EP2095341B1 - Method for identifying soiling and/or colour fading in the region of colour transitions on documents of value, and means for carrying out the method - Google Patents

Description

The present invention relates to a method for detecting soiling and / or color wear in the region of color transitions in at least one section of a value document, a device for carrying out the method, a computer program for carrying out the method and a data carrier with the computer program.

Under value documents are understood in the context of the invention card or sheet-shaped objects that represent, for example, a monetary value or an authorization and therefore should not be arbitrarily produced by unauthorized persons. They therefore have features which are not easy to manufacture, in particular to be copied, whose presence is an indication of the authenticity, i. the manufacture by an authorized agency. Important examples of such value documents are chip cards, coupons, vouchers, checks, stocks and in particular banknotes.

Value documents are typically provided with color design, for example more or less complex color patterns and / or color illustrations and / or characters and / or character combinations, for reasons of design, distinctness and protection against simple forgeries.

When using the value documents, they can be contaminated. In the context of the present invention, a contamination is understood to mean, in particular, a change in the color design of the value document caused by the introduction or introduction of substances onto or in value documents or by irradiation of the value documents with electromagnetic radiation. In particular, value documents can be considered as soiling Spots, for example, caused by the intentional or unintentional application of colored or color-changing liquids, or have colored markings.

In addition, color abrasions can occur. In the following, color changes are understood in particular to mean color changes due to fading, removal of printing ink and / or washing out of paint by means of water or other solvents for colors of the value document. Such color abrasions can occur, in particular with banknotes with polymer substrates, if printing inks with which the banknotes are printed do not adhere with sufficient stability to the polymer substrates.

In order to ensure the distinctness of value documents and / or the recognizability of the value documents with regard to authenticity for users, in particular also without the aid of technical devices, it is necessary to be able to recognize soiled value documents. Due to the very large number of documents of value circulating in the case of banknotes, for example, automatic or automatic recognition is desirable.

The automatic detection of soiling and / or Farbabnutzungen with preferably high speed, however, is complicated by the fact that occur on the value documents color transitions, which either the normal color design of the document of value or caused by dirt and / or Farbabnutzungen not over the entire surface of the entire area of the value document. The color transitions need not be sharp, such as in the range of edges of an image on a document of value, to be formed, but can also over a predetermined distance on the value document done slowly. Soiling and / or color wear must therefore be recognized.

In US 6,798,900 B1 US 4,848,160 A are descriptive of a paper sheet identification method and apparatus that statistically process information, represent characteristics of a bill, compact it into a smaller number of characteristics, and determine the authenticity of the bill using those characteristics. The paper sheet identifying method comprises separately performing (a) a detecting step and (b) a processing step, which will be described below, for a genuine paper sheet and a paper sheet to be inspected, determining reference data based on the detecting step and the processing step for the true paper sheet and the determination of the authenticity of the paper sheet by comparing the data obtained with the reference data. The paper sheet identifying device includes signal generating units that generate a plurality of pattern signals to be checked for a plurality of physical characteristics representing a pattern printed in a predetermined area on the paper sheet. It further comprises a computing unit, a reference data storage unit a determination unit.

In EP 0 947 964 A1 It is described that in order to identify securities and / or the authenticity check of the authenticity of securities from at least one areal detail, a line vector is generated whose components each represent a total value taking into account all the samples of a column of the section. So that the detection can be largely independent of location, it is first determined in which of several predetermined angular ranges the angular position of the security lies. For pattern matching, a reference pattern associated with the selected angle range is used.

In EP 0 718 808 A2 A system for authenticating documents, such as banknotes, checks, or the like, is described, in which a document to be inspected is transported past a sensor station which provides a digital color image of the document, which is stored in a memory and in the form of several digital images was obtained by illuminating the document with light of different colors. The stored color image is processed using, for example, template matching to determine the type and orientation of the document. The type and orientation information is used to select a plurality of regions of the digital color image for color intensity analysis. For example, a statistical test method may be used that uses stored values derived from the processing of real documents. Documents of different types are processed in a uniform way.

In EP 0 560 023 A1 a device is described for classifying a pattern, in particular a banknote or a coin, in terms of value and authenticity. The apparatus comprises a recording system for metrological detection of vectors of a specimen, a preprocessing system for transforming the measured vectors into local feature vectors, and a learning classification system for performing a plurality of tests. A first activity compares each of the local feature vectors with a vectorial setpoint in a first test. Only when the first check succeeds does the first activity associate the local feature vectors to global row feature vectors using first estimates stored in a database. In a second test, a third activity compares the global row feature vectors with corresponding target sizes and, if the second check is successful, calculates a single global area vector, of which a fourth activity in a third check is its distance to Mahalanobis to an estimated area Vector compares to a setpoint. The examinee is safely classified if all three exams are successful.

In EP 0 280 435 A2 there is described a monitoring system, in particular for use with document sorting machines, comprising a sensor array having a sensor for detecting a respective feature of an article and generating a corresponding first output signal and having a detector controller for applying at least one algorithm of a given set of Algorithms to the first output signal of the sensor in connection with the detector controller supplied data to produce one or more second output. A detector processor, common to the sensor arrays, provides the data and algorithm selection information to the detector controllers to control which of the algorithms of the set of algorithms is applied by the detector controller and in response to the or each second output signal and reference data to produce a third output representative of the feature of the article to be monitored.

In US 4,587,343 there is described a banknote validator controlled by a microprocessor. This contains a transport for moving an inserted note longitudinally past an optical scanning station. Infrared and visible color reflectance readings and opacity readings are acquired along multiple longitudinal tracks on the note. The microprocessor normalizes the reflectance readings to account for variations in fouling, and compares the normalized reflectance readings and the opacity readings to stored acceptance tape data, with a correction for pattern shift, if necessary. The length of the note is also checked and a validation signal is given if the note passes the optical tests and the length test. During the cycle time in which no document is being tested, the microprocessor automatically adjusts the optical circuits to compensate for drift and dirt build-up. The microprocessor also provides a visual indication of detected malfunctions.

In US 2004/211644 A1 For example, a method and an apparatus for checking the authenticity of a document are described. The apparatus includes a profile processor adapted to collect a plurality of data profiles from suspect documents, a comparator adapted to compare the plurality of data profiles to a set of envelopes of a library document, and to determine that the document it is true that, when the plurality of profiles are in compliance with the amount of envelopes, and a pressure lock, which locks a cassette, which records authenticated documents, to a body of the device for checking the authenticity of the document.

In US 5,615,760 For example, a method and an apparatus for checking the authenticity of money objects are described. Money objects are thereby validated. determining whether n measured properties are within an n-dimensional ellipse associated with a particular monetary object. The center of the ellipse lies at the statistical mean of the property measurements for the specific object and each of the axes is related to the standard deviation for the respective property measurement.

It is therefore an object of the present invention to provide a method of detecting soils and / or color bleeding in the area of color transitions, in at least a portion of a value document that is readily operable, and to provide means for performing the method.

The object is achieved by a method for detecting soiling and / or color wear in the area of color transitions in at least one section of a value document of a predetermined value document type on the basis of processing data, the color coordinate values of picture elements in a color space depending on the position of the respective picture elements reproduce corresponding areas in the portion of the value document, and reference data representing a color reference distribution of color coordinate values in the color space depending on reference locations on a value document type value document predetermined for a type of value documents, wherein it is determined for each of the picture elements whether the Picture element associated color coordinate values in the color space of the color reference distribution. In this case, the color reference distribution is given by at least one predetermined, closed reference surface in the color space, which is given by at least one predetermined for the value document type linear distance and a predetermined distance of the points of the reference surface of the at least one linear distance. In the method, the positions of the picture elements whose color coordinate values were determined to be corresponding to the reference distribution are compared with predetermined reference positions on the value document, and depending on the As a result of the comparison, a presence or absence of staining and / or color deterioration in the area of a color transition is detected.

The object is further achieved by a device for detecting soiling and / or color wear in the region of color transitions in at least one section of a value document of a predetermined value document type having at least one interface for acquiring processing data, the color coordinate values of picture elements in a color space as a function of the position of represent respective areas corresponding to the picture elements in the portion of the value document, and having an evaluation means arranged on the basis of processing data acquired via the at least one interface and reference data representing a color reference distribution of color coordinate values prescribed for a type of value documents Play color space as a function of reference locations on a value document of the value document type to perform the inventive method. In particular, the evaluation device may be configured to check for each of the picture elements whether the color coordinate values associated with the picture element correspond to the color space of the color reference distribution, the color reference distribution being given by at least one predetermined closed reference area in the color space defined by at least one for the value document type given predetermined linear distance and a predetermined distance of the points of the reference surface of the at least one linear distance. Furthermore, it is then designed to compare the positions of the picture elements whose color coordinate values were determined to be corresponding to the reference distribution with predetermined reference positions on the value document, and depending on the result of the comparison, a presence or absence of one Identify contamination and / or color deterioration in the area of a color transition caused by soiling.

In order to detect colorings due to soiling or color wear, processing data are used according to the invention which reproduce or describe properties of picture elements which, combined according to their position, yield an image of the section. Specifically, the processing data represents the color coordinate values in a predetermined color space for picture elements in the predetermined portion of the document of value depending on the position of the picture elements corresponding to the area sections on the document of value.

In principle, it is sufficient for the method to examine only a predefined section of a value document, but preferably several sections or the entire value document are examined.

The section basically only needs to be one-dimensional or strip-shaped, so that the picture elements form only one row or column. Preferably, however, a two-dimensional image of a sheet portion is detected.

The color coordinate values can be generated in any manner. For example, images can be acquired simultaneously or sequentially in a plurality of spectral ranges, preferably as a function of the color space used. The images can be captured simultaneously for the entire section. However, it is also possible to acquire the data by moving a detection element row and the value document in a direction transverse to the detection element row at a predetermined timing, for example, a predetermined speed relative to each other, and thus detected line by line To compose data for the picture elements to form a picture or to think together. The color coordinate values can be obtained either directly or after transformation of other acquired image data by using suitable detection means.

The assignment of color coordinate values and position can take place in very different ways and in particular also depending on the way in which the color coordinate values are acquired. For example, the color coordinate values and the position of the pixel on the value document representing position coordinate values in a suitable coordinate system can be used as processing data for each picture element. However, when using pictures whose picture elements are detected in rows and columns in matrix form, for example, and linearly stored according to their order in passing successive lines or columns, it is also possible only the position in the order of data for the color coordinate values indicating the position on the page Value document to use. Further possibilities for the assignment are known to the person skilled in the art.

Color transitions and, in particular, color-related or color-wear-related color transitions are detected in the method by comparing the color coordinate values or the corresponding processing data with the color reference distribution. The color reference distribution as well as the corresponding reference positions are specified for a specific value document type. For value documents in the form of banknotes, the type can be predetermined, for example, by the type of currency and the denomination of the banknote. The value document to be examined possesses a given value document type, which, however, need not necessarily be known before the examination. The examination can then be done with color reference distributions for different value document types. Preferably, however, in the method, the type of the document of value to be examined is determined beforehand, for example using methods known to those skilled in the art, so that only a comparison with the color reference distribution of the identified document of value type needs to be made.

The color reference distribution is based on value documents of the given type. However, it does not necessarily require the state of a value document of the given type when new, i. to reproduce accurately after manufacture and before use; Rather, it may also take into account tolerances caused by commonly occurring contaminants and / or color abrasions that are not considered annoying. Only a color reference distribution is then used that allows for certain deviations from an ideal state. However, it is also possible to consider the occurrence of permissible stains and / or color abrasions by the criterion when color coordinate values correspond to the color reference distribution. A distribution of color coordinate values can be understood in particular to mean that the color coordinate values can lie within a predetermined volume of the color space that defines the distribution.

In order to achieve a high processing speed, in the method the color reference distribution is given by at least one predetermined closed reference surface in the color space, which is given by at least one predetermined for the value document type linear distance in the color space and a predetermined distance of the points of the reference surface of at least one linear distance is. The reference surface therefore encloses a region of the color space in which color coordinate values of picture elements corresponding to a color transition, which on one as still acceptable or not considered dirty value document occurs in the examined section. The reference surface can be given by only a linear distance. However, it is also possible that the reference surface is given by a plurality of distances whose distance from each other may be greater or preferably smaller than the predetermined distance. In particular, the reference surface may be defined by a polygon in conjunction with the predetermined distance, so that even complicated, predetermined for a value document type color transitions, for example, between three colors, can be reproduced and recognized.

For each of the picture elements, it is therefore determined whether the color coordinate values associated with the picture element in the color space correspond to the color reference distribution. For this purpose, a suitable criterion can be used, in which the reference data are received.

The advantage of this type of comparison and in particular the representation of the reference surface has the advantage that a check as to whether a color value coordinate values correspond to the color reference distribution can be carried out very quickly and easily, since the geometric structure in the color space is very simple.

Surprisingly, it has been shown that this type of specification of the color reference distribution in the most important color spaces, in particular for value documents in the form of banknotes, can be used very well. This is probably due to the fact that not very saturated colors are used on or in banknotes.

The method can in particular be carried out automatically by means of a corresponding device. In the device according to the invention In principle, the evaluation device can be designed as any analog, mixed analog-digital or purely digital circuit. It is also possible that this alone comprises a so-called "field programmable gate array" (FPGA), which has the advantage that corresponding components can be adapted solely by preprogramming for the method to be performed, but in the implementation operate as a digital circuit , In this way, the production costs can be kept low for small batches. Preferably, however, the evaluation device has at least one processor and a memory in which an inventive computer program executable by the processor or a program or computer program for carrying out the method according to the invention is stored.

The object is therefore further achieved by a computer program for detecting soiling and / or color wear in the region of color transitions in at least one section of a value document of a predetermined value document type, with instructions for at least one processor, in the execution of which the processor executes the method according to the invention and in particular the basis of processing data representing color coordinate values of picture elements in a color space depending on the position of the respective portions corresponding to the picture elements in the portion of the value document, and reference data representing a color reference distribution of color coordinate values in the color space in dependence on a type of value documents of reference documents on a value document of the value document type, determines for each of the picture elements whether the color coordinate values associated with the picture element are in the color space of the color reference distribution speak, wherein the color reference distribution is given by a predetermined closed reference surface in the color space, which by at least one given for the value document type linear distance and a predetermined distance of the points of the reference surface of the at least one linear distance is, the positions of the pixels whose color coordinate values were determined according to the color reference distribution compares with predetermined reference positions on the value document, and in dependence determines, from the result of the comparison, a presence or absence of soiling and / or color deterioration in the area of a color transition. Namely, soiling and / or color deterioration in the area of color transition will be manifested in that the corresponding color coordinate values do not correspond to the color reference distribution.

Such a computer program may in particular be stored in the memory of the device.

Another object of the invention is a disk on which a computer program according to the invention is stored. As a data carrier are in particular optical media such as CD or DVD, magneto-optical media, magnetic media such as hard disks and semiconductor memory, such as EEPROMs or flash memory into consideration, the content can be accessed by a corresponding device of a computer.

The term "processor" in the context of the invention means any processor, for example a microcontroller or a general-purpose processor or a digital signal processor, or a combination with a general-purpose processor and / or a signal processor and / or a microcontroller and / or an FPGA. The computer program is then designed according to the existing processor. In particular, the Evaluation have at least one FPGA, which is programmed so that at least parts of the method according to the invention by the FPGA can be performed. As a result, the execution speed of the computer program can be increased, since an FPGA may possibly have a higher execution speed for certain operations than a general purpose or signal processor.

The use of a programmable evaluation device has the advantage that the device can be easily adapted to new types of value documents.

The check as to whether the color coordinate values for the picture elements correspond to the color reference distribution can, in principle, be done in any desired manner. In the method, however, it is preferable that, for checking whether color coordinate values for a picture element correspond to the color reference distribution, it is checked whether a point corresponding to the color coordinate values in the color space is inside or outside the reference surface. In the computer program, the instructions are preferably such that, when executed, the processor checks to see whether color coordinate values for a picture element correspond to the color reference distribution, checks whether a point corresponding to the color coordinate values lies in the color space inside or outside the reference surface. Only one of the alternatives need be checked, since the fact that the point lies within the reference surface, which is also understood to lie on the reference surface, implies that it is not outside and vice versa. Such an examination can be carried out very quickly.

In principle, the reference surface can be specified in any desired manner, for example by interpolation points or an approximation by one- or multi-dimensional splines or a sum of orthogonal functions. However, it is preferable that, to check whether a point corresponding to the color coordinate values is within or outside the reference area, or the color coordinate values corresponding to the respective pixel correspond to the color reference distribution, a value is determined which determines the distance of the point corresponding to the color coordinate values from the color coordinates at least one route plays. In the computer program, the instructions are then preferably given in such a way that the processor determines a value when it is executed to check whether a point corresponding to the color coordinate values lies inside or outside the reference surface, or if the color coordinate values associated with the respective pixel correspond to the color reference distribution, which represents the distance of the point corresponding to the color coordinate values from the at least one path. Such a determination can be made very quickly. In addition, only a small amount of memory space is required to display the color reference distribution or the reference area. In particular, the distance in the metric of the color space can be used as the distance, preferably the Euclidean distance. The determined value, however, only needs to reproduce the distance; Namely, the determination of the distance usually requires complicated or slow operations such as dragging a root, so that when using, for example, the distance square as a value representing the distance, the execution of the program can be greatly accelerated. If the color reference distribution is given by two or more lines or a polygon, the criteria for color coordinate values for a picture element of the color reference distribution of at least one of the lines may be a distance smaller than the predetermined distance. By means of a criterion for the maximum permissible distance can then be decided for the respective color space point, whether he is the Reference distribution corresponds. It can then be marked or saved accordingly.

In principle, any color spaces can be used in the method. Preferably, however, color spaces with at least three dimensions are used, but it is also possible to use even higher-dimensional color spaces. Further, as color space, a color space specific to the sensor device used to acquire the processing data may be used. The color space can also be understood as any other space in which points are assigned by a bijective transformation respectively corresponding points in another color space. In particular, the color space used can be, for example, the RGB or the HSI color space.

In order to facilitate a transferability of the color reference distribution between different devices, in the method preferably a device-independently defined color space is used as the color space. In the computer program, the instructions are then preferably given in such a way that the processor uses a device-independently defined color space as its color space when executed. In particular, a normalized color space such as the CIE XYZ color space can be used as color space, for example.

Whether or not staining or color deterioration is troublesome to the visual appearance of a value document depends on an observer's ability to distinguish colors. In the method, therefore, a color space which is linearized with respect to the perception of color differences by humans is preferably used as the color space. In the computer program, the instructions are then preferred in such a way that when executed as a color space, the processor uses a color space that is linearized with respect to the perception of color differences by humans. By this is meant in particular that the coordinates of the color space are chosen so that visual perceived by human observer distances between colors are approximately proportional to the distances between the colors in the color space. In particular, the color space used may be, for example, a color space such as the CIE L * a * b color space in one of the known variants, the Hunter Lab color space or the CIE L * u * v color space. The use of such color spaces has the advantage that the criterion of when color value coordinate values correspond to the color reference distribution can be simply, but true for the actual handling of the value documents by humans, formulated by distances in the color space.

Thus, in the case where one of said special color spaces is used, in the method it is preferred that the given distance be given as a function of a minimum distance of two color coordinate values in the color space for colors from a given viewer under given viewing conditions can still be recognized as different. For this purpose, in the computer program, preferably the instructions are given so that in their execution by the processor the distance is given as a function of a minimum distance of two color coordinate values in the color space for colors that can be recognized as different by a given viewer under predetermined viewing conditions , As a given observer, a fictive observer may in particular be considered whose perception is given by an average value via the perception properties of a plurality of real people is. In particular, for example, when using a CIE L * a * b color space, the color reference distribution and the above criterion may be selected such that color coordinate values for a picture element are considered corresponding to the color reference distribution when the distance in the color space is from the at least one of the distances at least one of a plurality of distances is less than a value that is the sum of the predetermined distance and a value ΔE that is between 1 and 2, depending on the severity of the required distinction. If color deviations due to production fluctuations are also to be taken into account, the value can also be selected larger.

In principle, the same predetermined distance can be used for different value document types. In a development of the method, however, it is preferred that the distance is predetermined as a function of the value document type. In the computer program, the instructions are preferably given in such a way that, when executed by the processor, the distance is predetermined as a function of the value-document type. This embodiment has the advantage that for different value document types respectively different criteria for the permissible contamination and / or color wear can be defined, whereby the assessment of the state of the value documents can be improved. Depending on the application, the type of value document to be examined can be entered manually into the device or automatically determined by a machine with which the device is coupled.

In principle, image data which has been directly detected and possibly transformed into the color space used can be used as processing data in the method. However, it is preferred in the process that the formation the processing data are used image data of pixels of a captured image of the portion from which the color coordinate values and the positions can be determined, and the processing data for a picture element are detected using low-pass filtering of the image data. In the computer program, the instructions are then preferably such that the processor, when executing it to form the processing data, uses image data of pixels of a captured image of the portion from which the color coordinate values and the locations are determinable and the processing data for a pixel using a Low pass filtering of the image data determined. This embodiment has for many types of value documents with very fine patterns, especially banknotes, the advantage that due to the fine patterning color transitions need not be taken into account in the investigation, which the detection of dirt and / or Farbabnutzungen, which is usually a large area , in particular on an area greater than 0.5 mm ² , occur, relieved. In the context of low-pass filtering is then preferably the spatial resolution, ie the number of pixels per mapped area, reduced.

The device may preferably be combined with a sensor for acquiring image data from pixels. The invention therefore also provides an examination device with a sensor for acquiring image data of pixels which correspond to regions in a section of a value document, and a detection device according to the invention which is connected to the sensor for transmitting the image data.

More specifically, in the method of forming the processing data, image data of pixels of a captured image of the portion from which the color coordinate values and the positions can be obtained can be used and the processing data for one pixel are formed by using the image data of at least two pixels. In the computer program, the instructions may be such that the processor, when executing it to form the processing data, uses image data of pixels of a captured image of the portion from which the color coordinate values and the locations are determinable and the processing data for a pixel using the Forms image data of at least two pixels. In particular, the image data of the at least two pixels corresponding to the same area of the value document as the pixel may be used for local low-pass filtering, for example to form an optionally weighted average. Such a local low-pass filtering can often be carried out much faster than a non-local low-pass filtering, as it can be carried out, for example, in the spatial frequency space. Preferably, more than two pixels are used, more preferably at least the next adjacent pixels.

If the color coordinate values or picture elements corresponding to the color reference distribution are determined, their position, which can be more accurately understood as the position of the area section corresponding to the picture elements on the banknote, is compared with reference positions for them. As a result, it can be determined whether the recognized color transition is at a predetermined location.

Since the position of a document of value examined can vary in general relative to the sensor used to acquire the image data, before or during the comparison of the layers, a transformation of the layers of the picture elements or of the reference layers can take place, so that a better matching of the layers with the reference layers is achieved becomes.

The reference positions can be given for example by corresponding position coordinate values. In this case, it can be checked for comparison whether the positions determined lie within a predetermined distance from these reference positions. However, it is also possible that the reference layers are given by a, preferably two-dimensional, area. For comparison, it then only needs to be determined whether the layers of the picture elements lie within the area.

The result of the comparison may be that the picture elements whose color coordinates correspond to the reference distribution and whose positions correspond to the reference distribution are determined. Depending on the comparison, the method determines whether presence or absence of fouling and / or color degradation in the area of the color transition is detected. In principle, any criteria can be used for this purpose.

Preferably, the criterion for a permissible state of the value document is a criterion which is dependent on the number of picture elements which have been determined to correspond to the predetermined color transition and / or the number of picture elements which were not correspondingly determined for this color transition. For example, a maximum number of picture elements may be specified which do not correspond to the predetermined color transition. The criterion, in particular also the number, can be predetermined depending on the value document type.

Another object of the invention is a device for processing documents of value with an examination device according to the invention.

In addition, the subject matter of the invention is a method for determining a reference surface for a method according to the invention for detecting soiling and / or color wear, in which for a given number of value documents of a predetermined value document type respectively associated processing data, the color coordinate values of picture elements in a color space in dependence from the position of the areas corresponding in each case to the picture elements in the section of the value document, in which averages over the color coordinates reproducing first components and correlations between the color coordinates representing components of a matrix in the color space or another color space are determined from the processing data; in which directional data, which determine the eigenvectors associated with the two largest eigenvalues of the matrix, are determined, in the case of the data defining the linear distance from the directional data be determined so that the center of the route is determined by the average values, the direction of the route by the direction of the first eigenvector, and in which the length of the route and the predetermined distance from the route in dependence on the largest eigenvalue or the second largest Eigenvalue or or the correlation along the route and the largest correlation in a subspace orthogonal to the route reproducing size is determined. The subspace has one dimension less than the original color space, but may have emerged from this by bijective transformation.

This method makes it possible to easily determine the reference area or the data defining it. As a basis, value documents of a given type can be specified. Further, at least the portion of the value documents is specified, which later in the recognition of soiling and / or paint wear is also used.

As components of the matrix, according to a first alternative, the components of the correlation matrix can be determined and used.

However, it is also possible to identify and use as components of the matrix the components of the covariance matrix.

In principle, the detected color coordinate values can be used as processing data or part of the processing data for the determination of the eigenvalues. However, it may be advantageous to first transform the processing data by transforming the color coordinates into a different color space, which is, for example, device-independent. In particular, processing data used to determine the averaging components and correlations between color coordinate reproducing components may be color coordinates in a Lab color space.

The data defining the reference surface can basically be stored for the color space used to determine it. Preferably, the data defining the reference surface is stored transformed into values valid for the color space, which is later also used for the detection of soiling and / or color wear.

Although the methods described above have been described for colors in the narrower sense, ie in the visible range, the color space can also be at least one dimension for non-visible optical radiation, for example IR radiation in a predetermined wavelength range, and then, for example, be four-dimensional.

Fig. 1

eine schematische Darstellung einer Banknotenbearbeitungsvorrichtung,

Fig. 2

eine schematische Darstellung eines optischen Sensors und einer Farbwertauswerteeinrichtung in einer Steuer- und Auswerteeinrichtung der Banknotenbearbeitungsvorrichtung in Fig. 1,

Fig. 3

eine schematische teilweise Darstellung von drei Detektionselementzeilen der Banknotenverarbeitungsvorrichtung in Fig. 1 aus Richtung eines einfallenden Strahlenbündels,

Fig. 4

eine schematische Darstellung eines Beispiels für ein zu untersuchendes Wertdokument in Form einer Banknote,

Fig. 5

eine schematische Darstellung eines Farbübergangs mit Hilfe einer Dichte von Punkten,

Fig. 6

eine schematische Darstellung von Farbkoordinatenwerten von Bildelementen des Farbübergangs in Fig. 5 in einem L*a*b-Farbraum,

Fig. 7

ein stark vereinfachtes Ablaufdiagramm für ein Verfahren zur Erkennung von Farbübergängen, das in der Banknotenbearbeitungsvorrichtung in Fig. 1 durchgeführt werden kann,

Fig. 8

eine schematische Darstellung von Farbkoordinatenwerten von Bildelementen des Farbübergangs in Fig. 5 in einem RGB-Farbraum, und

Fig. 9

eine schematische Darstellung einer Referenzfläche in einem L*a*b-Farbraum für zwei Farbübergänge, denen eine Farbe gemeinsam ist.

The invention will be explained in more detail below by way of example with reference to the drawings. Show it:

Fig. 1

a schematic representation of a banknote processing apparatus,

Fig. 2

a schematic representation of an optical sensor and a Farbwertauswerteinrichtung in a control and evaluation of the banknote processing device in Fig. 1 .

Fig. 3

a schematic partial representation of three detection element rows of the banknote processing apparatus in Fig. 1 from the direction of an incident beam,

Fig. 4

a schematic representation of an example of a value document to be examined in the form of a banknote,

Fig. 5

a schematic representation of a color transition using a density of points,

Fig. 6

a schematic representation of color coordinate values of pixels of the color transition in Fig. 5 in a L * a * b color space,

Fig. 7

a highly simplified flow chart for a method for detecting color transitions, which in the banknote processing apparatus in Fig. 1 can be carried out,

Fig. 8

a schematic representation of color coordinate values of pixels of the color transition in Fig. 5 in an RGB color space, and

Fig. 9

a schematic representation of a reference surface in a L * a * b color space for two color transitions, which is a color in common.

Fig. 1 shows a device 10 for determining a state of documents of value, in the example a banknote processing device, which serves among other things for determining the state of value documents 12 in the form of banknotes. The device 10 has an input pocket 14 for input of value documents 12 to be processed, a singulator 16, which can access value documents 12 in the input tray 14, a transport device 18 with a switch 20, and after the switch 20 an output tray 26 and a Shredder 28 for the destruction of banknotes. Along a given by the transport device 18 transport path 22 a sensor assembly 24 is disposed in front of the switch 20 and after the singler 16, which serves to detect characteristics of individually supplied value documents 12 and formation of the characteristics reproducing sensor signals. A control and evaluation device 30 is connected at least to the sensor arrangement 24 and the switch 20 via signal connections and is used to evaluate sensor signals of the sensor arrangement 24 and control at least the switch 20 as a function of the result of the evaluation of the sensor signals.

The sensor arrangement 24 comprises at least one sensor for this purpose; In this exemplary embodiment, three sensors are provided, namely a first sensor 32, in the example an optical sensor for detecting color properties, which detects optical radiation reflected from the value document second sensor 34, in the example also an optical sensor for detecting special spectral security features of the value documents, which also detects the optical document remitted optical radiation, and a third sensor 36, in the example an acoustic sensor, more precisely an ultrasonic sensor, which originates from the document of value, in particular detected by a value document transmitted, ultrasonic signals detected.

During the advance transport of a value document, the sensors 32, 34 and 36 acquire, according to their function, properties of scanning ranges determined by the relative position of the sensors to the value document on the value document, wherein the corresponding sensor signals are formed. Each of the sensors can have a different spatial resolution, i. the size and distribution of the detected scan areas on the value document may vary depending on the particular sensor and the transport speed used. Each of the scanning areas is assigned a location which reproduces the position of the scanning areas for the respective sensor relative to one another and / or relative to the value document.

From the analog or digital sensor signals of the sensors 32, 34, 36, at least one characteristic of at least one scanning region and / or at least one value document characteristic are determined by the control alignment 30 in a sensor signal evaluation, which are relevant for checking the bank notes with respect to their state , Preferably, several of these properties are determined. Furthermore, the authenticity of the value documents is checked by means of the signals of the sensor 34. The value document properties characterize the state of the value documents, in this example the state of the banknotes with regard to the marketability or fitness for circulation, ie the suitability continues to be used as a means of payment to be able to. Corresponding value-document properties in this example are, in particular, the presence of soiling and / or color abrasions or stains and the presence of cracks, adhesive strips, dog-ears and / or holes, and / or the absence of components of the value documents. These value-document properties can be determined as a function of sensor signals of only one of the sensors 32 or 34 or at least two of the sensors.

For this purpose, the control and evaluation device 30 has, in addition to corresponding interfaces for the sensors, a processor 38 and a memory 40 connected to the processor 38 in which at least one computer program with program code is stored, in the execution of which the processor 38 controls the device or the sensor signals, in particular for determining a total state of a tested value document, evaluates and corresponding to the evaluation, the transport device 18 controls.

In particular, the control and evaluation device 30, or more precisely the processor 38 therein, after determining the value document properties, can check a criterion for the overall state of the value document into which at least one of the value document properties is received or which depends on at least one of the value document properties. In particular, further reference data for determining a still permissible state of the value document, which are predefined and stored in the memory 40, can enter into the criterion. The overall state can be given for example by two categories "still fit for use" or "marketable" or "to be destroyed". Depending on the determined state controls the control and evaluation device 30, in particular the processor 38 therein the transport means 18, more precisely the switch 20, so that the tested Value document according to its determined overall state for storage in the output tray 26 or for destruction to the shredder 28 is transported.

For processing documents of value 12 are separated from the singler 16 in the input tray 14 as a stack or individually inserted value documents 12 and occasionally the transport device 18 which supplies the isolated value documents 12 of the sensor assembly 24. This detects at least one property of the value documents 12, wherein sensor signals are formed, which reflect the property of the value document. The control and evaluation device 30 detects the sensor signals, determined in dependence on these a state and the authenticity of the respective value document and controls depending on the result, the switch 20 so that, for example, still usable value documents the output tray 26 and shredded documents of value to the shredder 28 are fed to destruction.

Adhesive strips on the value documents 12 can be detected, for example, by means of the sensor 36. To characterize the state of the banknotes, the control and evaluation device 30 can determine, for example, the number of adhesive strips or the total length or total area of the adhesive strips from the sensor signals of the sensor 36.

To determine the overall state of the banknotes, the control and evaluation device 30 uses the criterion already mentioned, into which at least one of the properties can enter. The individual values can be linked, for example, in a criterion, for example by means of a linear combination. The control and evaluation device 30 then compares the linear combination to determine the overall state of the banknotes the properties characterizing the state of the banknotes with a predetermined value and, for example, decides whether the state of the banknotes is good or bad, ie whether they are fit or not. This ensures that a banknote, which already has a considerable contamination and / or color wear, but which would not in itself lead to the fact that the state of the banknote would be determined as bad, is determined as bad, if the banknote in addition z. B. also has few stains and / or cracks, etc.

The sensor 32, which serves for the spatially resolved detection of colors of the value documents 12, is in the FIGS. 2 and 3 shown in more detail. It is described only briefly below, a detailed description can be found in the filed by the Applicant patent application WO2006 / 018283 , the contents of which are hereby incorporated by reference into the description. The sensor 32 is designed as a line sensor to which a document of value is transported at a constant speed in order to acquire an image. During the pre-transport, the sensor 32 acquires line images that, when combined in accordance with the acquisition order, yield a two-dimensional image of the value document.

The sensor 32 has a light source 42 for illuminating a value document 12 with optical radiation 44 in the visible wavelength range, preferably white light. In the beam path of the illumination radiation 44, an optional condenser optics (not shown in the figures) for bundling the emitted illumination radiation 44 is arranged. For detecting the optical radiation remitted by the value document 12, also referred to below as detection radiation, a color detection device 46 is provided.

The color detection device 46 has along a detection beam path provided for limiting the field of view, an entrance slit forming aperture 48 and a field of self-focusing lenses 50, from the in Fig. 2 by way of example only one row is shown, of which in turn only the outermost lens can be seen. The self-focusing lenses 50 direct the detection radiation onto a spatially spectral dispersing device 52, which decomposes the optical radiation into spectral components which propagate along different spatial directions according to their spectral composition. A detection optics (not shown in the figures for clarity) focuses the spectral components on a spatially resolving detection device 54 having a plurality of lines 56, 58 and 60 of detection elements aligned parallel to the direction of the entrance slit and corresponding to the number of colors to be detected are referred to as detector lines which detect the intensity of the spectral components along a respective line and form corresponding detection signals.

The aperture 48 arranged in the vicinity of the document of value 12 to be tested preferably forms an entrance slit with a gap width between 0.1 and 0.2 mm and a typical length corresponding to an expected width of the value documents, in the example of banknotes between 10 and 200 mm, preferably at about 100 mm.

Self-focusing lenses 50 are generally cylindrical optical elements made of a material having a refractive index that parabolically decreases from the optical axis of the cylinder toward the cladding thereof. By the use of such lenses 50 is independent of the distance between value document and image and achieved adjustment-free 1: 1 mapping of the examined portion of the value document 12 to the dispersing device 52.

As the dispersing means 52, for example, a diffractive element such as an optical grating may be used. In this embodiment, however, a prism made of crown glass with a prism angle of about 60 ° is used. The dispersing means 52 is arranged so that the spectral components are parallel to a plane that is, to a good approximation, orthogonal to the direction of the entrance slit.

For detection of the spectral components, which is spatially resolved along a line, the detection device 54, which is used to generate image data which serves the colors of a document of value examined, is used with a color evaluation device 62, i. a device for the detection of dirt and / or Farbabnutzungen in the range of color transitions according to a first preferred embodiment of the invention, connected, which is integrated in this example in the control and evaluation device 30, but this is not necessary.

The detection element lines 56, 58 and 60 are on a common, for the sake of clarity only in Fig. 3 shown carrier 63 attached.

In the direction of the lines, the extents of the detection element lines 56, 58 and 60 are constant. The width of the detection elements of a line, ie the extension in the direction of the line, and their distance in the direction of the line are in each case the same and predetermined by the required resolution, in the example about 0.2 mm for a resolution of 125 dpi.

In order to be able to obtain as directly as possible detection signals which correspond as well as possible to human color perception, the detector lines 56, 58 and 60 differ in the height h of the detection elements of the respective line, ie their extent orthogonal to the direction of the line (cf. Fig. 3 ). As a result, the detection elements of different lines receive different width spectral ranges according to their height, so that the sensitivity spectrum of the detection device 54 is influenced accordingly. For selecting the position of the spectral bands, the distances d of the detector lines 56, 58 and 60 can differ from one another. The heights of the detection elements and the distances of the detector lines in the direction of the spatial fanning of the spectral components, ie transversely to the direction of the lines are chosen so that a detection is at least approximately close to that of the color perception of man or that the detected spectrum at least approximately adapted to the color perception of the human eye.

The individual detector rows can be based on silicon, for example. The detector lines 56, 58 and 60 must have a comparatively high level for approximating the color perception of the human eye to detect spectral components from the "blue" and "infrared" spectral regions, since silicon is less sensitive to these wavelength ranges than to other wavelength ranges ,

Further approximation to human color perception can be achieved by weighting the detected spectral components depending on or independently of the geometry of the detector lines 56, 58 and 60 in the color evaluator 62. The spectral components can in particular be weighted individually depending on their intensities by means of multiplicative weighting factors, the weighting factors being dependent of the spectrum that should be approximated. In this case, for example, it is found in a silicon detector that the spectral component in the "red" spectral range has an overall intensity value list, but the value should be I _soll . Then, for calibration, the weighting factor is initially set to convert a detected intensity value to a calibrated value using the weighting factor. This adjustment is made for all spectral components to be detected in the calibration of the overall device.

Here, it is assumed that after calibration, the color evaluator 62 can generate from the detection signals of the detector lines 56, 58 and 60, image data which can be used to a good approximation as color coordinates in the standard CIE XYZ color space.

In order to acquire a color image of a value document 12, it is conveyed past the color detection device 46 at a constant speed, wherein intensity data are recorded at constant time intervals with the detection element lines 56, 58 and 60 with spatial and color resolution. The intensity data represents image data describing the properties of pixels of a line image representing the line-shaped area of the value document 12 detected by the color detector 46. By juxtaposing the line images according to the time order of detection, i. corresponding assignment of the image data, then an image of the value document is obtained with pixels.

To carry out a method for detecting color transitions on examined value documents, the device 62, which, as already explained, is integrated in the control and evaluation device 30. To detect color transitions in the memory 40 is a computer program stored in the execution by the processor 38, the below-described, in Fig. 7 illustrated method is performed. The processor 38 forms, with a corresponding software module of the computer program, an interface for the acquisition of processing data, which is not explicitly shown in the figures.

The method is based inter alia on the following principles, which are exemplified by an in Fig. 4 shown banknote 64 are described as value document 12. Banknote 64 has an image area 65 that has areas of different colors.

In particular, a section 66, on which there is a color transition from a first color, a deep purple, to a second color, a bright yellow, is located on the genuine banknote 64 of a given type of banknotes assumed to be fresh in print. In Fig. 5 Part of the color transition is shown again in using black dots to illustrate the deep purple proportions.

By decomposing this banknote-type section 66 into quadratic pixels 68 completely overlapping the section 66 without overlapping, each of the picture elements can have color coordinate values or a color point represented by them in a color space, in this example the CIE L * a * b color space color space defined as device-independent, which is linearized in relation to the perception of color differences by humans, and a corresponding location or a corresponding position on the value document 12 are assigned.

In a representation of the color points of the section 66 in the coordinate system of the color space (see. Fig. 6 ), the color points are in a very good approximation on a path 70 that connects two endpoints T ₁ and T ₂ that correspond to the colors between which the transition occurs.

In particular, a feature portion of the color space may be indicated by the distance 70, more precisely its endpoints, and a smallest distance Δe chosen such that the distance of the predetermined points from the distance is not greater than the smallest distance. The distance and the smallest distance Δe define a closed area in the color space.

By only slight and uniform soiling and / or color wear of the banknote, which still allow further use of the banknote, the detected color points will not lie within the area and in particular exactly on the route 70, but have a certain distance from the same. If the distance is only small, a human being will observe no or only a very small deviation from a fresh banknote when viewing the banknote. For quantification, for example, the criterion can be used that a detected color point does not represent a relevant deviation in the color space if its distance from the distance 70 is less than a predetermined maximum distance ΔE, in the example a maximum distance of 2. The maximum distance is smaller than Δe, which in this embodiment need not be calculated in the recognition, but plays only a role for the choice of ΔE. If the distance is greater, a distinct color deviation can be detected, so that the banknote is no longer suitable for circulation because a person would notice color deviations from freshly printed banknotes. In another embodiment, the maximum distance ΔE can also be determined as the sum of Δe and a further value which describes a minimum distance of two color points in the color space, which the observer just recognizes as being different.

In order to be able to determine whether a color point lies within the maximum distance, a color reference distribution for the color points or the corresponding color coordinate values of the given section is defined for a fresh printing document or a freshly printed banknote of a given type. This is given by a closed reference surface 72, which may be given by the maximum distance and the distance, for example by the end points and whose points in particular have exactly the maximum distance from the distance 70. Color points lying within or on the reference surface 72 correspond to the color reference distribution, others not.

It must also be stated that the picture elements corresponding to the color dots corresponding to the color reference distribution are also distributed in the section of the banknote corresponding to reference positions predetermined for the given banknote type.

If, therefore, a banknote of the given banknote type is examined and if color dots are located outside the reference area, these are due to soiling and / or color abrasions. In contrast, color points within or on the reference surface belong to a color transition in the section of a banknote of the given type, if their local distribution corresponds to the local reference distribution predetermined by the reference layers.

In order to enable a fast implementation of the method, it proves to be favorable to check for a color point only if it has a distance from the distance 70 which is smaller than the maximum distance ΔE. There are two cases to be distinguished. If a color point P _{1 is} such that the perpendicular passing through the color point intersects the line on a straight line running coaxially to the line, then the distance between the color point P ₁ and the base point of the Lots is the sought distance (compare the color point P ₁ in Fig. 6 ).

On the other hand, if the plumb does not cut the line (compare color point P ₂ in Fig. 6 ), the distance of the color point from the distance to be used is the smaller of the two distances of the color point from the end points T ₁ and T _{2 of} the distance.

wobei $$e_{21}=\frac{t_2-t_1}{\left|t_2-t_1\right|}$$

der Einheitsvektor in Richtung von t₂-t₁ ist.The distance of a point P with color space vector p from the origin of the color space or coordinate system to the point P of a distance with the end points T ₁ (with color space vector ti) and T ₂ (with color space vector t ₂ ) can be determined as the length of the following distance vector D. : $$D=p-t_1-e_{21}\min \left(\left|t_2-t_1\right|.\mathrm{<figure-callout\; id="0"\; label="Max"\; filenames="imgf0002.png,imgf0005.png"\; state="\{\{state\}\}">Max</figure-callout>}\left[0.\left(p-t_1\right)e_{21}\right]\right).$$

in which $$e_{21}=\frac{t_2-t_1}{\left|t_2-t_1\right|}$$

is the unit vector in the direction of t ₂ -t ₁ .

If ΔE denotes the maximum distance, a color point corresponds to the color reference distribution if the criterion for the maximum permissible distance D ² <(ΔE) ^{2 is} fulfilled.

In this example, for the sake of simplicity, it is assumed that a section is selected for all given types of value documents is that in this only a predetermined color transition should occur. In general, several sections can be examined; If the absence of a color transition specified for the section is then detected for at least one of the sections, the value document is rejected as unfit for use. The method in this first embodiment is as follows (see. Fig. 7 ).

In step S10, first, an image of the portion of the value document is acquired. For this purpose, the value document is moved past the color detection device 46, in particular the detection device 54, with a transport speed constant in the example; during this movement image data are recorded in dependence on the transport speed at predetermined time intervals by means of the detector lines or detection element lines 56, 58 and 60, which reproduce colors of pixels corresponding to the respective detector line, which, when arranged in the order of their acquisition, form an image of the examined section of the image Play the value document. The image data for a pixel thus comprise the color values, in the example the color coordinate values in the CIE-XYZ color space, as well as the location or the position of the pixel on the banknote.

This image data is then subjected to low-pass filtering in step S12. In the example, for each selected pixel, the values of each of the color coordinates are replaced by the average of the values of the respective color coordinates over the selected pixel and pixels at a predetermined averaging distance. Here, for example, for each selected pixel, the values of each of the color coordinates may be replaced by the average of the values of the respective color coordinate over the pixels in a square centered on the selected pixel. The side length of the square is set in units of the length of the pixel edges and is in the example at least 5 pixels. By this low-pass filtering can be achieved that fine lines, the color of which deviates from the neighboring areas of the lines are not recognized as individual color transitions. Such color transitions are irrelevant for the examination of the condition of the banknotes.

In step S14, the low pass filtered image data of the pixels in processing data from the pixels, i. identical areas of the image, corresponding pixels are transformed by the color coordinate values are transformed in the image data according to known formulas in the color space CIE L * a * b. As stated above, these color coordinate values correspond to a color point in the color space. The CIE L * a * b color space is characterized by the fact that distances between color points in this color space at least approximately reflect human-perceived, quantified differences between colors. The processing data further includes the location of the pixel on the bill.

In step S16, a type of the examined value document is determined based on the image data of the pixels by known methods; an example of such a procedure is in DE 100 45 360 A1 described. This determined type is used to specify the color reference distribution and the reference positions. For this purpose, in the memory 40 of the control and evaluation device 30 is a list of banknote types comprising different denominations for at least one currency area, with data on corresponding color reference distributions given by the color coordinate values of endpoints of links in the CIE-L * a * b color space and the maximum distance .DELTA.E, and corresponding reference positions are stored.

In step S18, it is checked for which picture elements located in the predetermined section the color coordinate values in the processing data and the corresponding color points of the color reference distribution respectively correspond to the banknote type determined in step S16.

For this purpose, as described above, the squares of the distances of the color dots from the respective distance are determined and compared with the square of the maximum distance. If a color point or the corresponding picture element is smaller than the square of the maximum distance, it is correspondingly marked as corresponding to the color reference distribution.

In step S20, it is checked whether the picture elements corresponding to the color reference distribution are at locations on the banknote or in the section corresponding to a spatial distribution predetermined by the color reference distribution, given by reference positions. The example examines in more detail which of the locations lie within a given range. Pixels corresponding to the color reference distribution whose locations do not correspond to the location distribution and pixels not corresponding to the color reference distribution are then marked as color deviations. In this embodiment, the reference positions are given by position coordinates in the section. A position of a picture element corresponds to one of the reference positions if its Euclidean distance is smaller than a predetermined maximum distance in the position space, which can be selected, for example, as a function of the expected accuracy of the orientation of the color detection device to the value document.

In step S22, it is decided on the basis of a predetermined criterion for the picture elements not corresponding to the color reference distribution whether or not the value document is to be classified as unsoiled. In this example For this purpose, the number of picture elements marked as color deviation is compared with a maximum number N given for the value document type.

If the number of deviations exceeds the maximum number N, the value document is considered to be too dirty for further use and the control and evaluation device 36 controls the switch 20 so that the value document is fed to the shredder 26. Otherwise, the control and evaluation device 36 controls the switch 20 so that the value document is transported to the tray 26.

In a second embodiment, the RGB color space is chosen as the color space instead of the CIE-L * a * b color space. In Fig. 8 is the distribution of the color space points in Fig. 6 shown color transition shown. As can easily be seen, this results in a very good approximation by a straight line or a stretch.

In the method, a transformation into the RGB color space only needs to be performed in step S14. Furthermore, data describing the color reference distribution must be changed accordingly.

All other details correspond to those of the first embodiment.

A third preferred embodiment differs from the first embodiment only in that more complicated color transitions can be checked.

An example of such a complicated color reference distribution for a more complicated color transition, here between three colors, is in Fig. 9 illustrated. The color transition corresponds to a color reference distribution that can be represented by a reference surface 74. The reference surface can itself be represented by two distances 76 and 76 'and a maximum distance ΔE. The color points of the reference surface are given by all color points whose distance from the two distances calculated by the above-described method is smaller than the maximum distance.

mit $$e_{i21}=\frac{t_{i2}-t_{i1}}{\left|t_{i2}-t_{i1}\right|}.$$

In general, the reference area can be represented by N distances with endpoints T _i1 and T _i2 , represented by vectors t _i1 and t _i2 , in the color space used, where i = 1,..., N is a natural number. The square of the distance of a color point from the distances is then given by $$D^2=\underset{i=1.....N}{\min }{\left\{p-t_{1i}-e_{i21}\min \left(\left|t_{i2}-t_{i1}\right|.\mathrm{<figure-callout\; id="0"\; label="Max"\; filenames="imgf0002.png,imgf0005.png"\; state="\{\{state\}\}">Max</figure-callout>}\left[0.\left(p-t_{i1}\right)e_{i21}\right]\right)\right\}}^2$$

With $$e_{i21}=\frac{t_{i2}-t_{i1}}{\left|t_{i2}-t_{i1}\right|},$$

This is in Fig. 8 for the case N = 2, where T ₁₂ = T ₂₁ .

The corresponding method differs from the method of the first embodiment only in that in step S18 now the last-mentioned distance determination is performed.

In other embodiments, not every pixel is assigned a picture element, so that the local resolution is reduced in the course of the low-pass filtering becomes. For example, the reduction can be made so that only half as many picture elements are formed as pixels are present.

In other embodiments, other color detection means are used, as for example in the WO2006 / 018283 are described. Furthermore, conventional color sensors with color filters can also be used.

In general, the image data can be obtained by relative movement of color detection device and value document, that is, for example, by movement of the color detection device.

In further embodiments, it is also possible to capture the image data at a time, for which purpose a corresponding color detection device is used for detecting a two-dimensional color image.

Other embodiments may differ from the above-described embodiments in that the value document type, in the example the currency and denomination of a banknote being examined, is determined by detecting geometric dimensions of the banknote and comparing with predetermined dimensions. Of course, any other method is suitable.

It is also possible that the control and evaluation device 30 is designed to read in data representing the value document type in step S16. This is useful, for example, if only value documents of a given value document type are processed.

The determination of the value document type can be completely omitted if the value-document processing device is designed or used for processing value documents of only one type.

The reference area or the data defining this can be determined as follows by means of a correspondingly programmed data processing device, which or its processor carries out the calculation steps.

First of all, the largest possible number of value documents of the same value document type, in this example banknotes of the same denomination but different degree of soiling and / or color depreciation, are specified. The degree of soiling or color wear thereby preferably varies between "new" that is to say fresh from the press and without dirt and heavily soiled and / or worn, ie. a level of fouling or color deterioration which is considered just acceptable, for example by the institute examining the value documents in circulation, such as a central bank.

Then, the steps S10 to S14 are performed for the predetermined value documents and stored the predetermined portion of a respective value document.

Thus, as processing data for each of the securities, there are color space coordinates in the CIE Lab space. The color space coordinates are to be understood as simplifying the description of the method as coordinates of a three-dimensional vector.

In the next step, a mean vector is then determined as the mean value of the vectors and stored. From the vectors corresponding to the processing data, the mean vector is subtracted so that the resulting shifted vectors have the zero vector as an average.

In the next step, the correlation matrix for the shifted processing data is determined, which is a 3x3 matrix according to the dimension of the vectors.

In the next step, the two largest eigenvalues and their respective orthogonal eigenvectors of this matrix are determined.

In the following step, the coordinates of the distance determining the reference surface are determined by first of all bringing the eigenvector to the largest eigenvalue by multiplication by a corresponding scalar factor, which is determined by the largest eigenvalue. In the present case, the scalar factor is determined so that the square of the length is twice the largest eigenvalue. Thereafter, the resulting vector is shifted so that the shift performed prior to the formation of the correlation matrix is canceled.

The coordinates of the start and end points of the resulting vector are then the coordinates of the endpoints of the searched route.

The square of the distance in the direction orthogonal to the distance is given by the second largest eigenvalue.

The coordinates of the start and end points and the value for the given distance are then stored.

Another variant of the method differs from the method just described in that instead of the correlation matrix, the covariance matrix is used. The eigenvalues must then be scaled accordingly.

The determination of the eigenvalues and eigenvectors can be obtained, for example, by means of a singular value decomposition.

However, in another embodiment, it is possible to use other, iterative methods to obtain directional data that determines the eigenvector to be the largest eigenvalue. For example, an algorithm known as NIPALS (Nonlinear Iterative Partial Least Squares) or variants thereof may be used.

The determination of the length of the distance and of the predetermined distance can then be carried out in each case in a space which corresponds to the straight line through the path or to a subarray orthogonal to the straight line which has one dimension less than the color space.

Another embodiment differs from the first embodiment in that the color space now includes a further dimension for the intensity of non-visible optical radiation in a predetermined wavelength range, in the example in the infrared. The color space therefore includes the usual CIE XYZ dimensions and another IR dimension. Each picture element is therefore assigned not only the three coordinates for the (visible) colors, but also an additional coordinate for the intensity of the IR radiation in the given area.

The described methods are now carried out in the four-dimensional space. The described transformations of the subspace for the (visible) colors can also be carried out according to the transformations of the color space in the first embodiment, wherein the additional dimension or the corresponding additional coordinate can be left invariant.

In this way, also wear of IR printing inks can be detected.

Claims (17)

1. A method for detecting soiling and/or color wear in the area of color transitions in at least one portion of a document of value (12) of a predetermined type of document of value on the basis of processing data reproducing color coordinate values of image elements (66) in a color space in dependence on the position of the areas in the portion (66) of the document of value (12) that correspond respectively to the image elements (66), and of reference data reproducing for a type of documents of value (12) a predetermined color reference distribution of color coordinate values in the color space in dependence on reference positions on a document of value (12) of the type of document of value, wherein it is determined for each of the image elements (68) whether the color coordinate values in the color space that are allocated to the image element correspond to the color reference distribution, wherein the color reference distribution is given by at least one predetermined, closed reference surface (72), given by at least one linear segment (70) predetermined for the type of document of value, and a predetermined distance of the points of the reference surface (72) from the at least one linear segment, wherein the positions of the image elements (68) whose color coordinate values were determined as corresponding to the reference distribution are compared to predetermined reference positions on the document of value, and wherein, in dependence on the result of the comparison, a presence or an absence of soiling and/or color wear in the area of a color transition is detected.
2. The method according to claim 1, wherein for checking whether color coordinate values for an image element (66) correspond to the color reference distribution it is checked whether a point corresponding to the color coordinate values is disposed inside or outside of the reference surface (72).
3. The method according to claim 1 or 2, wherein for checking whether the color coordinate values allocated to the respective image element (66) correspond to the color reference distribution, a value is determined which reproduces the distance of the point corresponding to the color coordinate values from the at least one segment (70).
4. The method according to any of the preceding claims, wherein the color space used is a device-independently defined color space.
5. The method according to any of the preceding claims, wherein the distance is given in dependence on a minimum distance of two color coordinate values in the color space for colors which can still be recognized as different by a predetermined viewer under predetermined viewing conditions.
6. The method according to any of the preceding claims, wherein the distance is predetermined in dependence on the type of document of value.
7. The method according to any of the preceding claims, wherein for generating the processing data, image data of pixels of a captured image of the portion (66) are used from which the color coordinate values and the positions can be determined, and wherein the processing data for an image element (68) are determined using a low-pass filtering of the image data.
8. The method according to any of the preceding claims, wherein for generating the processing data, image data of pixels of a captured image of the portion (66) are used from which the color coordinate values and the positions can be determined, and wherein the processing data for an image element (68) are determined using the image data of at least two pixels.
9. The method according to any of the preceding claims, wherein the color space has, in addition to dimensions for the visible colors, a further dimension for non-visible optical radiation in a predetermined wavelength range, the image elements comprise data relating to the non-visible optical radiation in the predetermined wavelength range, and the reference surface is given by at least one linear segment predetermined for the type of document of value and a predetermined distance of the points of the reference surface from the at least one linear segment.
10. An apparatus for detecting soiling and/ or color wear in the area of color transitions in at least one portion (66) of a document of value (12) of a predetermined type of document of value having at least one interface for capturing processing data reproducing color coordinate values of image elements (68) in a color space in dependence on the position of the areas in the portion (66) of the document of value (12) that correspond respectively to the image elements (68), and an evaluation device (62) configured to carry out a method according to any of the preceding claims on the basis of processing data captured via the at least one interface and of reference data reproducing for a type of document of value (12) a predetermined color reference distribution of color coordinate values in the color space in dependence on reference positions on a document of value (12) of the type of document of value, and in particular to determine for each of the image elements (68) whether the color coordinate values in the color space that are allocated to the image element (68) correspond to the color reference distribution, wherein the color reference distribution is given by a predetermined closed reference surface (72) in the color space, which is given by at least one linear segment (70) predetermined for the type of document of value and a predetermined distance of the points of the reference surface (72) from the at least one linear segment, to compare the positions of the image elements (68) whose color coordinate values were determined as corresponding to the reference distribution with predetermined reference positions on the document of value (12), and to detect, in dependence on the result of the comparison, a presence or an absence of soiling and/or color wear in the area of a color transition.
11. The apparatus according to claim 10, in which the evaluation device (62) comprises at least one processor (38) and a memory (40) which stores a computer program according to claim 12 which is executable by the processor (48) or a different program for carrying out the method according to one of claims 1 to 9.
12. A computer program for detecting soiling and/or color wear in the area of color transitions in at least one portion of a document of value (12) of a predetermined type of document of value, having instructions for at least one processor (38), upon whose execution the processor (38) executes the method according to any of claims 1 to 9.
13. A method for determining a reference surface for a method according to one of claims 1 to 9, wherein for a predetermined number of documents of value of a predetermined type of document of value, respectively allocated processing data are determined which reproduce color coordinate values of image elements (66) in a color space in dependence on the position of the areas in the portion (66) of the document of value (12) that correspond respectively to the image elements (66), wherein first components reproducing average values over the color coordinates, and components reproducing correlations between the color coordinates, of a matrix in the color space or a different color space are determined from the processing data, wherein direction data are determined which specify the eigenvectors associated with the two largest eigenvalues of the matrix, wherein data defining the linear segment are determined from the direction data in such a fashion that the middle of the segment is specified by the average values, the direction of the segment is specified by the direction of the first eigenvector, and wherein the length of the segment and the predetermined distance from the segment is determined in dependence on the largest eigenvalue or second-largest eigenvalue, or a parameter reproducing the correlation along the segment and the largest correlation in a sub-space orthogonal to the segment.
14. The method according to claim 13, wherein the components of the matrix that are determined are the components of the correlation matrix.
15. The method according to claim 13, wherein the components of the matrix that are determined are the components of the covariance matrix.
16. The method according to one of claims 13 to 15, wherein the processing data used for determining the components reproducing the average values, and components reproducing correlations between color coordinates, are color coordinates in a Lab color space.
17. The method according to one of claims 13 to 15, wherein the color space has, in addition to dimensions for the visible colors, a further dimension for non-visible optical radiation in a predetermined wavelength range, the image elements comprise data relating to the non-visible optical radiation in the predetermined wavelength range, and the reference surface is given by at least one linear segment predetermined for the type of document of value and a predetermined distance of the points of the reference surface from the at least one linear segment.

Images