EP3210195B1 - Device and method for testing valuable documents and system for handling of valuable documents - Google Patents

Description

The invention relates to a device and a method for checking valuable documents, in particular banknotes, as well as a valuable document processing system.

In banknote processing systems, properties of banknotes, such as printed image, denomination, authenticity and condition, are determined by recording the physical properties of the banknotes using sensors and evaluating the sensor data generated in the process.

When checking banknotes, their reflectance and/or transmission properties are often used. For this purpose, one banknote is irradiated with the light of one or more light sources and the light remitted by the banknote, i.e. diffusely reflected or transmitted, is detected using one or more sensors. Depending on the type of light sources used, the reflectance or transmission curves determined in this way can deviate from the actual reflectance or transmission behavior of the banknote. For example, when using light-emitting diodes (LEDs) as light sources, artifacts can occur in certain areas of the reflectance or transmission curves that do not correspond to the actual properties of the banknote.

It is the object of the present invention to provide a device, a method and a valuable document processing system which allows the reflection and/or transmission properties of valuable documents to be determined as precisely as possible.

This task is solved by the device, the method and the value document processing system according to the independent claims. The device according to the invention for checking documents of value, in particular banknotes, according to claim 1 has: at least two radiation sources for emitting electromagnetic radiation, with which a document of value is irradiated; at least one sensor for detecting the electromagnetic radiation reflected and/or transmitted directed or diffusely by the document of value and generating corresponding sensor signals, with components assigned to the radiation sources, the sensor signals being corresponding reflection and/or transmission signals; an evaluation device which is designed to derive corrected sensor signals from the sensor signals generated by the at least one sensor, taking into account at least one spectral property of the electromagnetic radiation of the at least two radiation sources, wherein in deriving the corrected sensor signals at least one linear combination of components assigned to the different radiation sources the sensor signals are formed.

The method according to the invention for checking documents of value, in particular banknotes, according to claim 16 has the following steps: irradiation of a document of value with electromagnetic radiation from at least two radiation sources; Detecting the electromagnetic radiation reflected and/or transmitted in a directed or diffuse manner by the document of value and generating corresponding sensor signals, with components assigned to the radiation sources, the sensor signals being corresponding reflection and/or transmission signals; Derivation of corrected sensor signals from the sensor signals generated by the at least one sensor, taking into account at least one spectral property of the electromagnetic radiation of the at least two radiation sources, wherein in the derivation of the corrected sensor signals at least one linear combination is formed from components of the sensor signals assigned to the different radiation sources.

The valuable document processing system according to the invention has at least one device for processing, in particular for transporting and/or counting and/or sorting, valuable documents, in particular banknotes, and is characterized by the device according to the invention for checking valuable documents.

The invention is based on the idea of subjecting the reflection or transmission signals generated by sensors when detecting the light reflected and/or transmitted by the document of value, which together preferably represent a spectral reflection and/or transmission signal curve, to a correction in which corrected Reflection or transmission signals, which together preferably represent a corrected spectral reflection or transmission signal curve, are obtained. When correcting the reflection or transmission signals, at least one spectral property of the electromagnetic radiation emitted by the at least two radiation sources is used.

The spectral property of the electromagnetic radiation taken into account can refer to any property, in particular to the intensity, of the electromagnetic radiation emitted by the radiation source at one or more wavelengths or in one or more wavelength ranges. For example, the spectral property of the electromagnetic radiation taken into account relates to a value for the radiation intensity of a radiation source in the range of a first wavelength of a main emission and to a corresponding value in the range of a second wavelength of a further emission, which is also a secondary emission referred to as. In another example, the spectral property of the electromagnetic radiation taken into account relates to a value for the radiation intensity in the range of a first wavelength of a main emission as well as to several corresponding values in ranges of further wavelengths of further emissions, which are also referred to as secondary emissions.

Alternatively or additionally, the spectral property can also relate to a wavelength-dependent intensity curve of the electromagnetic radiation emitted by the radiation sources in a broader wavelength range, in which in particular the main emission and the secondary emission or the secondary emissions are included.

When correcting the sensor signals according to the invention, the spectral property of the electromagnetic radiation can also be taken into account in the form of parameters that are derived from the above-mentioned properties, in particular the intensity values at certain wavelengths or in certain wavelength ranges, such as quotients, differences or sums the intensity values mentioned.

In a preferred embodiment, when the document of value to be checked is illuminated sequentially by the at least two light sources with main emissions in different wavelength ranges, the spectral property of at least one light source can be taken into account when evaluating the sensor signals.

This ensures in a simple and reliable manner that the corrected reflection or transmission signals correspond significantly better to the actual reflection or transmission behavior of the document of value than without correction. In particular, this makes it possible Influence due to the type of light sources used, such as LEDs, is eliminated or at least reduced. Overall, the invention allows a much more precise determination of the reflection and/or transmission properties of documents of value.

Preferably, the at least one spectral property of the electromagnetic radiation from the at least two radiation sources is given by at least one spectral distribution of the electromagnetic radiation from the at least two radiation sources. The spectral distribution of a kth ( k = 1 ... n ) radiation source can preferably be specified in the form of a continuous intensity profile S _k ( λ ) dependent on the wavelength λ . Alternatively, the spectral distribution of the kth radiation source can also be given by a large number of intensity values S _ki at discrete wavelengths λ _i ( i = 1 ... m ). In a preferred embodiment, the spectral distributions of the n radiation sources differ from one another. Furthermore, the variant m=n is particularly preferred, since the correction of the reflection or transmission signals can then be determined particularly easily. By taking the spectral distribution of the radiation sources into account, the corrected sensor signals match the actual reflection or Transmission course of the value document corresponds.

It is further preferred that at least one spectral distribution of the electromagnetic radiation of the radiation sources is given by a first spectral distribution of the electromagnetic radiation emitted by the radiation sources and a second spectral distribution, which is different from the first spectral distribution. Preferably, a first spectral distribution of the electromagnetic radiation emitted by the radiation source corresponds to a spectral distribution with a main emission and at least one secondary emission. A second spectral distribution preferably corresponds to the first spectral distribution, but without having at least one secondary emission. The first spectral distribution is preferably determined by measuring, for example using a spectrometer, the radiation source or using associated data sheets. The second spectral distribution can then be derived from the first spectral distribution by eliminating the spurious emission. Using the first and/or second spectral distribution, a particularly reliable and precise correction of the sensor signals can be achieved, particularly with regard to disruptive influences due to secondary emissions.

According to a further preferred embodiment of the invention, the corrected sensor signals are calculated by multiplying the generated sensor signals R with a correction matrix B. The vector R is formed by the intensity values measured at one location with the n radiation sources. Each signal component R _k of the vector R therefore corresponds to the intensity when measured with the respective radiation source k = 1...n.

The correction matrix B is made from the at least one discrete spectral distribution S _ki , with k = 1 ... n and i = 1 ... m , of the electromagnetic radiation from a first number n of radiation sources at a second number m of discrete wavelengths λ _i and at least one spectral profile D _i of the sensitivity of the at least one sensor for electromagnetic radiation. The correction matrix B has at least one non-diagonal element that is different from 0. This means that when calculating the corrected sensor signals BR, at least one linear combination is always formed from components of the sensor signals Rk assigned to the different radiation sources. This linear combination is non-trivial, i.e. a total of at least two coefficients are not equal to 0, so that one The sum or difference is formed from at least two different components of the sensor signals Rk (with k = 1 .. n ).

The correction matrix B preferably corresponds to the product B = A ₀ A ⁺ from a second matrix A ₀ and a pseudoinverse A ⁺ a first matrix A describing the sensor system, the matrix elements A _ki of which are determined by the product of the first spectral distribution S _ki of the n radiation sources at m discrete wavelengths λ _i emitted electromagnetic radiation with the spectral profile D _i the sensitivity of the at least one sensor and a wavelength distance value Δλ between two discrete wavelengths λ _i are given: A _ki = S _ki D _i Δλ . The matrix elements A _0ki of the second matrix A ₀ preferably correspond to the product of the second spectral distribution S ' _ki of the electromagnetic radiation from which the at least one secondary emission was eliminated, with the spectral profile D _i of the sensitivity of the at least one sensor and the wavelength distance value Δλ between two discrete wavelengths λ _i : A _0ki = S ' _ki D _i Δλ. By multiplying the sensor signals R with the correction matrix B, the influence of the spectral behavior of the radiation sources, in particular secondary emissions, on the sensor signals can be corrected with particularly high accuracy.

Alternatively or in addition to the preferred embodiments described above, the at least one spectral property of the electromagnetic radiation of the radiation sources is given by at least one parameter which has one or more spectral components, in particular the intensity, of the electromagnetic radiation of the radiation source, in particular at one or more wavelengths or Wavelength ranges, characterized. Preferably, the sensitivity of the respective sensor, in particular at the wavelengths or wavelength ranges mentioned, can also be taken into account in the parameter. The parameter corresponds then preferably a product of the intensity of the radiation emitted by a radiation source at a specific wavelength and the sensitivity of the respective sensor at this wavelength. Alternatively or additionally, the at least one parameter can also be derived from two or more intensity values and possibly sensor sensitivity values at different wavelengths, for example by forming a quotient. By using one or more such parameters, the relevant spectral properties of the radiation sources can be easily taken into account when correcting the sensor signals, so that relatively low computing capacities are sufficient even for spectral reflection or transmission curves in a wide spectral range, for example between 400 and 1100 nm to enable correction of the sensor signals in real time.

wobei der Korrekturparameter a direkt durch Messung der spektralen Verteilung des von den Strahlungsquellen abgegebenen Lichts und der Detektorempfindlichkeit erhalten werden kann. Alternativ kann dieser auch aus dem gemessenen Sensorsignal R und den mittels Spektrometermessung an einem Kalibrierdokument erhaltenen Werten r₁ und r₂ gemäß $$a=-\frac{r_1-R}{r_2-R}$$

berechnet werden.According to the invention, at least a first parameter a ₁ characterizes the spectral portion of a main emission of the electromagnetic radiation of the radiation source and at least a second parameter a ₂ characterizes the spectral portion of an emission occurring in addition to the main emission, a so-called secondary emission, of the electromagnetic radiation of the radiation source. The evaluation device is designed in such a way that the corrected sensor signals are derived from the sensor signals, taking into account the first and second parameters a ₁ or a ₂ or a parameter a derived from the first and second parameters a ₁ or a ₂ , which in particular corresponds to the quotient a ₁ / a ₂ corresponds to the first and second parameters a ₁ and a ₂ , respectively. Preferably, the corrected (r ₁ ) and preferably standardized to a white reference (w ₁ ) sensor signals r ₁ / w ₁ are calculated from the measured sensor signals R, the preferably standardized value r ₂ / w ₂ of those actually remitted by the document of value in the area of secondary emission or transmitted radiation and the correction parameter a = a ₁ / a ₂ using the equation $$\frac{r_1}{w_1}=\left(1+a\right)R-{\mathit{<figure-callout\; id="2"\; label="ar"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">ar</figure-callout>}}_2=\left(1+a\right)R-a\frac{r_2}{{\mathrm{<figure-callout\; id="2"\; label="w"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">w</figure-callout>}}_2},$$

where the correction parameter a can be obtained directly by measuring the spectral distribution of the light emitted by the radiation sources and the detector sensitivity. Alternatively, this can also be based on the measured sensor signal R and the values r ₁ and r ₂ obtained using a spectrometer measurement on a calibration document $$a=-\frac{r_1-\mathrm{<figure-callout\; id="2"\; label="R\; r"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">R</figure-callout>}}{r_{}}$$

be calculated.

In a particularly preferred embodiment already mentioned, the corrected sensor signals are normalized based on corrected reference signals, the corrected reference signals being generated from reference signals that are generated by the at least one sensor when detecting the electromagnetic radiation emitted by a reference document, a so-called white reference, are derived taking into account the at least one spectral property of the electromagnetic radiation of the at least two radiation sources. When deriving the corrected reference signals from the reference signals, the sensitivity of the at least one sensor for electromagnetic radiation, in particular in the form of at least one spectral curve of the sensitivity, is preferably taken into account. The corrected reference signals used in the normalization of the corrected sensor signals are preferably corrected analogously to the sensor signals. The above statements and stated advantages in connection with the correction of the sensor signals apply accordingly to a corresponding correction of the reference signals.

Fig. 1

ein Beispiel eines schematischen Aufbaus eines Wertdokumentbearbeitungssystems;

Fig. 2

Beispiele einer nicht korrigierten Remissionskurve und einer mit einem Spektrometer gemessenen Remissionskurve;

Fig. 3

Beispiele einer korrigierten Remissionskurve und einer mit einem Spektrometer gemessenen Remissionskurve; und

Fig. 4

Beispiele für die Emission unterschiedlicher Lichtquellen.

Further advantages, features and possible applications of the present invention result from the following description in connection with the figures. Show it:

Fig. 1

an example of a schematic structure of a value document processing system;

Fig. 2

Examples of an uncorrected reflectance curve and a reflectance curve measured with a spectrometer;

Fig. 3

Examples of a corrected reflectance curve and a reflectance curve measured with a spectrometer; and

Fig. 4

Examples of the emission of different light sources.

Fig. 1 shows an example of a schematic structure of a value document processing system 1 with an input compartment 2, in which a stack of value documents to be processed, in particular banknotes 3, is provided, and a separator 8, from which the bottom banknote of the entered stack is recorded and sent to a - in The selected representation is only shown schematically - transport device 10 is handed over, which transports the banknote in the transport direction T to a sensor device 20.

In the example shown, the sensor device 20 comprises light sources 24 and 25 - which are shown only in a very schematic form - for irradiating the banknote with light, in particular in the visible and/or infrared and/or ultraviolet spectral range, as well as a first, second and third sensor 21, 22 respectively. 23, which is preferably designed as a so-called line camera and detects light emanating from the banknote, in particular in the visible and / or infrared and / or ultraviolet spectral range, by means of sensor elements arranged along a line and converts it into corresponding sensor signals.

Light-emitting diodes (LEDs) are preferably used as light sources 24 and 25. Even if two light sources 24 and 25 are indicated in the example shown, it may be preferred to provide more than two light sources. Likewise, any other light sources such as fluorescent lamps, flash lamps, (filtered) incandescent lamps or the like can be used for the inventive method instead of LEDs.

In principle, the at least two light sources can also be implemented by a light source in conjunction with at least one switchable filter, provided that at least two individually addressable, different spectra are thereby made available. In the further description, this constellation is further described as two light sources or several light sources.

The sensor device 20 preferably has a plurality of light sources which emit light in different spectral ranges. In particular, the respective spectral ranges of the light sources can be selected so that they together emit light in the spectral range in which the remission or transmission behavior of the banknote is to be tested. This spectral range is preferably between approximately 350 and 1100 nm. For example, three LEDs can be combined, each of which emits light in the ultraviolet, visible and near infrared spectral range.

In the example shown, the first and second sensors 21 and 22 detect light remitted, ie diffusely reflected and/or directed reflected, from the front and back of the banknote and convert this into corresponding sensor signals. The third sensor 23 located in the area of the front of the banknote, on the other hand, detects the light emitted by a light source 24 and preferably hitting the banknote at an angle and passing through it passing through, ie transmitted, light and converts this into corresponding sensor signals.

Preferably, the line with the sensor elements of the respective sensor 21, 22 or 23 runs essentially perpendicular to the transport direction T of the banknotes, so that with each reading process of the sensor line of the respective sensor 21, 22 or 23, a sensor signal curve is obtained along the sensor line, which corresponds to an intensity profile of the light that is transmitted or remitted by the banknote in a direction perpendicular to the transport direction T.

The sensor device 20 shown is preferably designed to check reflectance and/or transmission curves at different locations on a banknote. For this purpose, one location on the banknote is illuminated with light from one of the light sources 24, 25 at a specific wavelength λ and the light remitted or transmitted by the banknote is detected with one of the sensors 21, 22 or 23 and converted into corresponding sensor signals. Preferably, these sensor signals are then each divided by a reference signal determined using a white reference, whereby a standardized reflectance or transmission value is obtained at the location of the banknote at the wavelength λ .

Preferably, the banknote is successively illuminated with light of different wavelengths and the respectively remitted or transmitted light is detected. In another embodiment, several, up to (n-1) light sources can also be active at the same time.

The light sources 24 and 25 are clocked so quickly that the banknote has hardly moved despite the transport during a cycle in which all the different wavelengths are switched through, so that for all different wavelengths are measured at essentially the same location on the banknote. In this way, not only a possibly standardized reflectance or transmission value but also a possibly standardized reflectance or transmission curve is obtained for this point.

The sensor signals generated by the sensors 21 to 23 of the sensor device 20, in particular the corresponding reflectance or transmission curves, are forwarded to a control device 50 and an evaluation device 51. The evaluation device 51 can be contained in the control device 50 or can also form a unit separate from the control device 50. In particular, the evaluation device (51) has a memory function for storing previously calculated correction parameters that are used to calculate corrected sensor signals.

In the evaluation device 51, the sensor signals, in particular the reflectance or transmission curves, are used to check the banknote, with statements about various properties of the respective banknote being derived from the respective sensor signals, such as authenticity, degree of contamination, wear, defects and the presence of Foreign objects, such as adhesive strips. Depending on the properties of the respective banknote determined in the evaluation device 51, the transport device 10 and the switches 11 and 12 are controlled along the transport route by the control device 50 in such a way that the banknote is fed to one of several output compartments 30 and 31 and stored there. For example, 30 banknotes that were recognized as genuine are stored in a first output compartment, while banknotes that are classified as inauthentic or suspected of being counterfeit are stored in a second output compartment 31. The reference number 13 at the end of the transport route shown is intended to indicate that there are additional output compartments and/or other facilities, for example for storing or destroying banknotes, can be provided. If, for example, the examination of a banknote shows that it is genuine, but does not meet certain fitness criteria with regard to contamination, wear, defects or the presence of foreign objects, it can be sent directly to a shredder for destruction.

In the example shown, the value document processing system 1 further comprises an input/output device 40 for inputting data and/or control commands by an operator, for example using a keyboard or a touch screen, and outputting or displaying data and/or information about the processing process, in particular about the processed banknotes.

In the evaluation device 51, corrected sensor signals, in particular corresponding corrected reflectance or transmission curves, are preferably used to check the banknote, which reflect the actual reflectance or transmission behavior of the banknote much more accurately than the uncorrected reflectance or transmission curves. This is explained in more detail below.

Figure 2 shows an uncorrected reflectance curve 15 obtained with the sensor device 20 in the spectral range between approximately 400 and 1050 nm in comparison with a reflectance curve 16 measured with a calibrated spectrometer, which reflects the actual reflectance behavior of the location of the banknote under consideration. As can be seen from the comparison, the uncorrected reflectance curve 15 shows noticeable artifacts, which in this example appear as jagged reflectance peaks at approximately 590 nm and approximately 650 nm. As experiments have surprisingly shown, These reflectance peaks occur despite normalization of the reflectance curve 16 using reference signals that were determined on a white reference.

The invention achieves, among other things, that such reflectance peaks are eliminated from the reflectance curve 15 or at least significantly reduced, so that the corrected reflectance curve obtained here comes significantly closer to the actual reflectance curve 16.

Figure 3 shows a reflectance curve 17 corrected in accordance with the invention in the spectral range between approximately 400 and 1050 nm in comparison with the reflectance curve 16 measured with a spectrometer. As can be seen from the diagram, the course of the corrected reflectance curve 17 corresponds significantly better to the course of the reflectance curve 17 with the spectrometer measured reflectance curve 16 corresponds to that of the uncorrected reflectance curve 15 (cf. Fig. 2 ) the case is.

Ideally, the spectral illumination distributions of LEDs correspond to laser-like Dirac functions at the corresponding wavelengths, that is, they have a "needle-shaped" spectral intensity distribution of the emitted light around a nominal wavelength. However, because this is often not the case in practice, the reflectance curves obtained using LED lighting on banknotes are somewhat distorted. The spectral illumination distributions of real LEDs usually have a certain extent around the nominal wavelength, so that the remission spectrum is somewhat smoothed. This emission of light is also referred to as main emission in connection with the invention. In addition, it has been found that some LEDs, in addition to the main emission, also show secondary emissions in completely different wavelength ranges, which change the shape of the remission curve in a surprisingly noticeable and particularly disturbing manner.

The approach according to the invention for correcting the remission or transmission curves is based, among other things, on the knowledge that disturbing artifacts, in particular remission or transmission peaks, can be caused by secondary emissions from the respective light sources, in particular LEDs. The correction methods according to the invention for the mathematical elimination or at least reduction of these effects are explained in more detail below.

With a simple numerical correction method, the original reflectance or transmission curves in the wavelength range of LEDs with secondary emissions could simply be smoothed, e.g. with a moving average over three reference points. Although this smoothes the representation of the curves in a simple and quick manner, it also potentially creates new artifacts, particularly in the case of highly structured remission or transmission spectra with steep edges.

In order to better approximate the reflectance or transmission curves to the actual reflectance curves 16 (see Fig. 2 and 3 ) or transmission curves, a correction method is preferably used that takes physical properties into account in the emission, remission or transmission and detection processes.

This model is explained below as an example for reflectance measurements, but can of course also be applied in a completely analogous manner to transmission measurements.

With the help of this model, the generated sensor signals for a reflectance curve are simulated with knowledge of the illumination distributions and the detector sensitivity distribution for all LEDs. As previously explained, captured the sensor device 20 records both the main and secondary emissions of the light sources 24, 25 or the remission or transmission caused thereby.

wobei S_k (λ) die Beleuchtungsverteilung von Kanal k, d.h. der k-ten LED, D(λ) die Detektorempfindlichkeit, d.h. die Empfindlichkeit des Sensors, und r(λ) die tatsächliche Remissionskurve der Banknote ist.The formula applies to the sensor signal I _k , which is generated when a banknote is illuminated with the kth LED ( k = 1 ... n ). $$I_k={\displaystyle \int S_k\left(\lambda \right)D\left(\lambda \right)r\left(\lambda \right)\mathit{d\lambda }},$$

where S _k ( λ ) is the illumination distribution of channel k , i.e. the kth LED, D ( λ ) is the detector sensitivity, i.e. the sensitivity of the sensor, and r ( λ ) is the actual reflectance curve of the banknote.

d.h. das erhaltene Sensorsignal I_k würde bis auf den Kalibrierfaktor D(λ_k ) der tatsächlichen Remission r(λ_k ) entsprechen. Da sich bei einem Weißabgleich dieser Kalibrierfaktor wegkürzen würde, würde man also bei einer Beleuchtungsverteilung in Form von Dirac-Funktionen die exakten Remissionswerte erhalten.If S _k ( λ ) were a Dirac function at position λ _k , it would hold $$I_k=D\left({\lambda }_k\right)r\left({\lambda }_k\right),$$

That is, the sensor signal I _k obtained would correspond to the actual reflectance r ( λ _k ) except for the calibration factor D ( λ _k ). Since this calibration factor would be reduced in the case of a white balance, the exact reflectance values would be obtained with an illumination distribution in the form of Dirac functions.

In specific applications, r ( λ ) can often be at discrete, equidistant wavelengths λ _i (i = 1 ... m). Accordingly, S _k ( λ ) and D ( λ ) must also be determined for these wavelength values λ _i , if necessary by interpolation.

With the definitions S _ki = S _k ( λ _i ), D _i = D ( λ _i ) and r _i = r ( λ _i ) applies $$I_k={\displaystyle \sum _{i=1}^m{S}_{\mathit{ki}}{D}_i{r}_i\mathit{\Delta \lambda }}.$$

With A _ki = S _ki D _i Δλ applies $$I_k={\displaystyle \sum _{i=1}^m{A}_{\mathit{ki}}{r}_i}.$$

erhält man I als Matrixmultiplikation von r mit A $$I=\mathit{Ar}.$$

With the spellings $$I=\left(I_k\right),A=\left(A_{\mathit{ki}}\right),r=\left(r_i\right)$$

one obtains I as a matrix multiplication of r by A $$I=\mathit{Ar}.$$

berechnet, wobei $$w=\left(w_i\right)=\left(w\left({\lambda }_i\right)\right)$$

der tatsächlichen Remissionskurve einer sog. Weißreferenz, also einer Referenz mit im jeweils betrachteten Spektralbereich gleich hohen Remissionswerten nahe bei 1, entspricht.The vector I is preferably normalized using a white balance. For this purpose $$\frac{\mathit{Ar}}{\mathit{Aw}}$$

calculated, where $$w=\left(w_i\right)=\left(w\left({\lambda }_i\right)\right)$$

corresponds to the actual reflectance curve of a so-called white reference, i.e. a reference with the same high reflectance values close to 1 in the respective spectral range under consideration.

Taking this model into account, the generated sensor signals, i.e. the measured reflectance curves, can then be corrected as follows.

erhalten.Let A ₀ be the matrix analogous to A that is obtained if any secondary emissions in the data set are removed in the sensitivity curve S _ki = S _k ( λ _i ). With a sensor with corresponding LEDs, the actual remission curve would then be $$\frac{A_{0}r}{A_{0}w}$$

receive.

The following commutative diagram of images applies to the present model and the correction method derived from it:

The remission vector r is mapped onto the (correct) sensor signals BR in the n radiation channels with the discretization at m wavelengths either without secondary emissions with A ₀ , or alternatively via the measurement with secondary emissions (R) and their subsequent correction via the correction matrix B. Here B is defined as B = A ₀ A ⁺ , where A ⁺ is the pseudoinverse of A. For the case n = m, A ⁺ = A ^-1 is the inverse of the matrix A.

According to a mathematical definition, in the present case it is a pseudoinverse A ⁺ , which can also be referred to as a generalized inverse of A, if and only if: $${\mathit{AA}}^{+}A=A\mathrm{and}{A}^{+}{\mathit{AA}}^{+}=A^{+}.$$

also die richtigen, d.h. tatsächlichen Remissionswerte. Im Fall m ≠ n erhält man mit Hilfe der Pseudoinversen eine Näherung von $$\frac{A_{0}r}{A_{0}w}\mathit{durch}\frac{BR}{BW}.$$

The sensor signals R = Ar and W = Aω are obtained with the sensors 21, 22. Without correction according to the invention one would calculate R/W. However, with correction, BR/BW is calculated. In the case m = n this results $$\frac{bR}{bW}=\frac{A_0{A}^{-1}Ar}{A_0{A}^{-1}Aw}=\frac{A_{0}r}{A_{0}w},$$

i.e. the correct, ie actual, remission values. In the case m ≠ n, an approximation of is obtained using the pseudoinverses $$\frac{A_{0}r}{A_{0}w}\mathit{through}\frac{bR}{bW}.$$

In a further development of this method, it can be provided not only to remove the secondary emissions of the LEDs when creating A ₀ , but also to replace their Gaussian-like or even asymmetrical distributions with discrete Dirac functions in the entries for the respective wavelength range. This has the advantage that the edges of the reflectance curves become steeper and therefore more precise.

Overall, the correction method described allows a reliable elimination or at least reduction of remission or transmission peaks due to secondary emissions from the light sources, so that it can be used in an advantageous manner - especially in sensor and / or evaluation devices with sufficiently high computing power. A spectral correction is carried out by changing the shape of the remission spectrum. This correction is dynamic, i.e. the correction parameter depends not only on the systematic (static) alternating interference between the radiation source channels, but also on the current measured values of the radiation source channels involved.

In a preferred variant of this method, a reliable correction of the sensor signals can be carried out in real time even with lower computing capacities. Here, LEDs are used to irradiate the banknote, each of which has at most one secondary emission, which is preferably close to a wavelength at which one or more of the other LEDs is or are available, which in turn preferably has or have no secondary emission .

In a preferred case, the wavelength of the main emission of the other LED is shifted by less than 120nm from the wavelength of the secondary emission of the first LED, more preferably by less than 50nm, even more preferably by less than 10nm, depending on the desired spectral Resolution of the transmission or remission curves and the number of light sources.

mit einem Gewichtungsfaktor a_k . Wird die Remission relativ zu einer Weißreferenz gemessen, also mittels anhand der Weißreferenz erhaltener Referenzsignale normiert, so kürzt sich der Gewichtungsfaktor a_k weg: $$\frac{a_{k}r\left({\lambda }_k\right)}{a_{k}w\left({\lambda }_k\right)}=\frac{r\left({\lambda }_k\right)}{w\left({\lambda }_k\right)}=\frac{r_k}{w_k}$$

If one first considers the ideal case, that the illumination distribution is approximately a narrow Gaussian curve around the wavelength λ _k without any secondary emissions. Then you get approximately $$I_k=a_{k}r\left({\lambda }_k\right),$$

with a weighting factor a _k . If the reflectance is measured relative to a white reference, i.e. normalized using reference signals obtained from the white reference, the weighting factor a _k is reduced: $$\frac{a_{k}r\left({\lambda }_k\right)}{a_{k}w\left({\lambda }_k\right)}=\frac{r\left({\lambda }_k\right)}{w\left({\lambda }_k\right)}=\frac{r_k}{w_k}$$

In Figure 4 the intensity I of the emission from two light sources 24, 24' is shown as an example. To correct the first light source 24 with a first main emission with a wavelength λ ₁ and a first secondary emission with a wavelength λ ₂ , the second light source 24 'is required, which has a main emission with a wavelength that is the same or similar to the wavelength λ ₂ of the correcting secondary emission of the first light source 24. For LEDs with a main emission at 570 nm, for example, it has been shown that there is often a secondary emission at 850 - 870 nm. A second LED is therefore used for correction, which has a main emission at approx. 850 nm.

For this simplest case, in which a light source 24 has a main emission at a wavelength λ ₁ and only a secondary emission at a second wavelength λ ₂ , the quotient is obtained instead of the actual normalized reflectance value r ₁ / ω ₁ $$R=\frac{a_1{r}_1+a_2{r}_2}{a_1{w}_1+a_2{\mathrm{<figure-callout\; id="2"\; label="w"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">w</figure-callout>}}_2}=\frac{r_1+\frac{a_2}{a_1}{r}_2}{w_1+\frac{a_2}{a_1}{\mathrm{<figure-callout\; id="2"\; label="w"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">w</figure-callout>}}_2}.$$

With a ₂ / a ₁ = a and the assumptions ω ₁ = ω ₂ = 1 one obtains $$R=\frac{r_1+{\mathit{<figure-callout\; id="2"\; label="ar"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">ar</figure-callout>}}_2}{1+a}.$$

This results in a correction $$r_1=\left(1+a\right)R-{\mathit{<figure-callout\; id="2"\; label="ar"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">ar</figure-callout>}}_2=\left(1+a\right)R-a\frac{r_2}{{\mathrm{<figure-callout\; id="2"\; label="w"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">w</figure-callout>}}_2}.$$

From the reflectance R obtained using sensors, the unadulterated, ie actual, value r _{2 /} w ₂ and the correction parameter a , one can calculate the corrected, ie actual, value r ₁ / ω ₁ of the standardized reflectance.

The correction parameter a = a ₂ / a ₁ is preferably determined using two methods.

In the first method, the a _i are determined directly via the product of the measured spectral distributions of the light emission of the light source 24 with the measured sensitivity of the detector or detectors at the wavelength λ _i . In this case, the intensities of the main and secondary emission of the first light source 24 with the secondary emission as well as the detector sensitivities at the wavelengths λ ₁ and λ ₂ must be measured. The intensity of the further light source 24', which emits with its main emission in the range of λ ₂ , does not necessarily have to be measured.

With the second, measurement-technically simpler and more accurate method, no direct measurements of the intensities of the light sources or detector sensitivities are necessary. Here, the correction parameter a is calculated from the known variables R (sensor signals, if necessary standardized) and the actual reflectance r ₁ , r ₂ of a previously characterized test sample: $$a=-\frac{r_1-\mathrm{<figure-callout\; id="2"\; label="R\; r"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">R</figure-callout>}}{r_{}}.$$

The actual reflectance of the test sample can be known using standard color charts, or can be determined by direct measurement with a spectrometer on the test sample. The test sample preferably has sufficiently high reflectance values >0.2, particularly preferably >0.5, so that sufficiently high signal intensities and thus sufficient accuracy in determining a is achieved.

In the general case of several light sources with several secondary emissions, the first method for determining the correction matrix B can also be used: The S _ki for each wavelength λ ₁ are determined in turn by measuring the light emissions of the kth light source and with the associated detector sensitivity Di the matrix entries A- _ki =S _ki D _i Δλ are calculated. The associated adjusted matrix A ₀ is then defined and the correction matrix B=A ₀ A ⁺ is calculated.

Alternatively or in addition to the methods described above, the invention also includes the variants and embodiments set out below.

The spectral illumination contributions can come not only from the main and secondary emissions of individual light sources, but also from simultaneous illumination of the document of value with at least two light sources with different spectral distribution. In this case too, the correction of the sensor signals according to the invention via the algorithm according to the invention enables a correct extraction of the remission or transmission curves.

In one variant, a banknote is first illuminated with two different LEDs (LED A and LED B) at the same time and then only with one of the two LEDs, e.g. LED B, so that by subsequently forming the difference A = (A + B) - B on the Signal can be closed that would have been obtained with lighting with LED A alone.

In a preferred special case of this variant, LED A emits light in the UV range, LED B in the visible (VIS) or IR range. The UV signal can then be determined without the banknote having to be irradiated solely by the UV LED.

In a further variant, the sensor device 20 is designed so that the banknote is always simultaneously irradiated with LEDs of different wavelength ranges. For example, the banknote can be illuminated simultaneously with the various overlays of LEDs A + B + C, of LEDs A + B, and of LEDs A + C one after the other.

To correct the sensor signals, the algorithm according to the invention can then be used as described above if the spectra of the individual LED light emissions are used by the spectra of the kth combined LED overlays. The S _ki are determined one after the other for each wavelength λ _i and the matrix entries A _ki =S _ki D _i Δλ are calculated using the associated detector sensitivity D _i . The associated adjusted matrix A ₀ is then defined and the correction matrix B=A ₀ A ⁺ is calculated.

Claims (16)

1. Apparatus for checking value documents (3), in particular bank notes, comprising

   - at least two radiation sources (24, 25) for emitting electromagnetic radiation by means of which a value document (3) is irradiated,

   - at least one sensor (21-23) for detecting the electromagnetic radiation reflected and/or transmitted by the value document (3) and generating corresponding sensor signals (R), with components (R_k ) assigned to the radiation sources (24, 25), wherein the sensor signals are corresponding reflection and/or transmission signals, and

   - an evaluation device (51), which is designed to check the value documents on the basis of corrected sensor signals and to derive these corrected sensor signals (BR; r ₁) from the sensor signals taking into account at least one spectral property (S_k (λ), S_ki ; a_k , a) of the electromagnetic radiation of the at least two radiation sources (24, 25), wherein during the derivation of the corrected sensor signals (BR; r₁ ), at least one linear combination is formed from components of the sensor signals (BR; r ₁) assigned to the different radiation sources (24, 25) (R_k),

   wherein the at least one spectral property (a_k, a) of the electromagnetic radiation of the radiation source (24, 25) is given by at least one parameter (a_k , a) which characterizes one or more spectral parts of the electromagnetic radiation of the radiation source (24, 25), and

   wherein at least one first parameter (a₁) characterizes the spectral part of a main emission of the electromagnetic radiation of the radiation sources (24, 25) and at least one second parameter (a₂) characterizes the spectral part of a secondary emission of the electromagnetic radiation of the radiation sources (24, 25).
2. Apparatus according to claim 1, wherein the at least one spectral property (S_k (λ), S_ki ) of the electromagnetic radiation of the at least two radiation sources (24, 25) is given by at least one spectral distribution (S_k (λ), S_ki ) of the electromagnetic radiation from at least one radiation source (24, 25).
3. Apparatus according to claim 2, wherein the at least one spectral distribution (S_k (λ), S_ki ) of the electromagnetic radiation from at least one of the at least two radiation sources (24, 25) is given by a first spectral distribution (S_k (λ), S_ki ) of the electromagnetic radiation emitted by at least one of the at least two radiation sources (24, 25) and by a second spectral distribution of the electromagnetic radiation that is different from the first spectral distribution (S_k (λ), S_ki ).
4. Apparatus according to claim 3, wherein the first spectral distribution (S_k (λ), S_ki ) corresponds to a spectral distribution of the electromagnetic radiation emitted by at least one of the at least two radiation sources (24, 25) with a main emission and at least one secondary emission, and the second spectral distribution corresponds to the first spectral distribution (S'_k (λ), S'_ki ) without the at least one secondary emission.
5. Apparatus according to claim 4, wherein the wavelength of the main emission of a second radiation source (24, 25) is shifted by less than 120 nm, preferably by less than 50 nm, more preferably by less than 10 nm, from the wavelength of the secondary emission of a first radiation source.
6. Apparatus according to any of claims 2 to 5, wherein the evaluation device (51) is designed to detect the corrected sensor signals (BR) by multiplying the sensor signals (R) with a correction matrix (B) which is derived from the at least one spectral distribution (S_ki , k = 1 ... n, i = 1 ... M)) of the electromagnetic radiation of a first number (n) of radiation sources (24, 25) at a second number (m) of discrete wavelengths (λ_i ) and at least one spectral profile (D_i ) of the sensitivity of the at least one sensor (21-23) for electromagnetic radiation, and wherein the correction matrix (B) has at least one non-diagonal element different from 0.
7. Apparatus according to claim 6, wherein the spectral distributions (S_ki , i = 1 ... M)) of the electromagnetic radiation of the individual radiation sources (k = 1...n) are not all identical or are all different.
8. Apparatus according to either claim 6 or claim 7, wherein the correction matrix (B) corresponds to the product of a second matrix (A₀ ) and a pseudo inverse (A ⁺) of a first matrix ((B = A₀A ⁺), wherein the matrix elements (A_ki , k = 1 ... n, i = 1 ... M)) of the first matrix (A) correspond (A_ki = S_ki D_i Δλ) to the product of the first spectral distribution (S_ki , k = 1 ... n, i = 1 ... m) of the electromagnetic radiation emitted by the first number (n) of radiation sources (24, 25) at a second number (m) of discrete wavelengths (λ_i ), with the spectral profile (D_i ) of the sensitivity of the at least one sensor (21-23), and a wavelength distance value (Δλ) between two discrete wavelengths (λ_i ) in each case.
9. Apparatus according to claim 8, wherein the matrix elements of the second matrix (A₀ ) correspond (A_ki = S'_ki D_i Δλ) to the product of the second spectral distribution (S' _ki ) of the electromagnetic radiation, with the spectral profile (D_i ) of the sensitivity of the at least one sensor (21-23), and a wavelength distance value (Δλ) between each two discrete wavelengths (λ_i , i = 1 ... m).
10. Apparatus according to claim 1, wherein the evaluation device (51) is designed to derive the corrected sensor signals (r₁) from the sensor signals (R) taking into account the first and a second parameter (a₁ , a₂ ) or a parameter (a) derived from the first and second parameters ((a₁ , a₂ ) which in particular corresponds to the quotient (a ₁/a ₂) from the first and second parameters (a₁, a₂).
11. Apparatus according to claim 10, wherein the evaluation device (51) is designed, during the derivation of the corrected sensor signals (r₁) from the sensor signals (R), to also take into account a value (R₂ ) which represents a measure for the electromagnetic radiation originating from the value document (3), in particular reflected and/or transmitted from the value document (3), in the region of the secondary emission of the electromagnetic radiation of the radiation source (24, 25).
12. Apparatus according to any of the preceding claims, wherein the evaluation device (51) is designed to normalize (BR/BW; r₁ /w₁) the corrected sensor signals (BR; r₁ ) on the basis of corrected reference signals (BW; w₁ ), wherein the corrected reference signals (BW; w₁ ) are derived from reference signals (W), which are generated by the at least one sensor (21-23) during the detection of the electromagnetic radiation emitted from a reference document, taking into account the at least one spectral property (S_k (λ), S_ki ; a ₁ , a₂ ) of the electromagnetic radiation of the at least two radiation sources (24, 25).
13. Apparatus according to claim 12, wherein the evaluation device (51) is designed, during the derivation of the corrected reference signals (BW) from the reference signals (W), to take into account the sensitivity of the at least one sensor (21-23) for electromagnetic radiation, in particular in the form of a spectral profile (D(A), D_i) of the sensitivity of the at least one sensor (21-23).
14. Apparatus according to any of the preceding claims, wherein the evaluation device (51) comprises a memory function for providing previously calculated correction parameters (B, a_k, a).
15. Value document processing system (1) comprising at least one apparatus (2, 8.10-13, 30, 31, 50) for processing, in particular for conveying and/or counting and/or sorting, value documents (3), in particular banknotes, and an apparatus (21-23, 51) for checking value documents (3) according to any of the preceding claims.
16. Method for checking value documents (3), in particular banknotes, by means of an apparatus according to any of claims 1 to 14, comprising the following steps:

    - irradiating a value document (3) with electromagnetic radiation of at least two radiation sources (24, 25),

    - detecting the electromagnetic radiation reflected and/or transmitted by the value document (3) and generating corresponding sensor signals, with components assigned to the radiation sources, wherein the sensor signals are corresponding reflection and/or transmission signals, and

    - deriving corrected sensor signals from the sensor signals, taking into account at least one spectral property (S_k (λ), S_ki; a_k , a) of the electromagnetic radiation of the at least two radiation sources (24, 25), wherein, during the derivation of the corrected sensor signals, at least one linear combination is formed from components of the sensor signals assigned to the different radiation sources,

    wherein the at least one spectral property (a_k, a) of the electromagnetic radiation of the radiation source (24, 25) is given by at least one parameter (a_k, a) which characterizes one or more spectral parts of the electromagnetic radiation of the radiation source (24, 25), and

    wherein at least one first parameter (a₁) characterizes the spectral part of a main emission of the electromagnetic radiation of the radiation sources (24, 25) and at least one second parameter (a₂) characterizes the spectral part of a secondary emission of the electromagnetic radiation of the radiation sources (24, 25), and

    - checking the value documents on the basis of the corrected sensor signals.

Images