JPS59231692A - Discriminator for sheet papers - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of the Invention] The present invention relates to a paper sheet discriminating device for discriminating the type and direction of a banknote, for example in a banknote processing machine.

[Technical background of the invention and its problems]

Conventionally, various types of discrimination devices of this type have been proposed, one of which is a method of extracting characteristics of banknotes and discriminating the type and orientation based on the extracted data. This method is important for identifying banknotes and will be essential in the future. However, in this method, since the location where the partial feature exists usually differs depending on the type of banknote, it is very difficult to distinguish the type and direction when there are many types to be handled. Furthermore, when improving a banknote, it is necessary to devise a new detection unit. Therefore, it is appropriate to use the method to detect authenticity of banknotes by detecting their magnetism, fluorescence, etc., rather than distinguishing only one type and direction.

Another type of discrimination device is the one shown in Japanese Patent Application No. 80069/1983. This separates the reflected light from banknotes into three colors: red, green, and 1R.

By performing a difference calculation between the interval-integrated value for each color and a standard pattern, nine types of genuine and false banknotes, nine fronts and backs, are determined. Although this discrimination device compensates for the drawbacks of the method using feature extraction described above, it has the following problems.

In other words, in order to distinguish various banknotes (including foreign banknotes) using the difference calculation method described above, it is difficult to perform interval integration for each color, so a combination of each color for one banknote is used as a new t4 turn for discrimination. There is a need.

[Purpose of the invention]

The present invention has been made in view of the above circumstances, and an object thereof is to provide a paper sheet discrimination device that can be easily adapted to cases where there are many types of paper sheets or when improvements are made.

[Summary of the invention]

The paper sheet discrimination device according to the present invention photoelectrically converts each reflected light from the front and back surfaces of paper sheets, samples each of the photoelectric conversion outputs, and uses each sampling output as data for use in type discrimination. is configured to determine the type of paper sheet by calculating the absolute difference for each sample data between each converted data and a preset standard pattern. It is.

[Embodiments of the invention]

An embodiment of the present invention will be described below with reference to the drawings.

In FIG. 1, P is a banknote conveyed in the direction of the arrow shown in the figure;
11 is a light source that illuminates the surface of this banknote P, 21 is a banknote P
12 is a light source that illuminates the back side of the bill P; 22 is a bill P;
This is a backside detector that separates the reflected light from the backside of the camera and converts it into electricity. The front side detector 21 converts the reflected light from the surface of the banknote P into infrared (IR) light using a guichroic mirror or a glass filter, for example.

It separates the colors into red (R), green (G), and blue (B), and photoelectrically converts each of them using a photoelectric conversion element.
B is obtained. Further, the back side detector 22 has a similar configuration, and similarly has four types of /4 turn data IR, R, G.
, B are obtained. Here, the Il't of the gloss turn data obtained from the front side detector 21 is shown in FIG. In this way, the data IR, R, G, and B obtained from each detector 21*22 are sent to sample hold circuits (S/H) 31 and 32, respectively. The sample hold circuit 31+32 is a sub CPU (central processing unit) 41 e 42
Each of the input data IR, R, G, and B'i is sampled and held by the Carano't ringing pulse, and sample data as shown in FIG. 3 is output, for example. 1 to 64 indicate the sample positions. For example, when the length of the banknote P is 192 dew, sampling is performed 64 times at 3-peg pitch. It is determined by the processing speed and the decomposition ability required for banknote discrimination.

Thus, the sub CPU 41*4* receives each sample data IR, R, G, and B obtained from the sample hold circuits 31 and 32, and converts them into A/D using an A/D converter built in, By performing the calculation shown in the following formula using each data IR, R, G, and B after A/D conversion, new front/turn data FP, 1%, and back zJ? Create turn data BP, n. This FP
, n, BP, n is a formula determined in advance by carefully examining banknotes.

vp, n=Z (IR*R+G*B)nBP, n=
f (IR, R, G, B) nn: Number of sampling (1
,2,...64) An example of the new surface nocturne data FP,n created in this way is shown in FIG.

The most types of banknotes P we handle are FP and 1m.

There are cases in which discrimination is difficult using only BPln, and cases in which FP and BP2n are used to improve discrimination accuracy.

FP2n== f (IR,R,GIB)nBP2n
``''' (■R*ReG,B)n In this way, each pattern data FP created by the sub CPU 41e 4m
, BPln are sent to the normalization/canonicalization circuit 5K+52, respectively. Normalization/canonical circuit 5. ．． 5@ is each pattern data FP, n, B
P and n are each converted into t4 dern data used for type discrimination. That is, there are new banknotes and old banknotes, and therefore /IP turn data FP + BP
, n vary greatly depending on the quality n of these banknotes. Therefore, by dividing by the total area of the arm turn data FP, n, BP, n, the difference output due to the quality of the banknotes is corrected and normalized correctly), and the alternating current component of the waveform obtained in this way is amplified (canonicalization). This makes the 4-turn data NFP, u easier to distinguish between bright and dark.

NBP and n are respectively made. An example of the surface/main turn data NFP,n created in this way is shown in the fifth section.
As shown in the figure. Therefore, the normalization/canonicalization circuit 51e5
・Mother turn data created with NFp1. , NBP
, n are stored in a RAM (random access memory) 61e62, respectively.

On the other hand, 7 is a RAM, which is managed by the main CPU 8. That is, the sub CPU 41+ 4t is
When the conversion processing of the normalization/canonicalization circuits S1 and St is completed, CAM (direct
RAM 61 by making a memory access) request.
I6. Each 24 turn data NFP, nS NB
Transfer P and n to RAM 7 at high speed. When this data transfer is completed, the main CPU 8 transfers each data NFP, n, NIIP, n in the RAM 7 and the standard/turn data SFP, n, and front surface previously stored in the ROM (read-only memory) 9. The standard number and turn data SBP, n on the back side are respectively supplied to the difference calculation circuit IO. As shown in FIG.
/Archive data from AM7 NFP, NBP,
An operation is performed to find the sum of the absolute values of the differences between n and the standard J-turn data SFP, SBP, and n from the ROM 9 for each sample data. That is, the sum S of the absolute values of the differences is determined by the following formula.

In addition, K is the number of standard pattern data, for example, 7 types,
In the case of 4 directions, 1 no parable turn, 28 (7X4X1) - f
Since there are the same number of turns on both sides, a total of 56 pieces of standard 14 turn data exist. Each of these 56 standard arm turn data consists of 64 bytes, so 56x64=3584, 3
This results in 584 difference calculations. 2 for one difference operation
If μl is required, it will take 7 ms for 3584×2 μm, and because it is necessary to process banknotes at high speed, the difference calculation circuit 10 is often configured by hardware logic.

When the sum of the absolute values of the differences is determined by the difference calculation circuit 10, the overall judgment circuit 11 uses the calculation result to judge the type and direction. That is, since 28 pieces of S1 are determined as described above based on the surface turn data NFP,n, the type and direction are determined from the standard arm turn data whose value of S is close to that of ceramic. The same determination is made in the case of the back side data NBP,n. Specifically, the determination is made through the following steps. Namely.

Using the results for the front side of the banknote and the result for the back side of the banknote, for example, select 4 items for each front and back from those whose S value is close to l'' (those with high similarity), and give them a score of 4°2.1.0. The type and direction are determined when the total score is high and the type and direction match on both sides.

With the above configuration, the main CPU & sends the sub CPU 41 e 4 * the main turn data to be captured, the capture pitch, the number of captures, and the calculation formula (
vpe BP, n), etc., so if these data are set in, for example, a non-volatile memory, the data input conditions can be easily changed. Therefore, even when there are many types of banknotes to be discriminated or when banknotes are improved, it can be handled very easily.

In the above embodiment, a case has been described in which the type of banknote is determined, but the present invention is not limited to this. For example, the present invention is not limited to this. It can also be applied when

〔Effect of the invention〕

As described in detail above, according to the present invention, it is possible to provide a paper sheet discrimination device that can be easily adapted to the case where there are many types of paper sheets or when improvements are made.

[Brief explanation of the drawing]

The figures are for explaining one embodiment of the present invention, and FIG. 1 is an overall configuration diagram, FIG. 2 is a diagram showing an example of turn data obtained from banknotes, and FIG. Figure 4 shows an example of new mother turn data created based on the sample data. Figure 5 shows the normalized arm turn data in Figure 4. FIG. 6 is a diagram showing an example of pattern data after being normalized and normalized, and is a diagram for explaining difference calculation. P...Banknotes (paper sheets), II e12...Light source, 2
1+22...Detector, 31*32...Sample hold circuit, 41.42...Sub CPU, 51.5
2... Normalization/canonicalization circuit, 61.62.7...R
A.M. 8... Main CPU, 9... Standard motherboard RO
M. IO...Difference calculation circuit, ZI...Comprehensive judgment circuit.

Claims (2)

[Claims] (1) A first photoelectric conversion means that photoelectrically converts light reflected from the front surface of the paper sheet to be discriminated; a second photoelectric conversion means that photoelectrically converts light reflected from the back surface of the paper sheet to be discriminated; ．． @2 Sampling means for sampling each output of the photoelectric conversion means. By calculating the absolute difference for each sampled data between each data from this calculation means and the standard 4-turn data set in advance, 1. A paper sheet discriminating device, comprising: a determining means for determining the type of paper sheet. (2) Conversion by the calculation means into data used for determining the type of each output of the sampling means involves A/D converting each sample data obtained from the sampling means, and then calculating each sample data to generate new padded data. 2. The paper sheet discriminating device according to claim 1, wherein the /f-turn data is normalized and canonicalized.