JP3187698B2 - Paper sheet recognition 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 paper sheet recognizing apparatus which reads images from various kinds of paper sheets such as banknotes and securities as distribution media and discriminates the type or authenticity of the paper sheets.

[0002]

2. Description of the Related Art Machines such as currency exchange machines, automatic teller machines, ticket vending machines, and the like that handle paper sheets such as banknotes, securities, and gift certificates are provided with discrimination of the type of paper sheets to be handled or with authenticity. A paper sheet recognition device for performing discrimination is incorporated. In this paper sheet recognition device, an optical reflection / transmission pattern or a magnetic reflection pattern of a ticket pattern or a figure is read by a sensor or an image reader, and the pattern is converted into image data composed of an aggregate of a large number of pixels. Is converted into a signal, and based on information of many pixels constituting the image data,
The kind or authenticity is determined.

By the way, in discriminating the authenticity of a sheet, it is ideal to obtain high-resolution image data to increase the discrimination accuracy. However, as the resolution increases, the number of pixels constituting the image data increases, and the processing time for this large number of pieces of pixel information increases. On the other hand, for a general user, the shorter the processing time, the higher the convenience.

A conventional technology satisfying the conflicting demands of high resolution for improving the discrimination accuracy and high-speed processing for improving convenience is described in "Neural Network" by Shoichi Tejima and Yuho Kazuka. Research on Banknote Recognition Using GIS-Optimal Sensing Line Arrangement Problem-", IPSJ 46th Annual Meeting (Early 1993),
Proceedings, 2C-2. In the banknote recognition according to this conventional technique, a line-shaped area perpendicular to the banknote conveyance direction is used as a sensing line, and a line position with little additional noise due to dirt or scratches on the banknote is determined, and a predetermined sensing with little additional noise is specified. Identification is performed based on the pixel information of the line position. Therefore, according to this conventional technique, even if the resolution of the image data is increased in order to increase the discrimination accuracy, it is not necessary to perform the identification processing on all the pixels of the image data, and the discrimination work can be performed with higher precision and higher speed. Achieved.

[0005]

However, in the above-mentioned conventional technology, since the sensing line position of the bill is fixed, it is difficult to detect the position of the sensing line that is continuous in a band by trial and error. It is possible. For this reason, if a counterfeit ticket in which the corresponding band-like portion of the genuine note is attached only to the detected sensing line position is created, the use of the forged ticket cannot be excluded. Therefore, the emergence of a paper sheet recognition device that can reject the use of such a counterfeit ticket and is more excellent in safety has been desired.

[0006]

Means for Solving the Problems The present invention basically comprises:
The present invention is based on the idea that a different pixel group is used for each bill type or authenticity determination without fixedly specifying the pixels to be processed among the pixels constituting the image data. FIG. 1 is a drawing showing the concept of the present invention, and each rectangle indicated by numerals (1) to (L) corresponds to image data of a plurality of inspected objects, such as bills, to be respectively discriminated. Each of the image data (1 to L) is divided into a number of pixels, and the pixels shown in black in the figure indicate the selected pixels to be inspected, that is, the selected pixels, and are similarly shown in white. Pixels that have not been checked indicate non-inspection pixels.

As is evident from the change in the black and white selection pixel pattern indicated by each of the image data (1 to L), the position of the selected pixel to be inspected, that is, the pixel represented by black, changes every inspection. By circulating the change of the selected pixel pattern between (1) to (L) as shown in FIG. 1, or by selecting randomly, the pattern of the pixel group to be inspected can be changed for each inspection. In each selected pixel pattern, it is desirable that the black pixels as the selected pixels be evenly dispersed in the image data. The selected pixels to be inspected differ from one inspection to another, and the selected pixels do not simply continue in a strip shape. Doing so is virtually impossible.

The present invention employs the following configuration in order to realize the above-described concept. <Structure> The paper sheet recognition device of the present invention is a paper sheet that determines the bill type or authenticity of a paper sheet by processing pixel information of a pixel group constituting image data of the paper sheet to be inspected. A class recognition device, comprising: a selected pixel list storage unit storing a selected pixel list that specifies a plurality of pixel array patterns each including a combination of a plurality of pixels selected in advance from a pixel group; Pixel array pattern selection means for selecting a different pixel array pattern for each, based on the pixel information of the pixel group of the selected pixel array pattern, to determine the bill type or authenticity of paper sheets (Corresponding to claim 1). Hereinafter, each component will be described.

(Selected pixel list storage means) A selected pixel is, for example, a plurality of pixels constituting image data of an object to be inspected, among a plurality of pixels each specified by a pixel number, selected to be inspected. Pixel. Such a pixel group that is a combination of selected pixels selected based on, for example, a pixel number is a pixel array pattern. Some pixels may be redundantly selected by a plurality of pixel arrangement patterns. A group of selected pixel array patterns (1 to L) as shown in FIG. 1 is configured by the plurality of pixel array patterns, and information about a selected pixel list that specifies such a plurality of selected pixel patterns is stored in the selected pixel list. It is stored in the storage means.

(Pixel array pattern selecting means) The pixel array pattern selecting means is a pixel array for specifying one selected pixel pattern from a selected pixel list storing means in which a selected pixel list for specifying a plurality of selected pixel patterns is stored. This is a means for selecting information about the pattern, for example, as an array pattern number.

<Operation> In the paper sheet recognition apparatus of the present invention, the pixel array pattern selecting means selects one pixel array pattern from a plurality of pixel array patterns stored in the selected pixel list storage means. Since a plurality of pixel arrangement patterns, except for some pixels selected redundantly, as a whole constitute a pixel group composed of a combination of different pixels,
The selected pixels to be inspected differ depending on the selected pixel array pattern. Therefore, since the selected pixel to be inspected differs for each inspection, it is practically impossible to detect the inspection site even if the conventional trial and error for searching for the inspection site is repeated.

As a specific structure of the paper sheet recognition apparatus of the present invention, there is provided authenticity judgment reference data storage means for storing reference data for judging the authenticity of a paper sheet, and used as an authenticity discrimination apparatus. It is desirable (corresponding to claim 2). It is desirable that the selection of the selected pixel array pattern by the pixel array pattern selection unit be performed randomly based on a random number table (corresponding to claim 3). The random selection of the selected pixel pattern based on the random number table makes it more difficult to predict the repetition of the selected pixel pattern, that is, to predict the inspection region. Furthermore, it is desirable that the pixels specified by each selected pixel array pattern, that is, the selected pixels of each selected pixel pattern be evenly dispersed in the image data. By evenly distributing the selected pixels, it is possible to reliably prevent the selected pixels from being simply continuous in a band shape, which may be a more effective countermeasure against trial and error for detecting the inspection site. .

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below in detail with reference to the illustrated embodiments. FIG. 2 is a schematic diagram generally showing a general sheet recognition processing step of a sheet handling apparatus such as the sheet recognition apparatus according to the present invention. In the paper sheet recognition step, first, a planar image of a bill to be inspected is converted into an electric signal as image data as an aggregate of pixels by an image reading unit such as a sensor, and each pixel constituting the image data is converted. The signal is treated as input information.
The image data obtained from the input information undergoes preprocessing (step S1). This preprocessing is performed in addition to various correction processes such as shading correction, which is a shading correction of an image due to sensor characteristics or deterioration, and skew correction, which is a correction of an initial position such as a posture angle of a test object. Feature extraction processing such as differentiation / integration for extracting the feature based on the corrected image data is included.

The feature data extracted by the preprocessing is subjected to a bill type discrimination process by comparing it with preset bill type discrimination reference data (step S2). The image data that has undergone this type determination is subsequently compared with the reference data of the corresponding type from the authenticity determination reference data prepared for each type, thereby performing the true / false determination processing. Receive (Step S3). If the determination result is "true", the result is output for subsequent processing, while if the determination result is "false", acceptance of the bill as the test object is rejected (step S4).

The present invention can be applied to the bill type discriminating process (step S2), as will be described later, in such a paper sheet recognizing process. An example applied to S3) will be described. FIG.
1 is a configuration diagram schematically showing a paper sheet authenticity discriminating apparatus 10 according to the present invention. The paper sheet authenticity discriminating apparatus 10 handles, as an input signal, a large number of pixel signals constituting image data of a banknote to be inspected. FIG. 4 is an explanatory diagram showing an example of reading the image data. In FIG. 4, for example, when a paper sheet 11 such as a bill is conveyed in the longitudinal direction that is its long side 11 </ b> A and accepted by the paper sheet authenticity discriminating apparatus 10,
The image reading means (not shown) reads the image data along many reading lines parallel to the short side 11B perpendicular to the transport direction. In the example of FIG. 4, this image data is composed of an aggregate of N pixels each specified by a pixel number i of 1 to N. Electric signals of a series of pixels having pixel numbers i (1 to N) constituting the image data are input to the paper sheet authenticity discriminating apparatus 10 as input information, that is, input pixels.

Referring to FIG. 3 again, the paper sheet authenticity discriminating apparatus 10 includes a data storage section 12 for storing data necessary for authenticity discrimination, a pixel array pattern selecting section 13, a pixel processing section. 14, a pixel processing result accumulation unit 15, and a true / false determination unit 16. The data storage unit 12 includes a selected pixel list storage unit 17 that is a selected pixel list storage unit, a pixel reference data storage unit 18 that stores reference data for each pixel, and an authenticity determination reference data storage unit 19.

The selected pixel list storage section 17 stores, for example, the selected pixel array pattern shown in FIG. 1 as a column of each pixel number in order to select a pixel group to be inspected from the pixel groups of the read image data. FIG. 5 is an explanatory diagram showing an example of the selected pixel list stored in the selected pixel list storage section 17 as the selected pixel list storage means. The selected pixel list includes a pattern number (SYN) column indicated by a pixel array pattern number k (1 to L) and a selection order number j (1 to M).
The selected pixel number S _kj is determined by the two numbers kj. The array pattern number (1 to L) specifies, for example, a pixel array pattern number k corresponding to the selected pixel array pattern (1 to L) shown in FIG. 1, and a selected pixel number for each pattern number order in each pattern number k. (SPN) S _kj corresponds to the pixel number i (1 to N) shown in FIG.

For example, when the numerical value 1 of the pixel array pattern number k is selected, in this pixel array pattern (k = 1), each pixel whose pixel number i is 5, 11, 23,. A pixel array group to be the selected pixel is selected. When the pixel array pattern number (k = 2) is selected, the pixel number i is 2,
A pixel array group in which each of the pixels 27, 81,... 88 is a selected pixel is selected. As shown in FIG. 1, the pixel array group including the selected pixels specified by each pixel array pattern number k is set so as to be substantially evenly distributed over the entire area of the image data for each pixel array pattern. In particular, it is desirable to avoid that the selected pixel becomes a large number of continuous strips. These make it more difficult to detect the inspection site by trial and error using a counterfeit ticket.

The pixel array pattern selector 13 shown in FIG.
Has a selection pattern number counter (k). The selected pattern number counter (k) reads the initial value of the set pixel array pattern number k, and sequentially updates the pixel array pattern number k every time the determination process is repeated. In order to sequentially read the pixel number i of the pixel array pattern with the pattern number k read by the selected pattern number counter (k), the pixel array pattern selecting unit 13 sequentially updates the selection order from 1 to M. Selection order number counter (j)
A pixel counter (i) for reading a pixel number S _kj specified by a pattern number k and a selection order number j is provided.

The pixel array pattern selection unit 13 sequentially converts the pixel row of the pattern number k read by the selection number counter (k) with the corresponding selection order number j in accordance with the update of the selection order number counter (j). The selected pixel number (SPN) S _kj is read into the pixel counter (i) and is sequentially read by the pixel processing unit 1.
4 is output.

The pixel processing unit 14 receives the pixel number of the pixel group selected for the determination processing, and the pixel processing result storage unit 1
5. The authenticity determining unit 16 compares the pixel information obtained from the selected pixels with the data in the pixel reference data storage unit 18 and the data in the authenticity determination reference data storage unit 19 of the data storage unit 12 to determine the target. The authenticity of the inspected banknote is determined.

The configuration and operation of the pixel processing section 14, the pixel processing result accumulating section 15, and the authenticity determining section 16, and the data stored in the pixel reference data storage section 18 and the authenticity determination reference data storage section 19 According to the determination processing method, the same functional parts and data as in the related art can be applied. Among the various determination methods, hereinafter, the total number of pixels in which the image density, which is one of the pixel information, is within a predetermined range is determined, and whether the total number has reached a predetermined number is determined as true or false. For the example of applying the judgment method,
The outline will be described.

FIGS. 6 and 7 are explanatory diagrams showing examples of data stored in the pixel reference data storage section 18 and the authenticity determination reference data storage section 19 of the data storage section 12, respectively. The pixel reference data storage part 18, as shown in FIG. 6, for each pixel number i which constitutes one of the image data, the threshold value c _i is set. Each threshold c _i is
The reference value c _i of the time to achieve the binarization of the pixel density, which is information corresponding pixels. The threshold c _i is provided for each note denomination, in the example shown, the threshold for the entire image data is set, the pixel array selection pattern k
Each time, only the threshold value of the selected pixel group can be used as data. Further, the pixel reference data can be constituted by a mask value.

As shown in FIG. 7, the authenticity discrimination reference data storage section 19 stores the authenticity discrimination threshold value t for the number of pixels exceeding the threshold value shown in FIG. It is set for each. In the example of FIG. 6, in the pixel array selection pattern 1, if the number of pixels having a pixel value of “1” among the binarized pixels based on the pixel reference data is equal to or greater than the authenticity determination threshold value 35, for example, Indicates that it is a ticket.

The pixel processing unit 14 shown in FIG. 3 stores the pixel density, which is the pixel information, in the pixel reference data storage unit 18 for the pixel having the pixel number i read from the pixel array pattern selection unit 13 in FIG. The pixel density, that is, the pixel value is binarized in comparison with the corresponding threshold value of the pixel reference data shown. Further, the pixel processing result accumulating unit 15 accumulates each binarized pixel value for one pixel array pattern. The authenticity determination unit 16 stores the pixel values stored in the pixel processing result storage unit 15 and the authenticity determination reference data storage unit 19 shown in FIG.
Is compared with the corresponding true / false determination threshold value of the true / false determination reference data shown in FIG.

FIG. 8 shows a paper sheet authenticity discriminating apparatus 10 according to the present invention.
Is a flowchart showing the true / false determination processing operation, and the operation of the paper sheet true / false determination apparatus 10 will be described below along this flowchart. Step S11: First, in the selection number counter (k) of the pixel array pattern selection unit 13, the pattern selection number (SY
N) k is determined. By the determination of the pattern selection number, the corresponding pixel array pattern (1 to L) is selected.
The pattern selection number is sequentially set to 1 by the selection number counter (k) every time bills are handled.
ＬL, ie, increment by one.

Step S12: Selection order counter (j)
By this, the selection order number (SXN) j is initialized to 0. Step S13: The selection order number (SXN) j is incremented by 1 from 1 to M by the selection order counter (j). Step S14: The pixel array selection number (SY) is stored in the selected pixel list storage portion 17 by the pixel number counter (i).
N) k and selection order number (SXN) pixel number corresponding to the j (SPN) i = S _kj is read. Step S15: the pixel processing unit 14 reads the pixel values x _i of the pixel number i from the pixel array k of input media.

[0028] Step S16: the pixel processing unit 14 reads the pixel reference value c _i of the pixel number i corresponding the pixel reference data storage section 18. Step S17: Pixel value x read by the pixel processing unit 14
_i is compared with the pixel reference value c _i, and if x _i ≧ c _i is satisfied, “1” is output to the pixel processing result accumulation unit 15. If this expression condition is not satisfied, “0” is similarly calculated. Is output. Step S18: The pixel processing result accumulation unit 15 receives the output result of the pixel processing unit 14 and accumulates the output result. Step S19: It is checked whether or not the selection order number (SXN) j is the last number M. If there is an unprocessed pixel, that is, if it is not M, the process returns to step S13. When the processing of all the pixels of the selected pixel array k is completed, that is, when the selection order number is M, the process proceeds to step S20.

Step S20: The authenticity judging section 16 compares the pixel accumulation value, which is the accumulation result output from the pixel processing result accumulating section 15, with the authenticity judgment threshold value t of the corresponding pattern number of the authenticity judgment reference data. And the pixel integration value is equal to the authenticity determination threshold t.
If so, the process proceeds to step S21. On the other hand, if the pixel integration value is less than the authenticity determination threshold t, the process proceeds to step S22. Step S21: Execute a genuine note determination process. Step S22: Execute counterfeit note determination processing.

In the paper sheet authenticity discriminating apparatus 10 according to the present invention, as described above, the pixel array pattern selection unit 13 selects a pixel array including a different pixel group for each processing of a banknote.
The discrimination processing is performed based on the pixel information from the pixel group of the selected pixel array. Therefore, the discrimination process is not performed only based on the image information of the fixedly specified portion as in the related art, and the discrimination portion changes each time the banknote is handled, without impairing the speeding up of the determination process. This makes it difficult to detect the inspection part by trial and error from the outside, thereby eliminating the use of counterfeit bills and improving the security against counterfeit bills. In addition, when selecting a pixel array pattern, by selecting a pixel array number based on a random number table, the repetition of the pixel array pattern can be made irregular, thereby increasing the security against counterfeit bills. .

In the example described above, if the pixel value x _i read by the pixel processing unit 14 is equal to or _larger than the pixel reference value c _i in step S17, “1” is stored in the pixel processing result storage unit 15.
Although to output, instead of this, for each pixel, the upper threshold is provided in addition to the lower threshold, the pixel value x _i is the lower limit threshold value ≦
“1” can be output only when the condition of x _i ≦ upper threshold is satisfied, and “0” can be output when the condition is not satisfied.

Further, an example of a discrimination method different from the above will be described. The following determination method, and normalized out with "standard deviation sigma _i" pixel value "average value mu _i" of the pixel value the pixel value x _i of the pixel group of the pixel array pattern k selected, this normalization If the sum of the squares of the set values is equal to or smaller than a predetermined value, it is determined that the bill is genuine. In this determination method, as the pixel reference data stored in the pixel reference data storage section 18, a list of average value data μ _i of pixel values for each pixel specified by each pixel number i and a list specified by each pixel number i And a list of standard deviation value data σ _i of pixel values for each pixel. Truth determination reference data storage part 1
9 is a data list similar to that shown in FIG.

FIG. 9 is an explanatory diagram of the arithmetic processing performed by the pixel processing unit 14, the pixel processing result accumulating unit 15, and the authenticity determining unit 16. In step S17 shown in FIG. 8, the pixel processing unit 14 calculates the “average value μ _i ” and the “standard deviation value”, which are the statistics of each pixel i stored in the pixel reference data storage unit 18 of the data storage unit 12. σ _i ”and the pixel value x
_i is normalized {(x _i −μ _i ) / σ _i }, and the square value y _i shown in the equation (a) of FIG. 9 is obtained.

The pixel processing result accumulation unit 15 determines in step S1
In FIG. 8, as shown in FIG. 9B, the square of the normalized value of each pixel i of the selected pixel array list k is added and accumulated. In step S19, the authenticity judgment unit 16 compares the accumulation result value S obtained by the pixel processing result accumulation unit 15 with the threshold value t of the authenticity judgment reference data stored in the authenticity judgment reference data storage unit 19. . As a result of this comparison, as shown in FIG. 9 (c), if the accumulation result value S is equal to or less than the threshold value t, it is determined to be "true", and if the accumulation result value S exceeds the threshold value t, it is determined to be "false". Is determined.

As described above, the paper sheet authenticity discriminating apparatus 10 of the present invention is not limited to the above-described judging method, and various other judging methods can be applied. Further, it is possible to determine the type of a paper sheet based on the pixel information from the pixel group selected by the pixel array pattern selection unit 13, so that the present invention can be applied to a type determination apparatus. it can.

<Effect of the Invention> In the paper sheet recognition apparatus of the present invention, since the selected pixel to be inspected selected from the image data is different for each inspection of the paper sheet, the region to be inspected is fixed. It is not specified, and it is impossible to detect the inspection site from the outside, and the detection of the inspection site can be reliably prevented. Further, since the inspection site changes for each inspection, it is possible to reliably prevent a large number of counterfeit tickets having the same pattern from being used continuously. Therefore, according to the present invention, since information about all the pixels of the image data is not handled, the high-speed processing for the bill type or the authenticity is not hindered as in the past, and the inspection site is detected. Can be reliably prevented and the use of counterfeit tickets can be reliably eliminated, reducing the false accommodation rate of counterfeit tickets,
The reliability and convenience of the paper sheet recognition device can be improved.

[Brief description of the drawings]

FIG. 1 is a drawing of an example of a selected pixel array pattern showing a basic principle of the present invention.

FIG. 2 is an explanatory view schematically showing a general sheet recognition process.

FIG. 3 is a configuration diagram schematically showing a paper sheet authenticity discriminating apparatus according to the present invention.

FIG. 4 is an explanatory diagram showing an example of reading image data according to the present invention.

FIG. 5 is an explanatory diagram showing an example of a selected pixel list according to the present invention.

FIG. 6 is an explanatory diagram showing an example of pixel reference data according to the present invention.

FIG. 7 is an explanatory diagram showing an example of authenticity determination reference data according to the present invention.

FIG. 8 is a flowchart showing the operation of the paper sheet authenticity discriminating apparatus according to the present invention.

FIG. 9 is an explanatory diagram of an arithmetic expression showing another example of the authenticity determination method.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 10 Paper sheet authenticity discrimination apparatus 11 Paper sheet 13 (pixel arrangement pattern selection means) Pixel arrangement pattern selection part 17 (selected pixel list storage means) Selected pixel list storage portion

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G07D 7/20 G07D 7/02 G07D 7/10 G07D 7/12 G06T 7/00

Claims (4)

(57) [Claims] 1. A paper sheet recognition apparatus for determining the type or authenticity of a paper sheet by processing pixel information of a pixel group constituting image data of the paper sheet to be inspected, A selected pixel list storage unit that stores a selected pixel list that specifies a plurality of pixel array patterns each composed of a combination of a plurality of pixels selected in advance from the pixel group; And a pixel array pattern selecting means for selecting a pattern, wherein the sheet type or authenticity of the paper sheet is determined based on pixel information of a pixel group of the selected pixel array pattern. apparatus. 2. The apparatus according to claim 1, further comprising a determination data storage unit for storing data for determining the authenticity of the paper sheet, wherein the information of the pixel of the pixel array pattern selected by the pixel array pattern selection unit is included. A sheet recognition apparatus for determining whether the sheet is true or false based on the authenticity of the sheet. 3. The paper sheet recognition apparatus according to claim 1, wherein the selection of the pixel array pattern by the pixel array pattern selection unit is based on a random number table. 4. The method according to claim 3, wherein each pixel specified by the pixel array pattern is selected so as to be substantially evenly distributed in the image data for each pixel array pattern. Paper sheet recognition device.