CN1959744B - Identification device - Google Patents

Abstract

In recent years, the number of unauthorized access to automatic cash handling machines, for example, the use of counterfeit bills has increased extremely, and the methods have become more sophisticated. For example, it is the following method: carrying a large number of counterfeit bills and counterfeit bills of various types and putting them into the automatic cash processing machine, investigating the counterfeit bills and counterfeit bills accepted, and only using a large number of counterfeit bills of the type at the next opportunity. The counterfeit coupons and counterfeit banknotes are put into the automatic cash processing machine. In order to solve such a problem, in this invention, the whole information of a banknote is acquired first by a sensor, and each authenticity judgment is performed using this whole information in the some judgment part prepared beforehand. Then, one of the plurality of determination results is selected at a certain timing. In general, the areas where counterfeit bills are received and the area where genuine bills are rejected differ depending on the method of estimating the distribution. Therefore, for example, even if the area where a counterfeit bill has been received is approved, it can be dealt with by changing the judgment method at an appropriate timing.

Description

Identification device

technical field

The invention relates to a method and a device for reading the surface information of valuable securities and other items to determine their authenticity.

Background technique

The use of electronic money or credit cards to purchase items is becoming more and more popular, but the use of banknotes to purchase items is still prevalent. This point is very clear from the increase in the number of banknotes issued. On the other hand, in recent years, unauthorized access to automatic cash handling machines, for example, the number of counterfeit bills used is also extremely increasing. Moreover, its methods are getting more and more ingenious. For example, counterfeit bills and counterfeit bills of various types are counterfeited and put into an automatic cash handling machine, and the accepted counterfeit bills and counterfeit bills are inspected. On this basis, at the next opportunity, a large number of counterfeit bills and counterfeit banknotes that have been accepted are forged, and they are re-introduced into automatic cash processing machines to steal money, which is becoming more and more problematic.

In order to prevent such crimes, more and more attention has been paid to the method of improving the reliability itself of automatic cash handling machines. For example, there is a method of preventing counterfeit bills from being received by changing the method of receiving banknotes used for identification by operating sensors when bills pass (for example, refer to Patent Document 1: JP-A-8-249512). In this technique, the purpose is to prevent a counterfeiter from passing through the device by forging information on only a specific part of a banknote. In addition, there is a technique of storing encrypted information in a ticket and appropriately changing the encrypted information even when the corresponding reading is performed in an automatic ticket gate (for example, refer to Patent Document 2: JP-A-8-305904). .

The method described above is characterized by the following two points: (1) changing the information reading position of the object at an appropriate timing (random timing), and (2) writing encrypted information in the object. However, in order to apply it to the determination of the authenticity of highly forged banknotes, the following points must be taken into consideration. (1) Highly counterfeit banknotes are mostly finely imitated on the front of the banknote. Therefore, for example, even if the information reading position of the object is changed, there is a high possibility that it will be falsified, so there are many cases where it cannot be dealt with. In addition, in the case of current banknotes, it is difficult to adaptively write encrypted information because a storage device is not provided. Furthermore, changing the information reading position or making the timing of changing the encrypted information random generally results in low accuracy of discrimination.

Contents of the invention

In order to solve such a problem, in the present invention, the overall information of paper sheets of securities such as banknotes is acquired by a sensor, and the overall information is used in a plurality of prepared judging units having different characteristics to perform respective judgments. Authenticity determination. Then, one of the plurality of determination results is selected at a certain point of time.

According to the present invention, for example, even if there is a weak point of a certain judgment method, that is, there is a judgment area where counterfeit bills are received, by switching the judgment method at an appropriate time, even if a large amount of counterfeit bills intentionally entered into the judgment area and forged are put in, can be rejected. Furthermore, in the case of such a typical crime, the criminal loses confidence in reading the identification and determination characteristics of the automatic cash handling machine that he considers, and as a result, there is an effect of reducing the number of crimes.

Description of drawings

FIG. 1 is a diagram showing an example of the configuration of an authenticity judging device in this embodiment.

Fig. 2 is a diagram for explaining the principle and effect of the discrimination method in Fig. 1 .

FIG. 3 is a diagram showing an overview of the automatic payout machine in this embodiment.

FIG. 4 is a diagram showing an example of a processing flow of the authenticity judging device in the present embodiment.

Detailed ways

In the present invention, a sensor for acquiring various physical information from banknotes or the like and a computer such as a microcomputer process the physical information in a coded manner to express them in a specific form.

【Example 1】

Embodiments of the present invention will be described in detail with reference to the drawings. First, FIG. 3 is a diagram showing an overview of an automatic cash dispenser 31 to which an example of the present invention is applied. A CPU 13 is provided in the frame of 31. 311 is a deposit port for the user to insert cash, 312 is a keyboard for inputting necessary passwords or amounts when depositing and withdrawing, and 313 and 314 are bank card input ports and withdrawal ports. 315 is a display device for displaying information to the user.

The basic principle of the present invention will be described using FIG. 2 . Generally speaking, the counterfeit bill acceptance rate (the ratio of genuine bills received into the device by mistake) and the genuine bill rejection rate (the ratio of genuine bills that are mistakenly judged as counterfeit bills or other paper sheets and rejected) by the judging unit ) is generally a small value, statistically having some value other than zero. Examples of several determination methods are shown in FIG. 2 . 21 in FIG. 2 shows the actual distribution of the authentic banknotes of the banknotes in the space of the feature quantities obtained from the information on the entire paper sheet. However, since the true shape of the distribution is generally unknown, the current state of affairs is to make guesses using a large number of samples. 22 in FIG. 2 shows the decision boundary inferred by a certain inference method 1 with a dotted line. In general, the bias or number of collected samples is limited, so there are areas 221 where counterfeit bills are accepted and areas 222 where genuine bills are rejected. It is quite difficult for a perpetrator to make a counterfeit ticket to enter this area 221 by ordinary means. Therefore, recently, counterfeit bills that have entered this area 221 are frequently searched for by a somewhat forceful method. That is, counterfeit bills of various types are prepared in advance, and the counterfeit bills of the collected types are checked once by putting them into the automatic cash processing machine. Next, a large number of counterfeit bills of the type to be collected are prepared and put into the automatic cash processing machine. Usually, the transaction is subsequently canceled and the return of the genuine coupons equivalent to the amount charged for the counterfeit coupons is accepted. As a method of preventing such a crime or relieving the criminal, in the present invention, a plurality of judging sections with different characteristics are prepared in advance, and a result from the judging section is selected at a certain appropriate timing. 23 in FIG. 2 shows the decision boundary inferred by a certain inference method 2 with a dotted line. In this case, there are also an area 231 where counterfeit bills were accepted and an area 232 where genuine bills were rejected. In addition, 24 in FIG. 2 shows a decision boundary estimated by a certain inference method 3 with a dotted line. In this case, there are also an area 241 where counterfeit bills were accepted and an area 242 where genuine bills were rejected. In general, the area where counterfeit bills are usually accepted and the area where genuine bills are rejected differ depending on the method of inferring the distribution. Thus, for example, even if the area 221 for accepting counterfeit bills is approved, at the timing of accepting the next drop-in, if the judgment method is changed (the method of judging the authenticity inside and outside the dotted line obtained by using the judgment method), then the above-mentioned Such counterfeit banknotes are used for disposal. This is clear from the fact that the areas 221, 231, and 241 do not overlap. In this application, a good effect can be obtained by using a combination of judging methods with slightly different characteristics in judging genuine bills. Considering the above-mentioned typical crimes as the timing for converting multiple modes stored, each transaction is valid at the start of the next transaction that causes a large number of rejections or when there are few transactions such as nighttime business.

Next, the flow of the overall processing will be described using a block diagram showing an example of the configuration of authenticity determination in the present embodiment in FIG. 1 and a flowchart in FIG. 4 . In the present invention, the banknotes put in from the deposit port 311 are conveyed to the identification unit, and the physical information of the entire paper pieces such as banknotes is acquired by the sensor unit 111 in step 411 . Here, physical information is, for example, color image information or the like. In addition, the so-called whole banknote means that not only a specific partial area set in advance is set as the object of information acquisition, but the general whole of the banknote is set as the object of information acquisition, so that even if only a specific part is finely It will not be overlooked in the case of processing. This is because, of course, if only a specific partial area is targeted, it can only correspond to a specific counterfeit banknote. Thereafter, in the feature amount generating flow 416 , the feature amount is generated in the feature amount generating unit 116 based on the acquired physical information. In the feature amount generation unit 116 , a feature amount obtained by combining physical quantities, for example, a feature amount expressed in components such as principal component analysis, is generated. Usually, these feature quantities are generated by software processing because complicated processing is required. Next, authenticity judgments 412 , 413 , and 414 are performed, respectively, in the prepared plurality of judging units 112 , 113 , and 114 using the feature values. These authenticity determinations are processed by software in a different manner while using feature quantities using the same physical quantities obtained from the sensor 1 . Therefore, complex processing can be easily performed, and various processing means can be provided. In addition, in processing using hardware, it is good in terms of reliability and processing speed against problems such as occurrence of a failure due to a physical load. As described above, the present application is characterized by having a plurality of determination processes having different characteristics regarding authenticity determination.

Authenticity is determined based on whether or not this feature value is included in authenticity determination distributions 22 , 23 , and 24 indicated by dotted lines in determination modes 1 , 2 , and 3 shown in FIG. 2 . For example, authentic bill determination distributions 22 , 23 , and 24 are defined as regions determined by learning with a layered neural network using each feature quantity group obtained from a plurality of authentic bills as input. At this time, by preparing and learning three neural networks with different numbers of neurons or different structures, the distributions 22 , 23 , and 24 for authenticity determination can be obtained. Alternatively, the number and structure of neurons may be the same, and three different regions may be obtained by changing the initial values at the time of learning. In addition, instead of using three neural networks, it is also possible to use a single neural network for judgment such as using different types of judgment methods, such as using Determination of the size of the shortest Euclidean distance in the sample distribution of real coupons. In addition, about such a judgment method, for example, it is published in Tokyo University Press "Handbook of Image Analysis" (published January 17, 1991, edited by Takagi Mikio and Shimoda Yohisa, "Classification" page 641, "Neural Network" 190 page) described in the existing method. Then, one of the plurality of determination results is selected in the selection unit 115 ( 415 in FIG. 4 ) by a selection signal controlled by the CPU 13 . Here, the number of determination modes is not limited to three, as long as it is greater than or equal to two. In addition, one of the plurality of determination units prepared may be selected in advance based on the selection signal, and only the process in the determination unit may be performed.

The discriminated banknotes are processed based on the determination result. That is, in the case of genuine bills, they are counted as banknotes inserted for deposit or remittance, stored in a cash box, and the amount of insertion is displayed on the display device 315, and deposit or remittance processing is performed. On the other hand, when it is determined that it is not a genuine bill, it is generally rejected and returned to the money-receiving and withdrawing port 311 of FIG. 3 . In addition, when the number of rejections is large, a warning message notifying the abnormality of the device is displayed on the display device 315, and the information is conveyed to relevant persons.

In this embodiment, the result is selected after performing all three judgment methods, but even if one judgment method is selected and the result is used, exactly the same result can be obtained. The selection signal opens and closes the door 122 at a timing described as a rule in a selection timing generator controlled by the CPU 13 . As a rule, for example, it may be defined as the time when the next transaction that caused a large number of rejections exceeding the processing number starts. This utilizes the phenomenon that a large number of rejections start to occur when a type of counterfeit bill that is not rejected is inspected after inserting a large number of various counterfeit bills as described in the background art. Therefore, a reference value of the number of rejections within a predetermined period is set in advance, and a comparison with this is used to determine the switching timing. In addition, it may be at the start of each transaction when there are few transactions at night. This is because most of the counterfeit banknotes are used at night when there are few people. Alternatively, it is also possible to determine the change timing using a random number without setting a specific rule. If the control signal from the CPU 13 carries rejection information and transaction information, timing determination can be realized.

As a specific example of the above rule, the judgment results from the judging unit 11 are accumulated during the execution of one transaction, and the number judged to be rejected within a certain period of time is counted in advance. Then, when this value is greater than or equal to a certain threshold value, it is changed to the determination method of subsequent transactions. In addition, in the second rule, the time without transaction is counted in advance, and when the value is greater than a certain threshold value, or based on the number of securities or the number of transactions judged within a certain period of time, When this number is smaller than the threshold value, it is changed to the determination method in subsequent transactions. These rules can be implemented using a procedure as above.

On the other hand, the selection method signal generation unit 123 that generates the selection signal indicating which judgment result to select uses the control signal from the CPU 13 as a trigger signal to randomly determine one of the three judgment results without selecting the same method result, or the sequence to be selected periodically. Alternatively, the following methods are also possible. The number of times the determination is taken is detected for each determination method in advance. Then, the determination result with the least number of times is selected. By doing so, a specific determination result is not biased to be selected.

In addition, in this embodiment, three forms are prepared, but as long as there are plural forms, other forms may be used. In addition, in the determination unit 11, a plurality of feature amount generating units may be provided and prepared. That is, a plurality of feature quantity generators that generate different feature quantities from sensor physical quantities are prepared. The processing may be performed in advance in the judging unit corresponding to each feature generating unit, or the feature generating unit and the judging unit may be separately selected by a selection circuit, and multiple feature values and multiple judging methods are used to determine authenticity. Then, select a result from the overall judgment results. With this structure, it is possible to further determine that the characteristics are different, and the effect is improved.

In addition, this invention is not limited to the above-mentioned object of implementation, ie, securities, such as a banknote, and can be widely applied. As long as information such as the surface or shape of the object can be obtained, for example, the authenticity of brand-name goods and the like can be determined using the same structure as the present invention.

As described above, according to the present invention, a plurality of methods for authenticity judgment using physical information of the entire object are prepared, and the judgment result of one method is selected at an appropriate timing. By doing so, for example, even if the weak point of a certain determination method is known due to input of many types of forgery, it has the effect of preventing crime in advance because it is changed to another determination method at the next moment. The present invention can be applied to a discriminating unit of an automatic machine that handles securities such as banknotes, such as an automatic cash handling machine and an automatic vending machine, and contributes greatly to the further development of the industry and the prevention of crime in advance.

Claims (6)

1. identification device is characterized in that comprising:

Put into portion, accept putting into of securities;

Sensor scans the above-mentioned securities of putting into and obtains physical message;

The characteristic quantity generating unit generates characteristic quantity from above-mentioned physical message;

A plurality of discriminating portion uses above-mentioned characteristic quantity, has adopted the real and fake discrimination of the different decision procedure of the characteristic of identification each other respectively; And

Selection portion is selected which the discriminating portion that carries out in a plurality of discriminating portion.

2. the identification device described in claim 1 is characterized in that:

Above-mentioned characteristic quantity generating unit and above-mentioned real and fake discrimination are carried out software processes.

3. the identification device described in claim 1 or 2 is characterized in that:

The sensor is roughly all obtained physical message from above-mentioned securities.

4. the identification device described in claim 1 or 2 is characterized in that:

Above-mentioned selection portion content in above-mentioned discriminating portion is transformed into other discriminating portion under the situation of refusal more than or equal to set number of times of forging.

5. the identification device described in claim 1 or 2 is characterized in that:

Above-mentioned selection portion has surpassed the discriminating portion that is transformed into other under the situation of set value in the time of not concluding the business.

6. the identification device described in claim 1 or 2 is characterized in that:

Above-mentioned selection portion is differentiated the selection of portion based on the selection number of times in the past of above-mentioned a plurality of decision procedures.

Images