EP1844450B1 - Method for determining the exact centre of a coin introduced into a coin acceptor unit - Google Patents

Abstract

A method for determining an exact centre of a coin introduced into a coin acceptor unit, the method comprising: determining a probable centre of the coin by means of a sensor arrangement; recording an image of the coin to be examined; and developing an edge region of the coin around the probable centre and beyond a probable edge thereof using the image of the coin, the edge region of the coin being imaged as an at least approximately sinusoidal line upon displacement of the probable centre to the exact centre, and the exact centre being determined from analysis of the at least approximately sinusoidal line using an amplitude and phase angle of a chosen starting point.

Description

The invention relates to a method for determining the exact center of a coin entered into a coin validator according to the preamble of the main claim.

From the WO 2004/075124 A1 For example, a method of detecting coins is known in which the image of a coin is picked up by an image sensor. To determine the size of the receiving area, ie the coverage area of the image sensor by the coin, the diameter of the coin is determined by scanning the vertex by means of a column of the image sensor. In the known art, in addition to the determination of the timing, the velocity is calculated by scanning the leading edge on a line in the center of the coin. At the coverage area the picture of the coin is taken, wherein in a further processing step the embossing or pattern recognition is performed. This pattern recognition is based on the analysis of a transformed image of the coin, in which the circular area or approximate circular area of the coin is unwound over 360 ° around the center. It is important that the exact center is known, as an imprecise definition of the center directly goes into the reproducibility of said evaluation process. It has been found that the detection of the center according to the cited prior art is not sufficiently accurate.

The invention is therefore an object of the invention to provide a method for determining the exact center of an entered into a coin validator coin, which determines based on the detection of the probable center according to the prior art, the exact center without particularly large evaluation effort.

This object is achieved by the characterizing features of the main claim in conjunction with the features of the preamble.

The measures specified in the dependent claims advantageous refinements and improvements are possible.

By making the edge region of the coin wrap around the probable center around and beyond its probable edge, a map of the edge as an approximate sinusoidal line results from the analysis of the probable center to the exact center at least Approximated sinusoidal line using the amplitude and the phase angle of a selected starting point, the exact center can be determined. It can therefore be made with a low computational effort by a reduced transformation of the edge region over the probable edge away the precise determination of the center.

Advantageously, the magnitude of the deviation between the likely center and the exact center is determined from the difference of the largest and smallest amplitude, the phase angle of the offset is determined as the average of the phase angle of the largest amplitude and the phase angle of the opposite direction to the smallest amplitude. The exact radius of the coin is determined as the sum of the inner radius of the edge area and the mean of the smallest and largest amplitude.

The size of the deviation or the offset and the direction of the offset can also be determined in other ways, where appropriate, the quality of the sinusoidal line has an influence.

For example, the phase angle can be determined by determining the location of the largest or the smallest amplitude, the magnitude of the deviation can be calculated from the difference of the largest amplitude with given phase angle and the amplitude at a phase angle of - 90 ° and + 90 ° to the determined phase angle of Maximums or of the minimum. The exact radius of the coin can be determined as the sum of the inner radius of the edge region and the amplitude at a phase angle of -90 ° and + 90 °, respectively, to the phase angle of the location of the maximum amplitude.

It is possible to use the center only with part of the image, e.g. only with the upper half of the coin shown to save time for the image data transfers. Here then the transformed processing is only half of the sine function.

Advantageously, the inventive method can also be applied to coins with corners or waves, wherein the sinusoidal line of the edge of the coin superimposed on a curve whose period for the number of corners or waves and their amplitude for the depth of the corners or waves ,

In this case, the size of the deviation, the phase angle and the number of waves or corners of a coin can be determined, for example, by a Fourier transformation of the transformed edge line.

In principle, the inventive idea is that a substantially sinusoidal curve is generated by means of a polar transformation from an edge line of the coin, the analysis of which in a fast and accurate manner determines the center of the coin. For the analysis, other well-known methods for determining the parameters of the edge line can be used in addition to the methods already mentioned.

In addition to the exact form of calculation, approximations can be used which provide sufficient results for the boundary conditions given by the respective application, such as reproducibility in the series product and the like.

Fig. 1

die schematische Darstellung einer Münze mit den Koordinaten einer Polartransformation um einen genauen Mittelpunkt A und eines wahrscheinlichen Mittelpunktes B, der relativ zum genauen Mittelpunkt nach rechts unten verschoben ist,

Fig. 2

die Polartransformation des Randbereichs der Münze

1. a) um den genauen Mittelpunkt und
2. b) um den verschobenen wahrscheinlichen Mittelpunkt,

Fig. 3

eine Münze mit einer gewellten Randkontur,

Fig. 4

die Abwicklung des Randbereichs der Münze bei einem genauen Mittelpunkt und bei einem verschobenen wahrscheinlichen Mittelpunkt,

Fig. 5

Darstellung von Kreisen mit versetzten Mittelpunkten zur Erläuterung der Form einer transformierten Randlinie.

The inventive method is based on the drawing explained in more detail in the following description. Show it

Fig. 1

1 is a schematic representation of a coin having the coordinates of a polar transformation about an exact center A and a probable center B shifted down to the right relative to the exact center;

Fig. 2

the polar transformation of the edge area of the coin

1. a) around the exact center and
2. b) the displaced probable center,

Fig. 3

a coin with a wavy edge contour,

Fig. 4

the settlement of the edge area of the coin at an exact center point and a shifted probable center point,

Fig. 5

Representation of circles with offset centers to explain the shape of a transformed boundary line.

As with the prior art of WO 2004/075124 has been described, the center point or the diameter can be determined by the method disclosed therein. However, the diameter or the center can also be found by light barriers and sensors. However, this provision is for pattern recognition by means of image acquisition and processing of the circular area over 360 ° around the center, ie for a polar transformation not exactly enough. Therefore, this center is called the probable center. The exact center is the center that is actually present when the coin is picked up by an image sensor.

In Fig. 1 is a coin 1 with exact center A and a shifted by about 45 ° to the bottom right by δR center B, these centers are also the centers for a polar transformation. For the method according to the invention, starting from the probable center B and corresponding to the diameter, a polar transformation, ie a development of the edge region of the coin, is undertaken, in which the polar coordinates are converted into Cartesian coordinates.

In Fig. 2 On the one hand, the development of the edge region of the coin 1 with the exact center A and on the other hand about the probable center B displaced by ΔR is shown. In this case, in order to accelerate the transformation, the processing is carried out in such a way that an edge region is considered which is limited to the outside by a boundary line 3 having a radius R ₂ and inwardly by a boundary line 2 having the radius R ₁ . With R _M is the exact radius of the coin 1 and R _{O is} the probable radius, which is determined when the coin enters the measuring range. The width of the edge area must be selected such that the coin with the center shifted by δR is still encompassed by this edge area. Therefore applies to the radii of the edge area $${\mathrm{R}}_{}$$$${\mathrm{}}_{\mathrm{1}}\mathrm{<}{\mathrm{R}}_{\mathrm{O}}\mathrm{-}{\mathrm{.DELTA.R}}_{\mathrm{Max}}{\mathrm{and\; <figure-callout\; id="2"\; label="R"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">R</figure-callout>}}_{\mathrm{2}}\mathrm{>}{\mathrm{R}}_{\mathrm{O}}\mathrm{+}{\mathrm{.DELTA.R}}_{\mathrm{Max}}\mathrm{.}$$

Here ΔR _{max is} the sum of the maximum possible errors in determining the probable center δR _max and in determining the probable radius of the coin dR _max : $${\mathrm{.DELTA.R}}_{\mathrm{Max}}\mathrm{=}{\mathrm{.delta..sub.R}}_{\mathrm{Max}}\mathrm{+}{\mathrm{dR}}_{\mathrm{Max}}$$

With a match of the probable and the exact center, the edge of the coin 1 in the transformed representation gives a straight line 4 accordingly Fig. 2a ). In a settlement with a probable, ie shifted center of the polar transformation, an at least approximately sinusoidal line 5 results for the edge (FIG. Fig. 2b ), hereinafter referred to as a sinusoidal line.

Using the Fig. 5 the shape of the transformed boundary line (sinusoidal line 5) will be explained in more detail.

In the picture Fig. 5 O _{1 is} the center of the polar transformation and O ₀ is the center of the coin with a radius r ₀ . The offset between the transformation center O ₁ and the center O _{0 of} the coin is d.
An edge point of the coin P has the distance r ₀ to the center of the coin and r ₁ to the center of the transformation, whose phase angles are marked accordingly as φ ₀ and φ ₁ .

An equation for the distance to the center of the transformation r ₁ (ordinate in the transformed coordinate system of the diagrams Fig. 2 and Fig. 4 ) can be written as follows: $${\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="imgf0005.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1}}\mathrm{=}\sqrt{{\mathrm{<figure-callout\; id="0"\; label="r"\; filenames="imgf0002.png,imgf0005.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{0}}^{\mathrm{2}}\mathrm{+}{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">d</figure-callout>}}^{\mathrm{2}}\mathrm{+}\mathrm{2}{\mathrm{<figure-callout\; id="0"\; label="dr"\; filenames="imgf0002.png,imgf0005.png"\; state="\{\{state\}\}">dr</figure-callout>}}_{\mathrm{0}}{\mathrm{<figure-callout\; id="0"\; label="cos"\; filenames="imgf0002.png,imgf0005.png"\; state="\{\{state\}\}">cos</figure-callout>}}_{\mathrm{0}}}$$

dann ist auch der Unterschied zwischen den Phasenwinkeln gering: $${\phi }_1\approx {\phi }_0$$

Assuming that the offset between the transformation center O ₁ and the center of the coin O _{0 is} much smaller than the radius of the coin r ₀ , $$d<<{\mathrm{<figure-callout\; id="0"\; label="r"\; filenames="imgf0002.png,imgf0005.png"\; state="\{\{state\}\}">r</figure-callout>}}_0$$

then the difference between the phase angles is small as well: $${\phi }_1\approx {\mathrm{<figure-callout\; id="0"\; label="\phi "\; filenames="imgf0002.png,imgf0005.png"\; state="\{\{state\}\}">\phi </figure-callout>}}_0$$

oder, wenn wir nur Schwankungen des Abstandes betrachtet werden $$r_1-r_0\approx {d\mathrm{cos\phi }}_1$$

The above equation can be transformed as follows: $${\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="imgf0005.png"\; state="\{\{state\}\}">r</figure-callout>}}_1\approx {\mathrm{<figure-callout\; id="0"\; label="r"\; filenames="imgf0002.png,imgf0005.png"\; state="\{\{state\}\}">r</figure-callout>}}_0\left(1+\frac{\mathrm{<figure-callout\; id="0"\; label="d\; r"\; filenames="imgf0002.png,imgf0005.png"\; state="\{\{state\}\}">d\; </figure-callout>}}{r_{}}{\mathrm{<figure-callout\; id="1"\; label="cos"\; filenames="imgf0005.png"\; state="\{\{state\}\}">cos</figure-callout>}}_1\right)$$

or, if we are only considered variations of the distance $$r_1-{\mathrm{<figure-callout\; id="0"\; label="r"\; filenames="imgf0002.png,imgf0005.png"\; state="\{\{state\}\}">r</figure-callout>}}_0\approx {d\mathrm{<figure-callout\; id="1"\; label="cos"\; filenames="imgf0005.png"\; state="\{\{state\}\}">cos</figure-callout>}}_1$$

Thus, it is clear that, at least with a small offset between the center of the coin and the center of the transform compared to the radius of the coin, variations in the boundary line in the transformed map by a value r _{0 are} actually described with a sine function (or cosine function) can be.

However, the greater the offset between the centers, the more the boundary curve deviates from a sine function. From the picture Fig. 5 For example, it is clear that the angular range φ _pos , where r ₁ > r ₀ (part of the edge line from D to C counterclockwise), is smaller than the angular range φ _neg , where r ₁ <r ₀ , and the larger the offset d is, the larger the difference between the angular ranges becomes, and it is called an "approximately sinusoidal line".

After the clockwise starting and starting with a start phase angle, which is Φ = 0 °, with reference to Fig. 2b for the sinusoidal line, the amplitude is examined, in such a way that the maximum or the minimum of the amplitude and the associated phase angle are found. This is done by comparing the difference of the coordinate values on the sinusoidal line 5 at predetermined intervals starting from the starting angle.

$$\mathrm{\delta \Phi }\mathrm{=}\frac{{\mathrm{\Phi }}_{\max }\mathrm{+}{\mathrm{\Phi ʹ}}_{\min }}{\mathrm{2}}$$

$${\mathrm{R}}_{\mathrm{M}}\mathrm{=}{\mathrm{R}}_{\mathrm{1}}\mathrm{+}\frac{{\mathrm{A}}_{\max }\mathrm{+}{\mathrm{A}}_{\min }}{\mathrm{2}}\mathrm{,}$$

wobei Φ'_min der Phasenwinkel in Gegenrichtung zum Minimum ist. In Bezug auf die Fig. 2b lässt sich Φ'_min beispielsweise wie folgt berechnen Φ'_min = Φ_min + π, wobei der Winkel im Radian berechnet wird. R₁ ist hier der innere Radius des Randbereichs (s. Fign. 1 und 2).From the ordinate of the maximum A _max , the ordinate of the minimum A _min and their phase angle, both the exact radius R _{M of} the coin and the magnitude of the deviation or the offset of the probable center to the exact center δR and the phase angle δΦ of the Calculate offset to bring the center of the transformation into the exact center of the coin. $$\mathrm{.delta..sub.R}\mathrm{=}\frac{{\mathrm{A}}_{\mathrm{Max}}\mathrm{-}{\mathrm{<figure-callout\; id="2"\; label="A\; min"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">A\; </figure-callout>}}_{\min }}{\mathrm{2}}$$

$$\mathrm{\delta \Phi }\mathrm{=}\frac{{\mathrm{\Phi }}_{\mathrm{Max}}\mathrm{+}{\mathrm{\Phi \text{'}}}_{\min }}{\mathrm{2}}$$

$${\mathrm{R}}_{\mathrm{M}}\mathrm{=}{\mathrm{<figure-callout\; id="1"\; label="R"\; filenames="imgf0005.png"\; state="\{\{state\}\}">R</figure-callout>}}_{\mathrm{1}}\mathrm{+}\frac{{\mathrm{A}}_{\mathrm{Max}}\mathrm{+}{\mathrm{A}}_{\mathrm{<figure-callout\; id="2"\; label="min"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">min</figure-callout>}}}{\mathrm{2}}\mathrm{.}$$

where Φ ' _{min is} the phase angle in the opposite direction to the minimum. Regarding the Fig. 2b can be Φ _'min, for example, be calculated as follows Φ' Φ _min = _min + π, where the angle is calculated in radians. R ₁ here is the inner radius of the edge region (s. FIGS. 1 and 2 ).

The inventive method is also applicable to coins that have a non-circular contour, but are provided with corners or waves. Such a coin is for example in Fig. 3 shown.

Fig. 4 again shows the settlement of the edge area of the coin Fig. 3 , in which Fig. 4a the settlement around the exact center, ie the settlement at the correspondence of the centers of the transformation and the center of the coin shows, while Fig. 4b shows a settlement around the probable, shifted center. As in Fig. 4 1, the edge curve has repetitive maxima and minima, its period P standing for the number of waves and the amplitude between maxima and minima for the depth T of the contour.

In Fig. 4b is the boundary curve accordingly Fig. 4a superimposed on a sinusoidal line using an average filter to determine the offset of the probable center to the exact center, which can be used to calculate a balanced smooth curve. The maximum and minimum and the phase angles are determined in an analogous manner as described above for determining the magnitude and direction of the deviation of the probable center from the exact center.

For the determination of the number of waves, their depth T and their period P, the current boundary line ( Fig. 4b ) or the number of transitions of the current Edge line to be used by the balanced line.

1. a) Originalabbildung einer Münze wird einer Polartransformation unterzogen, dabei werden Vorkenntnisse über vermutliche Radius und Mittelpunkt der Münze benutzt, um den Transformationsbereich abzugrenzen. Praktisch ist es möglich, die relevanten Bereiche um die Randlinie der Münze auf einem Originalbild von der Größe 400x600 Pixel zu einem transformierten Bild von der Größe etwa 40x60 Pixel umzuwandeln (also Verkleinerung der Datenmenge um Faktor 100), das trotzdem die ganze relevante Information behält.
2. b) Im transformierten Bild wird die Randlinie gesucht, dafür wird z.B. in jeder Spalte von oben nach unten die Position des ersten Maximums, das eine vorgegebene Schwelle (Hintergrund) überschreitet, registriert.
3. c) Randlinie wird gesäubert (Ausreißer gelöscht) und ausgeglichen. Um im Falle von "eckigen" Münzen (Fign. 3 + 4) Wellen und Ecken des Randes zu registrieren, werden zwei unterschiedlich abgeglichene Kopien der Randlinie erstellt. Beispielweise kann eine abgeglichene Randkurve durch ein eindimensionales Mittelwert-Filter von der Größe 3 Pixel, und eine weitere Kopie der Randkurve durch ein wesentlich größeres Mittelwert-Filter (z.B. von der Größe 15 Pixel) erstellt werden. Dann kann durch den Vergleich der leicht und stark ausgeglichenen Randlinien die Information über Anzahl und Form der Ecken gewonnen werden (siehe Abbildungen Fign. 3 + 4).
4. d) Maximum und Minimum der abgeglichenen Randlinie werden berechnet und dann die gesuchten Parameter (Mittelpunkt und Radius) gefunden.

As an example, procedural steps are to be listed below in order to calculate the exact center and exact radius of an entered coin:

1. a) Original image of a coin is subjected to a polar transformation, using previous knowledge of probable radius and center of the coin to delineate the transformation region. In practice, it is possible to convert the relevant areas around the edge line of the coin on an original image of size 400x600 pixels to a transformed image of size about 40x60 pixels (ie reduction of the data amount by a factor of 100), which nevertheless retains all the relevant information.
2. b) In the transformed image, the boundary line is searched for, for example, in each column from top to bottom, the position of the first maximum, which exceeds a predetermined threshold (background) registered.
3. c) Edge line is cleaned (outliers cleared) and compensated. In the case of "square" coins ( FIGS. 3 + 4) To register waves and corners of the border, two differently matched copies of the border line are created. For example, a matched edge curve can be created by a one-dimensional 3-pixel average filter, and another copy of the edge curve by a much larger average filter (eg, 15 pixels). Then, by comparing the easy and strong balanced Edge lines the information about number and shape of the corners are obtained (see pictures FIGS. 3 + 4).
4. d) Maximum and minimum of the adjusted edge line are calculated and then the searched parameters (center point and radius) are found.

Points b) and c) can be realized with different known methods, therefore, it will not be discussed in more detail.

Claims (7)

1. Method for determining the exact centre of a coin introduced into a coin acceptor unit, in which the probable centre is determined by means of a sensor arrangement and the image of the coin to be examined is recorded,
   characterised in that
   using the image of the coin, a development of the edge region of the coin is undertaken around the probable centre and beyond the probable edge thereof, the edge of the coin being imaged as an at least approximately sinusoidal line upon displacement of the probable centre to the exact centre, and in that the exact centre is determined from the analysis of the at least approximately sinusoidal line using the amplitude and the phase angle of a chosen starting point.
2. Method according to claim 1, characterised in that the size of the deviation between probable centre and exact centre is determined from half the difference of the greatest and smallest amplitude of the at least approximately sinusoidal line.
3. Method according to claim 1 or claim 2, characterised in that the direction of the deviation between probable centre and exact centre is determined from the average of the phase angles indicating the location of the greatest amplitude and of the location of the smallest amplitude, offset by a half period, of the at least approximately sinusoidal line.
4. Method according to one of the claims 1 to 3, characterised in that, in the case of an edge contour of a coin provided with corners or undulations, the at least approximately sinusoidal line is overlapped by a curve representing the corners or undulations and an average value filter being used to determine the at least approximately sinusoidal line.
5. Method according to claim 4, characterised in that the number of corners or undulations is determined by the number of maxima and/or minima of the entire line relative to the filtered approximately sinusoidal line.
6. Method according to claim 1, characterised in that the greatest amplitude of the sinusoidal line relative to a boundary line of the development region is determined, and in that the exact centre is determined using the difference of the value of the greatest amplitude and of the value of the amplitude of the sinusoidal line relative to the boundary line at the phase angle -90°.
7. Method according to one of the claims 1 to 6, characterised in that the exact radius of the coin is determined from the sum of the radius of the inner boundary line of the edge region and half the sum of the greatest and smallest amplitude of the at least approximate sinusoidal line.

Images