JPH07311867A - Recognition device for paper sheet of the like - Google Patents

Abstract

PURPOSE:To improve the detection precision of skew or transverse deviation and to simplify a recognition device. CONSTITUTION:Positions of plural points on each side of paper sheet M are detected by an outline detection means 32 for paper sheets or the like. In a skew angle calculating means 50, an outline vector calculating means 51 obtains outline vectors in accordance with differences between positions of plural points, and an orthogonal component extracting means 52 obtains vectors which have 4-fold direction angles of these outline vectors as directions. An average vector calculating means 53 weights these 4-fold angle vectors to reduce the influences of damage, edge breaks, folding, etc., of paper M sheets or the like. A skew angle calculating means 54 obtains vectors in directions of 1/4 angles from 4-fold angle vectors by adopting the formula of the half angle twice, thus expressing the inclination of a rectangular figure like the paper sheet M or the like consisting of four orthogonal sides as one vector. It is taken as the average vector, and the inclination of this vector is calculated as the skew angle.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is incorporated in an automatic teller machine, an automatic teller machine, a money changer or the like of a financial institution, etc., and in particular, by detecting skew or lateral deviation of conveyed paper sheets, The present invention relates to a sheet recognition device that corrects the position of a sheet and performs a recognition process according to the position-corrected data of the sheet.

[0002]

2. Description of the Related Art Automatic teller machines, automatic teller machines, money changers and the like used in financial institutions are provided with a sheet recognition device for recognizing sheets, that is, banknotes. . With such a paper sheet recognition device, the type, authenticity, stain, damage, etc. of a bill are identified. In such a paper sheet recognition apparatus, there is a case where the paper sheet is skewed or laterally shifted when the paper sheet is conveyed, which may hinder the recognition of the paper sheet. Conventionally, in order to prevent this, for example,
An array of a plurality of photo interrupters is provided at the end portion of the paper sheet so that the position and angle of the edge side when the paper sheet is conveyed are detected. Then, the lateral deviation and the skew angle are obtained from the detected position and angle, and the cut-out position of the print pattern of the paper sheet is determined in accordance with this.

[0003]

However, the above-mentioned conventional paper sheet recognition device has the following problems. That is, in the above-mentioned device, there is no problem as long as the paper sheet is new, but if damage or corner folding occurs due to the distribution process etc., an error will occur in both skew detection and lateral deviation detection by the photo interrupter. Occurs. Then, if the allowable range of this error is narrowed, there is a problem that rejects in the recognition of paper sheets increase. On the other hand, if the allowable range of the error is widened, there is a problem that counterfeit bills and the like cannot be rejected. Further, in the above-described device, many sensors such as a photo interrupter for detecting skew and lateral deviation are required, and it is difficult to reduce the size of the device and it is difficult to reduce the cost. was there.

[0004]

In order to solve the above-mentioned problems, the paper sheet recognition apparatus of the present invention is characterized by the following points. (1) Conveying means for conveying paper sheets, illuminating means for irradiating either one or both of transmitted light and reflected light to the paper sheets conveyed by the conveying means, first conveying direction by the conveying means And a reading unit that optically reads the paper sheet image data in the direction of the second coordinate axis orthogonal to the coordinate axis of each of the above, and converts the image data into an electric signal at every predetermined sampling time. The paper sheet contour detection means detects the positions of a plurality of contour points on the respective sides of the paper sheet, which are separated from each other by a predetermined distance, by the reading means in the directions of the first and second coordinate axes.

A contour vector connecting two points of the contour points detected by the paper sheet contour detecting means is calculated, and a quadruple angle of the direction angle of the contour vector is calculated to obtain mutually orthogonal contours. A contour vector calculation means and a quadrature component extraction means for matching the directions of the vectors are provided. The contour vector whose direction is corrected by the orthogonal component extraction means is weighted according to the angular difference between the contour vector of interest and one of the contour vectors before and after the contour vector,
An average vector calculation means for averaging the contour vectors is provided. A skew angle calculating unit for calculating the skew angle of the paper sheet is provided by the contour vector averaged by the average vector calculating unit. A reference pattern prepared in advance corresponding to the position of the paper sheet detected by the paper sheet contour detection means and the skew angle calculated by the skew angle calculation means is selected, and the selected reference pattern and reading means are selected. It is provided with a discriminating means for discriminating the paper sheets by comparing and collating with the print pattern read from.

(2) In (1), the following features are provided. An affine transformation unit is provided for correcting the position of the print pattern read by the reading unit based on the contour detection unit and the skew angle calculation unit by rotating the position over the skew angle by affine transformation. The print pattern corrected by the affine transforming unit is provided with a discriminating unit that compares and collates with a reference pattern prepared in advance to discriminate the paper sheet.

(3) In (1) or (2), the following features are provided. An affine transformation means is provided for affine-transforming the coordinates of at least one of the corner positions of the read object read from the reading means based on the contour detection means and the skew angle calculation means over the skew angle. An affine inverse transforming unit is provided which obtains the coordinates of the position before affine transforming the data corresponding to the coordinate range specified by the position of the corner after the affine transforming by the affine inverse transform.

(4) In (2) or (3), the trigonometric function values used in the affine transformation and the affine inverse transformation are, for example, sin 2 ° = 0.035 and cos 2 ° = when the skew angle is 2 degrees. It is characterized in that a calculation value table stored beforehand such as 0.999 is provided.

[0009]

[Action]

(1) The position of a plurality of points on each side of the paper sheet is detected by the paper sheet contour detection means, and the contour vector is weighted by the average vector calculation means. As a result, it is possible to reduce the effects of damage to paper sheets, loss of edges, breakage, etc., calculate the quadruple angle of these contour vectors, and sum them to obtain paper sheets composed of four orthogonal sides. The inclination of the rectangular figure can be represented as one vector. This is used as an averaging vector, and the parallel component of the image data is effectively extracted by the inclination of this vector. Then, the authenticity of the paper sheet is determined by comparing and collating with a reference pattern prepared in advance according to the skew value.

(2) The skew angle is set to zero by affine transformation over the skew angle obtained in (1), and it is true or false by comparing and collating with the reference pattern when the skew angle is zero. To determine. As a result,
It is not necessary to prepare a reference pattern for each skew angle. (3) The coordinates of the corners of at least both ends or one end of the upper side of the paper sheet are affine-transformed over the skew angle obtained in (2), and the coordinates of the position after the rotation angle correction of the paper sheet are calculated. Ask.
Then, the position of the cut-out data that should come to that position is obtained by performing the affine inverse transformation over the skew angle. After that, the discriminating means of (1) or (2) cuts out the print pattern at the position required for discrimination, and compares it with the reference pattern to discriminate between true and false. As a result, it is possible to surely perform coordinate conversion of the minimum necessary data. (4) A sine value and a cosine value corresponding to the skew angle at the time of affine transformation or affine inverse transformation are prepared in advance, and high-speed transformation is performed. As a result, the discrimination processing time can be shortened.

[0011]

The present invention will be described in detail below with reference to the embodiments shown in the drawings. (First Embodiment) FIG. 1 is a block diagram showing a first embodiment of the present invention. First, the first embodiment will be described. The apparatus shown in the figure includes a conveying unit 1, an illuminating unit (not shown), a reading unit 3, an image data storage unit 20, a paper sheet contour detecting unit 32, a skew angle calculating unit 50, and reference data. It comprises a storage unit 80, a discrimination unit 70 and the like.

The transport means 1 comprises a transport path 11 and a plurality of drive rollers 12a and 12b. The conveyance path 11 has a width slightly wider than the width of the paper sheet M in the longitudinal direction. The paper sheets M are conveyed to the conveyance path one by one in the lateral direction with the paper surface horizontal. The drive rollers 12a and 12b are used for the transport path 1
A plurality of sheets are provided on the left and right sides of the sheet No. 1 at predetermined intervals in the transport direction so that the paper sheets can be reliably transported. The illuminating means is, for example, an infrared LED array, and a plurality of illuminating means are provided in the longitudinal direction of the paper sheet, and they are arranged below or above the paper sheet M. Reading means 3
For example, when a CCD image sensor is provided, the longitudinal direction of the paper sheet M, that is, the direction orthogonal to the paper sheet conveyance direction is reduced to a certain magnification using a condenser lens (not shown). The transmitted light or the reflected light from the illumination means is used to form an image on the image sensor, and the intensity of the light due to the density difference of the print pattern is output as the intensity of the electric signal.

The image data storage unit 20 is composed of a memory such as a RAM and stores the image data of the sheet M read by scanning the image sensor of the reading means 3. The paper sheet contour detection unit 32 mainly includes a long side (front) detection unit 33,
It is composed of a long side (rear) detection unit 34, a short side (right) detection unit 35, and a short side (left) detection unit 36, each of which searches for several contour points of a paper sheet to form a continuous contour vector. . The side position calculating means 40 is where the representative values (coordinates) of the long side and the short side are determined from the contour points. The skew angle calculation means 50 mainly includes a contour vector calculation means 51, a right angle component extraction means 52, an average vector calculation means 53, and a skew angle calculation means 54.

The reference data storage section 80 stores various reference data corresponding to various skew values. Then, the discriminating means 70 includes the skew angle calculating means 50 and the side position calculating means 4.
Based on the information obtained in 0, the required reference data is selected from the reference data storage unit 80, and this is compared and collated with the image data of the paper sheet to determine the authenticity of the paper sheet. Next, the operation of the first embodiment of the paper sheet recognition apparatus of the present invention will be described. First, the image data storage unit 20 monitors the output of the image sensor of the reading unit 3, the paper sheet M reaches the field of view of the reading unit 3, and the output of the image sensor becomes high (the light amount increases, the bill Each time the paper sheet M travels a predetermined distance, the image sensor is scanned, and the sensor output is sequentially stored in the memory. Eventually, when the sheet M has finished passing through the reading field, the image data of the entire sheet M is obtained in the memory of the image data storage unit 20.

Next, the paper sheet contour detecting means 32 detects the coordinates of the position (outer position) of the contour point of the paper sheet M based on the obtained image data. These are divided into four sides and detected by the long side (front), the long side (rear), the short side (right), and the short side (left), respectively. Next, each of these detection units will be described. FIG. 2A shows the image data of the paper sheet M stored in the image data storage unit. X ₂ axis is the main scanning direction (sensor array direction), X
_One axis is the direction opposite to the banknote transport direction. The point P indicates the position on the paper sheet M which reaches the sensor visual field region of the reading means 3 first. The point Q indicates the position on the paper sheet M that is the last to be separated from the sensor.

First, in the short side (left) detecting section 36, in order to prevent the point P from being influenced by the corner bending of the edge of the sheet when the left side is conveyed as shown in FIG. ,
X ₁ axially contour points from a predetermined distance D ₁ by the movement position X ₂ axially at appropriate intervals (R4,
Search (in order of R3, R2, R1). As a search method, for example, a method of differentiating in the X ₂ direction using a 3 × 3 first-order differential spatial operator and using the point at which the density value first sharply rises as the contour point can be considered. Similarly, in the long side (front) detector 33, D _{2 is} moved from R4 in the X ₂ axis direction.
Contour points from a distance in the X ₁ direction opposite to turn only at suitable intervals R5, R6, continue to search in the order of ... R12. The method in this case is determined by a value differentiated in the X ₁ direction using a 3 × 3 first-order differential space operator, similarly to the short side (left) detection unit 36 described above. Similarly, the short side (right) detection unit 35
With respect to the long side (rear) detection unit 34, in the case of left preceding as shown in FIG. Figure 2
(B) is a plot of the coordinates of the contour points thus obtained.

Next, the skew angle calculating means 50 will be described in detail. FIG. 3 shows a calculation example of skew value detection shown in FIG.
This is shown in the example of (b). First, as shown in FIG. 2B, the contour vector calculating means 51 sequentially joins the contour points by vectors such as R1 to R2 and R2 to R3. These are called contour vectors, and dx and dy in FIG.
Indicates the X ₂ direction component and the X ₁ direction component of each contour vector. As shown in FIG. 4, for example, R1 to R
The contour vector to 2 is dx = 2, dy = -6.

Next, in order to detect the quadrature component of the contour vector by the quadrature component extraction means 52, the vectors dx4 and dy4 in the quadruple angle direction of the contour vector are obtained by the following equation.
This is because the direction of the vector in each of the quadruple angles of the mutually orthogonal vectors coincides because the direction difference of 90 degrees becomes the direction difference of 360 degrees. The position of a general rectangular figure such as a class is detected. The following relationship is established between the contour vectors dx and dy and the vectors dx4 and dy4 in the quadruple angle direction. ^{dx4 = (dx 2 -dy 2)} 2 - (2dxdy) 2 ... (1) dy4 = 4dx · dy · (dx 2 -dy 2) ... (2)

Next, the average vector calculation means 53 weights the contour vectors and averages them. First, in FIG. 5, two unit contour vectors (dx (I) / r (I), dy
(I) / r (I)) and (dx (I-1) / r (I-1), dy (I-1) / r (I
The angle WANG (I) formed by -1)) is calculated by the following equation (3). WANG (I) = tan ^-1 [{DANG (I) ² / (1-DANG (I) / 4) ^1/2 } / 2] x 2 (3) Similarly, WANG (I + 1) is calculated. , Equation (4) gives the weight load
W (I) is obtained. W (I) = π / 2-WANG (I) -WANG (I + 1) (4) Here, W (I) may be negative, but in that case, it is set to 0.

The meaning of the equation (4) is that the weight load W (I) approaches 0 because the angle difference between the front and rear contour vectors becomes large with respect to the lack of edges and the damage of sides. 0, and if there is no angular difference between them, the weighting weight W (I) is π / 2. Therefore, even if a point other than the original contour point is searched for due to a folding defect of the paper sheet or the like, the weight load W (I) does not affect them.
Next, the weighted average vector A of the quadruple angle vector
The total sum of X and AY is calculated by the equations (5) and (6). The mean vector is dx4, dy4 a divided by R ^3, respectively. SX = Σw (i) · dx4 (i) / R (i) ³ (5) SY = Σw (i) · dy4 (i) / R (i) ³ (6)

Next, the skew angle calculating means 54 causes the SX, S
Return Y to the original vector direction using the half-width formula twice. In this case, if the direction of the original vector is within the range of 0 to 90 degrees, it returns to that range, but 90 degrees to 1
If it is in the range of 80 degrees, 180 degrees to 270 degrees, and 270 degrees to 360 degrees, the direction of the remainder angle divided by 90 is
It is in the range of 0 to 90 degrees. First, in the first half-width formula, X3 and Y3 are obtained as shown in equations (7) and (8). X3 = ± [{(SX ² + SY ² ) ^1/2 + SX} × (SX ² + SY ² ) ^1/2/2 ] ^1/2 (+ when SY> 0, − when SY <0) … (7) Y3 = [{(SX ² + SY ² ) ^1/2 -SX} × (SX ² + SY ² ) ^1/2/2 ] ¹ /2… (8)

Furthermore, in the second half-width formula application, RS,
RY is obtained from equations (9) and (10). RX = ± [{(X3 ² + Y3 ² ) ^1/2 + X3} × (X3 ² + Y3 ² ) ^1/2 / 2] ^1/2 (+ when Y3> 0, − when Y3 <0) ... (9) RY = [{ (X3 2 + Y3 2) 1/2 -X3} × (X3 2 + Y3 2) 1/2 / 2] 1/2 ... (10) the RX, RY is always positive And the skew angle is α, α = -tan ^-1 (RY / RX) (RX> RY) α = π / 2-tan ^-1 (RY / RX) (RY ≧ RX) ( 11) is required. When RX> RY, it indicates a left leading, and when RY> RX, it indicates a right leading. However, the range of α must be within 45 °.
By sequentially performing the above processing in the skew angle calculation means 50,
The skew angle and the leading direction of the paper sheet can be accurately detected.

In the example of FIG. 3, SX = 61.6586 and SY = 8
9.7929, RX = 105.743, RY = 26.1314, and α
= -Tan ^-1 (26.1314 / 105.743) =-0.242 (rad) =-1
It is 3.9 degrees, and it is skewed 13.9 degrees on the left. Next, the side position calculating means 40 will be described. here,
The representative coordinate value of the side is obtained, and the coordinates in which the weight load included in the element of each side is 1 or more are averaged and given. Explaining with the example in FIG. 3, the representative values of the short side (left) are x ₁ = (14 + 15) /2=14.5, y ₁ = (29 + 37) / 2 = 33, and the representative value of the long side (front) is The typical value is x ₂ = (38 + 47 + 60 + 69
_{) /4=53.5, y 2 = (41} + 39 + 35 + 34) /4=37.25,
The representative value of the short side (right) is x ₃ = (91 + 89) / 2 = 90, y ₃
= (8 + 1) /2=4.5.

Based on the information on the side position coordinates, the skew value calculated by the skew angle calculating means 50, and the information on the leading direction, the discriminating means 7 is described.
At 0, necessary reference data is selected from the reference data storage unit 80, and the final determination result is obtained by comparing and collating with this. As described above, it is possible to detect the skew angle without using the skew sensor while reducing the influence of the damage of the paper sheets, the edge loss, the break, and the like.

(Second Embodiment) Next, a second embodiment will be described with reference to FIG.
An example will be described. The second embodiment is the same as the first embodiment up to the skew angle calculation means 50, and thereafter, the image data is printed with the skew angle obtained by the skew angle calculation means 50 by the affine transformation means 60. Convert to a rectangular coordinate system with the long and short sides of the class as coordinate axes. This conversion formula is shown by the following formula. X = cos θx−sin θy (12) Y = sin θx + cos θy (13) x, y: input coordinates X, Y: output coordinates θ: rotation angle Data converted by equations (12) and (13) Finally, the discrimination result is obtained by comparing and collating with the data in the reference data storage unit 80. In this way, it is not necessary to prepare a reference pattern for each skew angle, and the capacity of the reference data storage unit 80 can be reduced.

(Third Embodiment) Next, a third embodiment will be described with reference to FIG. Similar to the first embodiment, the third embodiment is similar up to the skew angle calculating means 50, but the representative coordinate value of each side first obtained by the side position calculating means 40 is used to determine the paper sheet. The coordinates of at least one of the four corners are calculated. That is, first, the coordinates of the corners of these corners are transformed by the affine transformation means 60 using the equations (12) and (13), and then the new point is used as the reference of the coordinate frame range, and the image corresponding to each coordinate is obtained. The original coordinates of the data are obtained by the affine inverse transformation means 65 shown by the equations (14) and (15). x = Ysin θ + Xcos θ (14) y = Xcos θ−Xsin θ (15)

After that, the discrimination result is obtained by comparing and collating the image data existing at the coordinates obtained by the affine inverse transform with the reference data in the reference data storage unit 80 by the discriminating means 70. In this way, it is possible to surely perform coordinate conversion of the minimum image data required for true / false discrimination.

(Fourth Embodiment) In the fourth embodiment, the second,
In the third embodiment, the values of sin θ and cos θ used in the affine transformation means 60 and the affine inverse transformation means 65 are calculated in advance and stored in the calculated value table 90 of sin and cos. As a result, the calculation time can be shortened and the processing speed can be improved. It is needless to say that the present invention is not limited to the above-mentioned embodiments and various modifications can be made. For example, the conveyance of paper sheets is not limited to the case of carrying in the lateral direction, and the same applies to the case of carrying in the longitudinal direction.

[0029]

As described in detail above, according to the paper sheet recognition apparatus of the present invention, the quadruple angle of each contour vector of paper sheets is calculated, the directions of these contour vectors are aligned, and weighting is performed. The influence of corner folds and defects is removed to detect the skew angle and the amount of lateral deviation, and the printing pattern is corrected based on these to determine the authenticity of the paper sheet. It can be expressed as two vectors, and the processing can be simplified. In addition, since the contour vectors are weighted as described above, it is possible to perform reliable skew detection without being easily affected by the folding of paper sheets, the loss of edges, noise such as dust, and the distortion of sides. it can. Further, as for the sensor configuration, since all the processing can be performed by only one main image sensor, the recognition device itself can be designed compactly and the cost can be reduced.

[Brief description of drawings]

FIG. 1 is a block diagram of a first embodiment of a paper sheet recognition apparatus of the present invention.

FIG. 2 is an explanatory diagram of an operation of the paper sheet recognition apparatus of the present invention.

FIG. 3 is an explanatory diagram of a calculation example of skew value detection.

FIG. 4 is an explanatory diagram of a contour vector.

FIG. 5 is an explanatory diagram of weighting of a contour vector.

FIG. 6 is a block diagram of second and fourth embodiments of the paper sheet recognition apparatus of the present invention.

FIG. 7 is a block diagram of third and fourth embodiments of the paper sheet recognition apparatus of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Conveying means 3 Reading means 32 Paper sheet contour detecting means 52 Right angle component extracting means 53 Average vector calculating means 54 Skew angle calculating means 70 Discriminating means

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI technical display location 9061-5L G06F 15/70 460 F

Claims (4)

[Claims] 1. A transport means for transporting paper sheets, an illuminating means for irradiating either or both of transmitted light and reflected light to the paper sheets transported by the transport means, and the transport means. A reading unit that optically reads the paper sheet image data in the direction of the second coordinate axis orthogonal to the first coordinate axis in the transport direction at every predetermined sampling time and converts it into an electric signal, and the paper sheet by the reading unit. Paper sheet contour detecting means for detecting positions in the direction of the first and second coordinate axes of a plurality of contour points located at predetermined distances from each other on each side of the class, and the paper sheet contour A contour vector calculation unit that calculates a contour vector from each contour point detected by the detection unit, and a quadrangle of the direction angle of the contour vector calculated by the contour vector calculation unit are obtained to obtain mutually orthogonal contour vectors. The quadrature component extraction means that matches the direction of the torque and the contour vector whose direction has been corrected by the quadrature component extraction means are weighted according to the angular difference between the contour vector of interest and one of its contour vectors. An average vector calculation means for averaging the contour vectors, a skew angle calculation means for calculating a skew angle of the paper sheet by the contour vectors averaged by the average vector calculation means, A reference pattern prepared in advance corresponding to the position of the paper sheet detected by the class contour detection unit and the skew angle calculated by the skew angle calculation unit is selected, and the selected reference pattern and the reading unit are selected. A paper sheet recognizing device comprising: a discriminating means for discriminating the paper sheet by comparing and collating with the read print pattern. 2. An affine transformation unit for rotationally correcting the position of the print pattern read by the reading unit based on the contour detection unit and the skew angle calculation unit, and a print pattern rotationally corrected by the affine transformation unit in advance. 2. The paper sheet recognition apparatus according to claim 1, further comprising: a determination unit that compares the prepared reference pattern with a reference pattern to determine the paper sheet. 3. An affine transformation means for rotationally transforming the coordinates of at least one corner position of the read object read from the reading means based on the contour detection means and the skew angle calculation means over the skew angle. And affine inverse transforming means for obtaining the coordinates of the position before the rotational transformation of the data corresponding to the coordinate range specified by the position of the corner after the rotational transformation by the affine transforming means by the inverse rotational transformation. The paper sheet recognition apparatus according to claim 1 or 2, wherein the paper sheet recognition apparatus is provided. 4. The paper sheet according to claim 2, further comprising a calculation value table in which trigonometric function values used in rotation conversion by the affine transformation means and the affine inverse transformation means are stored in advance. Recognition device.