

FIG. 2 is an explanatory diagram of a read signal level of a group of genuine bills.

FIG. 3 is an explanatory diagram of a level determination process according to the present invention.

FIG. 4 is a diagram illustrating the contents of an accommodation probability.

FIG. 5 is an explanatory diagram of a determination level setting effect according to the present invention.

FIG. 6 is an explanatory diagram of a true / false judgment process.

FIG. 7 is an operation flowchart of the apparatus of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Medium 3 Level judgment part 4 Authenticity judgment part 5 Judgment result 11 Image reader 13, 17 Comparator 14 Upper limit value 15 Lower limit value 16 Pixel counter 18 Reference value

──────────────────────────────────────────────────続 き Continued on the front page (58) Fields investigated (Int. Cl. ⁷ , DB name) G07D ^7/ 00-7/20 G06F 19/00 G06T 7/00

Claims (1)

(57) [Claims] A level determining unit that reads an image of a medium to be discriminated and determines, for each pixel, whether or not a read signal level is within a reference density range; The number of pixels for which the level is determined to be within the reference density range is determined. If the number of pixels exceeds a threshold value, the medium is determined to be a genuine note; otherwise, the medium is determined to be a genuine note. A determination unit, the level determination unit, assuming that the read signal level of each corresponding pixel of a group of genuine bills for obtaining statistical data is normally distributed around an average value, The reference density range is determined by selecting the upper limit and the lower limit symmetrically on both sides around the average value, and the authenticity determination unit determines the probability that the read signal level of each pixel is accommodated in the reference density range. Called the probability of containment , The probability of accommodating a certain number of pixels
Obtained by the binomial distribution, and the accommodation probability of this fixed number of pixels is
Function of the number of pixels that should be judged to be within the reference density range
When the number of pixels takes a predetermined value,
And integrate the value to determine the required truth value.
The threshold of the number of pixels is defined so as to exceed the confidence level of
A true / false discrimination device for media.