GB2518107A

GB2518107A - Method and apparatus for checking valuable documents

Info

Publication number: GB2518107A
Application number: GB1500503.6A
Authority: GB
Inventors: Shanchuan Su
Original assignee: Giesecke and Devrient GmbH
Current assignee: Giesecke and Devrient GmbH
Priority date: 2012-08-24
Filing date: 2013-08-14
Publication date: 2015-03-11
Anticipated expiration: 2033-08-14
Also published as: GB201500503D0; GB2518107B; WO2014029476A1; DE102012016828A1

Abstract

The invention describes a method for checking valuable documents, in particular also soiled valuable documents, for authenticity, having the steps of acquiring pixel data relating to pixels in an image of the valuable document, determining pixels in one or at least two predefined examination areas of the image and determining, for the examination areas, error pixels among the pixels in the particular examination area, wherein the pixel data relating to the error pixels are each outside a predefined pixel-dependent and/or location-dependent authenticity range, checking whether or not the error pixels in the examination area(s) correspond to soiling, for example staining, on the valuable document, and assigning an authenticity class from predefined authenticity classes to the valuable document on the basis of the checking result.

Description

Method and apparatus for checking valuable documents 100011 The present invention relates to a method for checking value documents, in particular also soiled value documents, in particular for authenticity, and to means for carrying out said method.

100021 Value documents are understood here to be sheet-shaped objects that represent for example a monetary value or an authorization and thus should not be manufacturable arbitrarily by unauthorized persons. They hence have features or security features that are not easy to manufacture, in particular to copy, and whose presence is an indication of authenticity, i.e. manufacture by an authorized body.

Important examples of such value documents are coupons, vouchers, checks and in particular bank notes.

100031 Such value documents usually have a stipulated printed image in the visible wavelength region, with such a printed image frequently being colored. Some value documents are additionally equipped with an JR printed image, i.e. a printed image that is recognizable in the infrared wavelength region. Such an IR printed image need not match the visible printed image but can instead deviate therefrom, depending on the value-document type.

100041 Examinations of the printed image, in particular of the visible and/or of the IR printed image, if any, can give indications of the presence of a forged value document.

10005] However, such examinations on value documents, in particular bank notes, are impeded by soiling. Deviations in the printed image can be due to forgery or to soiling. Such soiling may involve for example a stain or writing done with a utensil such as a felt pen or pencil. Hereinafter soiling will be understood to refer to changes present on the value document in a non-uniform manner that have not arisen upon manufacture of the value document and lead to deviations in at least one printed image of the value document from a comTespondmg printed image stipulated for the value-document type of the value document. Value-document types are given within the framework of the present invention by value documents of the same value-document type having the same properties at least with regard to the printed images, preferably also with regard to other features. \Vith value documents in the form of bank notes, the value-document types can be given in particular by the currency and the value of the bank note, as well as the issue of the batik note, where applicable.

10006] The present invention is hence based on the object of providing a method for examining value documents, in particular also soiled value documents, which allows a recognition of value documents that are suspect, or also ones that are forged, where applicable. Further, means are to be provided for carrying out said method.

100071 The object is achieved by a method for checking value documents, in particular also soiled value documents, for authenticity, having the steps of capturing pixel data of pixels of an image of the value document, establishing pixels in one or at least two stipulated examination regions of the image, and establishing, for the one or at least two examination regions, error pixels among the pixels in the examination region or a respective one of the at least two examination regions, wherein the pixel data of the error pixels respectively lie outside a stipulated pixel-and/or location-dependent authenticity region, checking whether or not the error pixels in the examination region or regions correspond to soiling, for example a stain, on the value document, and assigning an authenticity class selected from stipulated authenticity classes to the value document in dependence on the result of the check. Preferably, there is formed an authenticity signal designating the assigned authenticity class, and/or there is preferably stored a datum designating the assigned authenticity class.

The method is carried out employing an evaluation device which can have at least one processor that can be programmed accordingly or that executes a corresponding program stored in the evaluation device for carrying out the method.

100081 The object is further achieved by a device for checking value documents, in particular also soiled value documents, for authenticity, having an optical sensor for capturing an image of the value document, wherein the image has pixels having pixel data assigned thereto, and an evaluation device which is arranged for canying out a method according to the invention. In particular, the evaluation device is configured for capturing pixel data of pixels of an image of the value document by means of the sensor, for establishing pixels in one or at least two stipulated examination regions of the image, and for establishing, for the one or at least two examination regions, error pixels among the pixels in the examination region or a respective one of the at least two examination regions, wherein the pixel data of the error pixels respectively lie outside a stipulated pixel-and/or location-dependent authenticity region, for checking svhether or not the error pixels in the examination region or regions correspond to soiling, for example a stain, on the value document, and for assigning an authenticity class selected from stipulated authenticity classes to the value document in dependence on the result of the check. The device for checking, hereinafter also designated as a checking device, is thus configured in particular for carrying out the method according to the invention. Preferably, the evaluation device is configured for forming an authenticity signal designating the assigned authenticity class and/or preferably for stonng a datum designating the assigned authenticity class.

10009] The invention relates to the check of value documents. In the check according to the invention, authenticity classes can be assigned to value documents when said value documents have a stipulated value-docmnent type or one of a plurality of stipulated value-document types.

10010] It is provided to first capture pixel data of pixels of an image of the value document, which can be done by reading in via an interface and preferably by means of the optical sensor. The image may be an image in the visible wavelength region, for example a strictly gray-scale image or also a colored image, with the sensor of the checking device then being configured for capturing the corresponding image.

Preferably, the image of the value document is an IR image of the value document, however. In the checking device, the optical sensor is then preferably so configured that an IR image of the value document can be captured by means of the JR sensor.

The image can then be an image in a stipulated wavelength region in the infrared wavelength region, said image comprising only intensity information or lightness information in said wavelength region. Spectral properties then play no part.

100111 In the usual way, the image is constructed of pixels whose properties are described by the pixel data. The pixel data can comprise a different number of components, depending on the type of the image; for example, the pixel data can respectively describe at least one color coordinate in a color space and/or a lightness for the pixels. In the ease of a pay-scale image in the visible wavelength region or an IR image, the pixel data can preferably describe a lightness of the pixels, while in the case of a colored image, the pixel data can be color coordinates for the pixels in a color space. The optical sensor of the checking device is configured for capturing colTesponding images and fonning colTesponding pixel data, i.e. it can be a gray-scale sensor working in the visible region, a color sensor working in the visible region, or a sensor working in the JR region.

100121 In the captured image there are now established pixels that lie in a stipulated examination region or at least two stipulated examination regions. The examination region or regions can be stipulated in dependence on the value-document type and preferably on the orientation of the value document, depending on the value-document type. The employment of at least two examination regions has the advantage of enabling a more differentiated check, since the check can respectively be effected in the examination regions in dependence OB the respective examination region or the properties of the value document in the respective examination region. The orientation of the value document, assuming a rectangular value document, is understood to be one of the four possible positions that can be obtained by respectively rotating the value document around its longitudinal and transverse axes by 1800. The examination region or regions are preferably chosen such that a good and reliable recognition of forged or suspect value documents is obtained. The examination regions preferably have respectively an area of at least 4 mm2. If only one examination region is employed, its area preferably amounts to more than 30% of the total area of the value document, and if a plurality of exammatioli regions are employed, their total area preferably amounts to more than 30% of the total area of the value document.

10013] For the examination region or regions, error pixels are established among the pixels in the respective examination region. For this purpose there is employed for the examination region or each of the at least two examination regions a stipulated authenticity region, preferably an authenticity region stipulated for a respective value-document type and/or a stipulated onentation of the value document, within which the pixel data must lie if the value document is to be regarded as authentic or neither as suspect nor as forged. The authenticity region is respectively preferably location-and/or pixel-dependent, i.e. the authenticity region can be given in another way or differently for different pixels of the image or locations in the image or on the value doctunent; but at the same time the authenticity region or regions can be the same for certain different pixels, depending on the value-document type. For example, in the case of one-dimensional pixel data, i.e. pixel data having only one component, the authenticity region can be given by an interval, whose limits can depend on the respective pixel or location. The size of the authenticity region can bc established for example by measurements on stipulated value documents, preferably value documents of the value-document type. Pixels whose pixel data will lie outside the respective authenticity region are established as enor pixels. Said pixels or their pixel data or pointers thereto can be stored separately or only marked as error pixels or pixel data of error pixels. The establishment of pixels in the examination region or regions and the establishment of the error pixels can be carried out in arbitraiy order and/or in one step. As a result there can thus be obtained an error image of the examination region or regions that shows the pixels whose pixel data lie outside the respective authenticity region.

100141 It is then checked whether or not the established error pixels in the examination region or at least two examination regions correspond to soiling, for example a stain, on the value document. For this purpose there can be employed at least one stipulated soiling criterion by means of which it is fixed when error pixels in

S

the examination region or regions are accordingly classified as soiling and when they are not. It can then thus be checked whether or not the established error pixels in the one or at least two examination regions correspond to soiling, for example a stain, on the value document according to at least one soiling critenon. It thus constitutes a cntenon for when error pixels in the examination region or regions represent a value-document region, corresponding to the respective examination region, with soiling, for example at least one stain. The check of whether or not the established error pixels in the examination region or regions correspond to soiling, iii particular a stain, on the value document can comprise a plurality of different checking substeps. The soiling criterion can then comprise subcriteria which must be met alternatively or cumulatively, depending on the embodiment, for a presence of soiling to be recognized. The soiling criterion or a subcriterion can in particular also be a criterion for the error pixels not being due to soiling or corresponding to soiling; in this case the soiling criterion or subcriterion would be met when the criterion is not met. The soiling criterion or subcriterion can also be stipulated in dependence on the examination region.

100151 This procedure exploits the fact that, in many cases, error pixels due to forgery have properties and/or a local distribution that are different from error pixels due to soiling, in particular in the form of stains.

100161 In dependence on the result of the check, one of at least two stipulated authenticity classes is assigned to the respective value document or its image. The authenticity classes can comprise for example the classes "recognized as authentic" and "forged", the classes "not suspect" and "suspect", or the classes "recognized as authentic" and "suspect". Which authenticity classes are stipulated can depend on whether the result of the check of the value document is linked with the results of other authenticity checks of the value document in order to obtain a total classification.

100171 In the method or the device, the authenticity class assigned can be in particular the class "suspect" when the check of whether or not the error pixels hi the examination region or regions correspond to soiling, for example a stain, on the value document has yielded that the error pixels do not correspond to soiling.

100181 However, it is also possible to assign the authenticity class "false", if this is provided, instead of the authenticity class "suspect".

100191 There can then preferably be formed an authenticity signal designating or identifying the result of the check, in particular the assignment to the authenticity class or the assigned authenticity class, or there can be stored a datum designating or identifying the assignment to the authenticity class or the assigned authenticity class.

In the checking device, the evaluation device can be arranged for forming an authenticity signal designating or identifying the result of the check, in particular the assignment to the authenticity class or the assigned authenticity class, or for stonng a datum designating or identifying the assignment to the authenticity class or the assigned authenticity class. The authenticity signal can be stored or further processed directly.

100201 To enable a good check for the presence of soiling, it is preferred, in the method, that in the step of establishing the error pixels or in the step of checking for the presence of soiling, i.e. whether or not the error pixels in the examination region or regions coiTespond to soiling, for example a stain, on the value document, the error pixels are divided into upper error pixels, whose pixel data lie outside the authenticity region in a first direction, and lower error pixels, whose pixel data lie outside the authenticity region in a second direction opposite to the first direction. In the checking device, the evaluation device is then preferably configured for, in the step of establishing the error pixels or in the step of checking for the presence of soiling, i.e. whether or not the error pixels in the examination region or regions correspond to soiling, for example a stain, on the value document, dividing the error pixels into upper error pixels, whose pixel data lie outside the authenticity region in a first direction, and lower error pixels, whose pixel data lie outside the authenticity region in a second direction opposite to the first direction. Said division can be realized for example by assigning to the error pixels or their pixel data a corresponding identifier which enables a distinction between upper and lower error pixels, or by storing upper error pixels and lower error pixels or their pixel data separately from each other, or by keeping lists for upper and lower error pixels in which pointers to the respective error pixels or pixel data of the respective error pixels are stored.

100211 Particularly preferably, in the method, the authenticity region can be one-dimensional. The upper error pixels can then be error pixels whose pixel data lie above the authenticity region, and the lower error pixels can be error pixels whose pixel data lie below the authenticity region. The pixel data are then likewise one-dimensional.

This embodiment is suited in particular for IR images which only have lightness information, i.e. whose pixels have lightness values as pixel data.

100221 The embodiments stated in the two preceding paragraphs have the advantage that a check for the presence of soiling can be effected in a more differentiated manner than without such a division.

100231 Preferably, in the method, in the checking step for the one or at least two examination regions, it is checked respectively in a first region check whether the upper error pixels correspond to at least one instance of soiling, for example at least one stain, and/or checked respectively in a second region check whether the lower error pixels correspond to at least one instance of soiling. It can then be decided that the error pixels in the examination region or regions correspond to soiling when at least a stipulated minimum number of region checks has yielded that the upper or lower error pixels in the respective examination region correspond to at least one instance of soiling, for example at least one stain, on the value document. The stipulated minimum number can in particular also be twice the number of examination regions, so that a presence of soiling is then only recognized when all region checks have yielded this. In an alternative embodiment, it can be decided that error pixels in the one or at least two examination regions do not correspond to at least one instance of soiling, for example at least one stain, on the value document when at least a stipulated minimum nmnber of region checks has yielded that the upper or lower error pixels in the respective examination region do not correspond to at least one instance of soiling, for example at least one starn, on the value document. The stipulated mhuimuni number can be chosen to be the same or different with respect to the fir st embodiment. In particular, the minimum number can be equal to I. 100241 The order of region checks is arbitrary, it being possible for example to carry out, for the examination region or at least two examination regions, the second region check first and the fir st region check thereafter, or the second region check first and the first region check thereafter. However, it is also possible to carry out for example the second region checks for all examination regions first and the first region checks for all examination regions thereafter. The first and second region checks can also be interchanged. Further, it is also conceivable to carry them out in parallel.

10025] The minimmn numbers can be chosen in dependence on the value-document types, and on a sensor employed for capturing the images, and on the desired strength of the check.

10026] The region checks can comprise in particular one or more of the hereinafter-mentioned checks alternatively or cumulatively. In said checks there can be recognized an indication of the presence of soiling or an indication of the presence of no soiling.

When only one of the checks is employed, the indication of a presence of soiling is recognized as the presence of soiling. \Vhen a plurality of the checks are employed cumulatively, the presence of soiling can be recognized in the respective examination region or the corresponding portion of the value document when a stipulated number of checks, preferably all employed checks, have yielded an indication of the presence of soiling. There can also be recognized the presence of no soiling in the respective examination region or the corresponding portion of the value document when at least one of the employed checks or a stipulated number of the employed checks has yielded an indication of the presence of no soiling.

10027] According to a first preferred possibility, in the method, in the first region check for the examination region or one of the at least two examination regions the nmnber of the upper error pixels in the respective examination region can be compared with a stipulated fir st maximum value, and/or in the second region check for the examination region or one of the at least two examination regions the number of the lower error pixels can be compared with a stipulated second maximum value, and an indication of soiling be recognized in the respective examination region when the nmnbers do not exceed the respective maxinium value, or an indication of no soiling be recognized in the respective examination region on the value document when the first and/or second numbers exceed the first or second maximum value. The maximum mnnber here is considerably smaller than the number of the pixels in the respective examination region, preferably smaller than 80% of the number of the pixels in the respective examination region. This check offers the advantage of being especially simple to carry out.

100281 For the above-descnbed possibility and in particular for the hereinafter described possibilities it is advantageous, in the method, when upon establishing or upon checking, distances of the pixel data from the authenticity regions are established for the upper and lower error pixels, and the distances are employed upon checking the region criteria. This offers the advantage that, when printed images have strong variations of the pixel data due to the printed image, the deviations can be employed for examination more easily.

100291 A second preferred possibility provides for checking in the first region check for the examination region or one of the at least two examination regions whether a distribution of the distances of the upper error pixels satisfies an upper distribution criterion stipulated for the respective examination region, and/or in the second region check whether a distribution of the distances of the lower error pixels satisfies a lower distribution criterion stipulated for the respective examination region for the presence of soiling. The upper and lower distribution criteria are stipulated for the examination region or respectively the examination regions; the distribution criteria can still be stipulated in dependence on the value-document type and preferably on the orientation, depending on the value-document type. The distribution criteria can differ or preferably be chosen to be the same. Furthermore, the distribution criteria can differ or be the same for different examination regions.

100301 The distribution criterion can in principle be selected arbitrarily. For example, there can be stipulated a distribution function describing the frequency of the occurrence of distances and it can be checked whether the distribution of the distances satisfies the distribution function. To allow a fast but sufficiently exact check, however, it is preferred that when checking the upper or lower distribution criterion it is respectively checked whether the width of the distribution of the respective distances exceeds or undershoots a stipulated upper or lower threshold value. For the width of the distribution there can be employed any suitable quantity that is a measure of the width of the distribution, for example the full width at half maximum (FWHM) or the variance or coefficient of variance (standard deviation/average). The upper and lower threshold values can be chosen to be the same or different, depending on the value-document type. Preferably, an indication of the presence of soiling is recognized when the respective threshold value is exceeded. The threshold values can be established by analyzing value documents of the stipulated value-document type. This embodiment is based on the observation that the deviations in a forgery are due to manufacture and hence systematic, which should yield a very narrow distribution of the distances for the examination region, in particular in the case of JR images.

100311 According to a third preferred possibility, in the method, there can be established in the first region check for the examination region or one of the at least two stipulated examination regions a correlation of the upper error pixels in the examination region with a stipulated reference image of the examination region, and/or in the second region check for one of the stipulated examination regions a correlation of the lower error pixels in the examination region with a stipulated reference image of the examination region. The reference image can be chosen differently depending on the value-document type. The reference image can be for example an image of the same type as the image captured for the check, i.e. in the case of a colored image the reference image can likewise be a colored image, in the case of an JR image an IR

II

image. Preferably, there is employed upon employment an JR image as a reference image the printed image in the visible region as a lightness image. The reference images can preferably be obtained by analyzing a plurality of authentic value documents of the respective value-document type, for example as the average. The colTelation is understood here to be a local or pixel-wise correlation of the respective error pixels with the reference image or corresponding pixels of the reference image.

There can be employed the correlation of the pixel data or of the distances of the respective error pixels with the pixel data of pixels of the reference image which are disposed in the same way relative to each other as the error pixels. Preferably, a set of displacement vectors can be stipulated and the correlation can be the greatest colTelation of the local or pixel-wise correlation of the pixel data or of the distances of the error pixels with the pixel data of the corresponding pixels of the reference image or of the corresponding pixels of the reference image which is displaced by a respective one of the stipulated displacement vectors. This is to say, it is checked how the pixel data or distances of the error pixels are correlated with the pixel data of pixels of the reference image that correspond to the error pixels in their position, except for a displacement, where applicable. Preferably, an indication of the presence of soiling is recognized in the first or second region checks when the correlations are smaller than upper or lower maximmn correlation values stipulated for the respective examination region. As a measure of the correlation there can be employed for example a cross-correlation of the deviations or of the pixel data of the error pixels with the pixel data of corresponding pixels of the reference image. Preferably, there is employed a maximum cross-correlation which is obtainable by stipulating a set of displacement vectors and establishing for the displacement vectors the greatest cross-colTelation between the deviations or of the pixel data of the error pixels with the pixel data of corresponding pixels of the reference image which are displaced by the respective displacement vector. Thus, an indication of the presence of soiling can preferably be recognized when the correlation, preferably the cross-correlation or the maximum cross-correlation, is smaller than a first or second correlation threshold value. The displacement vectors can be vectors on a grid given by the pixels of the captured image or of the reference image, the length thereof preferably being smaller than a stipulated maximum distance. Instead of the cross-correlation there can also be employed an absolute or average square error, with a small error corresponding to a high correlation.

100321 When at least two of the possibilities are employed cumulatively for at least one of the examination regions, the presence of soiling is preferably recognized when either the checks for the upper error pixels yield an indication of the presence of soiling or the checks for the lower error pixels yield an indication of soiling.

100331 Independently of the above-described embodiments employing the distinction between upper and lower error pixels, it may be provided in a preferred embodiment of the method that in the step of checking whether the error pixels in the examination region or at least two examination regions correspond to soiling, it is checked whether the number of the error pixels is smaller than a threshold value stipulated for the respective examination region, and when this is the case, preferably for the examination region or at least one of the examination regions, a presence of soiling is recognized or an indication of the presence of soiling is recognized. The check here can comprise all error pixels in one of the examination regions indiscriminately. The threshold value can be chosen here as in the above-described embodiments.

100341 Further subject matter of the invention is a method for checking value documents, in particular soiled value documents, having a portion with an IR-absorbent security feature, preferably a security thread, for authenticity, having the steps of capturing pixel data of pixels of an IR image of at least the portion of the value document, establishing authenticity pixels among the pixels in the JR image of the portion, wherein the pixel data of the authenticity pixels lie above a minimum threshold value, comparing the number of the authenticity pixels with a minimum number stipulated for the portion, and assigning an authenticity class selected from stipulated authenticity classes to the value document in dependence on the result of comparing the number of the authenticity pixels with the minimum number. The Ii minimum tlueshold value is preferably chosen so as to prevent pixel data corresponding only to noise of the sensor employed for capturing the JR image from being recognized as authenticity pixels. The minimum number is preferably so chosen that it corresponds to the number of the pixels in the portion of the value document that results for an authentic, freshly printed value document. For example, the minimum number can be greater than 85% of the number of the pixels in the portion of the value document that results for an authentic, freshly printed value document. This procedure has the advantage that the exact position of the security feature need not be known as long as said feature lies in the portion to be suitably chosen.

10035] The evaluation device of the checking device can in principle be of arbitrary configuration, but preferably possesses a data processing device having at least one processor and a memory storing program code upon whose execution the method according to the invention is carried out.

10036] The subject mailer of the present invention is hence also a computer program for execution by means of a data processing device having a processor, containing program code upon whose execution the processor executes the method according to the invention, in particular according to one or more of the above-described embodiments.

10037] Further subject mailer of the present invention is a computer-readable data carrier on which such a computer program is stored. The computer program is preferably stored on the data carrier not only in volatile form; the latter thus constitutes a non-volatile data memory or comprises such a memory.

10038] The checking device according to the invention can preferably be used in apparatuses for processing vaJue documents, in particular also soiled value documents.

The subject mailer of the present invention is hence also an apparatus for processing value documents, in particular also soiled value documents, having a feeding device for value documents to be processed, an output device for value documents which has at least two output portions for receiving processed value docmnents, a transport device for transporting singled value documents from the feeding device along a transport path to the output device, a checking device according to the invention, in particular according to one or more of the above-described embodiments, which is so disposed that the transport path extends through the capture region thereof, and a control device which actuates the transport device hi dependence on an authenticity signal of the checking apparatus for a value document transported by the transport device, such that the value document is transported into an output portion of the at least two output portions that con-esponds to the authenticity signal. Preferably, the apparatus is configured for processing value documents at a speed of more than 15 value documents per second.

100391 The invention will hereinafter be explained further by way of example with reference to the thawing s. There are shown: Fig. 1 a schematic view of a value-document processing apparatus, Fig. 2 a schematic representation of a digital colored image and of an JR image of a value document, Fig. 3 a flowchart for a first method for checking value documents by means of the apparatus in Fig. 1, Fig. 4 a schematic representation of authenticity regions and pixel data for the method hi Fig. 3, Fig. 5 aflowchartforastep 518 ofthemethodinFig. 3, Figs. 6A, B flowcharts for first and second region checks in step SI 8 in Fig. 5, Fig. 7 a flowchart for a second method for checking value documents by means of the apparatus hi Fig. 1, Fig. 8 a flowchart for a step S 18' of the method in Fig. 7, Figs. 9A, B flowcharts for first and second region checks in step S 18 in Fig. 8, Fig. 10 distance distributions for an examination region of an authentic, soiled bank note (a) and for the same examination region of a forged bank note (b), and Fig. 1! a flowchart for a further method for checking value documents by means of the apparatus in Fig. 1.

100401 An apparatus 10 for processing value documents 12, in the example a bank-note processing apparatus, in Fig. I serves, in/er a/ia, for checking the authenticity of value docmnents 12 in the fonn of bank notes and for sorting the value documents in dependence on the result of the authenticity check. The apparatus 10 has a feeding device 14 for feeding or inputting value documents 12 to be processed and a transport device 20, leading to an output device 18, for singly tTansporting value documents from the feeding device 14 along a transport path 22 to the output device 18. The feeding device 14 comprises in this example an input pocket 15 and a singler 16 which can access value documents 12 in the input pocket 14 in order to single them. In the example, the transport device 20 transports value documents 12 singled by the singler 16. The output device 18 serves for receiving and outputting checked value documents and has for this purpose at least two, in this example three, output regions 24, 24' and 24, in the example stack pockets; for elective transport into the output regions the transport device 20 comprises gates disposed successively along the transport path 22, hi the example two gates 26 and 26', so that value docmnents are feedable electively to one of the output regions by actuation of the gates. Along the transport path 22 given by the transport device 20 there is disposed after the feeding device 14, in the example in particular after the singler 16, and before the gates 26 and 26' a sensor assembly 28 which serves for capturing properties of singly fed value documents 12 and forming sensor signals rendering said properties. A control device 30 is connected at least to the sensor assembly 28 and the gates 26 and 26' via signal connections and serves for evaluating sensor signals of the sensor assembly 28, in particular for checking the authenticity of value documents, and for actuating at least the gates 26 and 26' in dependence on the result of evaluating the sensor signals, i.e. in particular the results of the check for authenticity.

10041] The sensor assembly 28 comprises at least one sensor; in this exemplary embodiment there are provided an optical sensor 32 for spatially resolved capture of color properties, i.e. of a colored image, and an optical sensor 33 for capturing IR properties, i.e. an IR image, which captures optical radiation in the visible wavelength region remitted by the value document and in particular JR radiation in a stipulated wavelength region. While a value document is being transported past, the optical sensor 32 captures a digital colored image of the value document in the three color channels, red, green and blue, and the optical sensor 33 a digital JR image in the thfrared spectral region; the images are respectively represented and described by corresponding sensor signals and pixel data produced therefrom.

100421 The digital images captured by the optical sensors 32 and 33 respectively comprise pixels whose properties are given by the pixel data which represent color properties or IR properties.

100431 For evaluating the sensor signals, the control device 30 has an evaluation device 34 which is integrated into the control device 30 in this example, but may also be part of the sensor assembly 28 in other exemplary embodiments.

10044] The control device 30 has, besides corresponding interfaces 36 for the sensor device 28 or the sensors 32 and 33 for receiving the digital images captured thereby, a processor 38 connected to the interfaces 36, and a memory 40 connected to the processor 38 and storing at least one computer program with program code upon whose execution the processor 38 in a first function as part of the evaluation device 34 evaluates the sensor signals, in particular for checking the authenticity of a captured value document, and thereby, inter a/ia, executes a hereinafter described method while employing the sensor signals or the pixel data. The evaluation device 34 also includes the interfaces 36. In a second function, the processor actuates the apparatus or, in accordance with the evaluation, the transport device 20. The evaluation device 34 hence forms a computer or a data processing device as intended by the present invention.

10045] During operation, the evaluation device 34 checks for each value document captured by the sensor assembly 28 at least one stipulated criterion for the authenticity of the value docmnent by means of the processor 38 on the basis of the digital images of the value document, and establishes one of stipulated authenticity classes for the value document.

10046] In dependence on the established authenticity class, the control device 30, in particular the processor 38 therein, actuates the transport device 20, more precisely the gates, such that the checked value document is transported into corresponding output regions to be deposited in accordance with its established authenticity class. For example, there can be stipulated as authenticity classes the classes "suspect" and "forged" and "recognized as authentic"; the authenticity class "recognized as authentic" can be assigned to those value documents that cannot be assigned to either of the other two authenticity classes. Then the value documents recognized as authentic can be output into the region 24, the value documents recognized as suspect into the region 24', and the value documents recognized as non-authentic or forged into the region 24".

10047] For value documents 12 to be processed, value documents 12 inserted as a stack or individually into the feeding device 14, more precisely the input pocket 15, are singled by the singler 16 and fed singly to the transport device 18, which feeds the singled value documents 12 to the sensor assembly 28. The latter captures optical properties of the value documents 12, in the example the colored image and the IR image, and forms corresponding digital images whose pixels or pixel data render the corresponding properties of the value document. The control device 30 captures the sensor signals, establishes in dependence thereon the authenticity of the respective value document, and actuates the gates in dependence on the result such that the examined value documents are fed to the output pockets in accordance with their established authenticity.

100481 The optical sensor 32 is configured for capturing images for three colors, or colored images, and the optical sensor 33 for capturing images for IR radiation in a stipulated wavelength region for infrared radiation, i.e. JR images, iii remission.

10049] In the example, said sensors are respectively configured as line sensors which, while a value document is being transported past the sensors 32 or 33, capture a sequence of line images which yield a line image of the value document in a direction transverse to the direction of the line.

10050] The sensor 32 comprises in the present example an illumination device 42, shown only very schematically in Fig. 1, for illuminating a strip, extending transversely to the transport direction I, for the value document 12 or in a plane of the value document t2 with convergent white light in the stipulated wavelength region, while the value document is being transported past, over the total extension thereof transversely to the transport direction 1. Analogously, the sensor 33 comprises in the present example an illumination device 43, shown only very schematically in Fig. 1, for illuminating a strip, extending transversely to the transport direction T, for the value document 12 or in a plane of the value document J2 with JR radiation in the stipulated wavelength region, while the value document is being transported past, over the total extension thereof transversely to the transport direction T. In other exemplary embodiments, the sensors 32 and 33 can also comprise a common illumination device which emits both white light and the desired JR radiation.

100511 Further, the sensor 32 comprises a capture device 44 for capturing illumination radiation which is remitted by a value-document region illuminated by the illumination device 42. Said capture device 44 has, in the example, three line-scan cameras, not specifically shown, having red, green, blue filters, likewise not shown, disposed in the ray path before said cameras, for capturing red, green and blue fractions of the optical radiation of the illumination device 42 which is remitted by the value document. Analogously, the sensor 33 comprises a capture device 45 for capturing illumination radiation which is remitted by a value-document region illuminated by the illumination device 43. Said capture device 43 has, in the example, a line-scan camera, not specifically shown, having an JR filter disposed therebefore for IR fractions of the optical radiation of the illumination device 45 which is remitted by the value document.

10052] Each of the line-scan cameras has a detector row with photodetection elements iii row-type arrangement, before which there is respectively disposed the filter that corresponds to that color fraction or JR fraction of the remitted optical radiation that is to be detected by the respective line-scan camera. The sensors 32 or 33 can also comprise further optical elements, in particular for imaging or focusing, which are not shown here. The detector rows of photodetection elements are disposed parallel to each other. Furthermore, the sensors 32 and 33 comprise signal processing devices, not shown in the figures, which produce corresponding digital images from the signals of the photodetection elements.

100531 For a colored image or an IR image of a value document 12 to be captured, said document is transported past the sensors 32 and 33 in transport direction T at constant speed, whereby intensity data are captured in a spatially resolved maimer for the four channels, red, green, blue and IR, at constant time intervals with the capture devices 44 or 45, in the example the line-scan cameras. The intensity data or color coordinates, established therefrom, in a stipulated color space or lightness values constitute pixel data describing the properties of pixels of a colored or JR line image that render the line-shaped region, captured by the sensor 32 or sensor 33, of the value document 12. Joining the line images together in accordance with the time sequence of capture, i.e. corresponding assignment of the pixel data, then obtains the digital colored and IR images of the value document with pixels having respectively assigned thereto pixel data describing or representing optical properties of the value document, namely color values for red, green and blue and a lightness in the stipulated IR wavelength region.

10054] A digital image captured by the sensor 32 or the sensor 33 is hence assembled from pixels disposed in a rectangular matrix and is described by the pixel data. The pixel data for the colored image are color coordinates in a suitable color space, the pixel data for the JR image are lightnesses or lightness values. The pixels are square in this exemplary embodiment, but can also have different side lengths. The length of a pixel is then the length of the longer side of the pixel.

10055] For checking the value documents, there is stored in the rnemoiy 40 in a portion serving as part of the evaluatioll device 34, and thus in this example in the control device 30, a program with program code which, upon execution by the evaluation device 34, i.e. here the processor 38, carries out the steps of a method for checking value documents, in particular soiled value documents, for authenticity, which are illustrated schematically in Fig. 3.

10056] Upon the check, examination regions are defined for each orientation for a stipulated value-document type. The present regions are an JR pnnted image region which has an IR printed image in authentic value documents of the stipulated value-document type, and an IR-transparent printed region which is transparent in the examined IR region in authentic value documents of the stipulated value-document type. Further, a security thread portion is given in which a security thread is present in authentic value documents of the stipulated value-document type. Said security thread portion, hereinafter also designated as a special examination region, is treated differently from the examination regions. Fig. 2 shows by way of example and schematically a value document 12 having a colored printed image 50 in the visible wavelength region, represented by continuous lines and hatching, and an IR image 52 represented by dotted lines and dotted regions, as well as -represented by dashed lines -an JR printed image region 54, an JR-transparent printed region 56 and a security thread portion 58 around a security thread 60. The security thread portion is not employed in the present first exemplary embodiment. Further, one can see an instance of soiling in the form of a stain 62 in the IR printed image region 54, which shows both in the colored image and in the IR image.

10057] For the described method portion relating to the check using the IR image, the two authenticity classes "suspect" and "recognized as authentic" are employed, with the authenticity class "recognized as authentic" being assigned whenever the authenticity class "suspect" is not assigned.

100581 First, in step S 10, a digital JR image of a value document is captured by the evaluation device 34 or the processor 38 by means of the optical sensor 33 and, in this exemplary embodiment, a digital colored image thereof by means of the optical sensor 32, which respectively comprise pixels. In so doing, corresponding pixel data are respectively captured for the pixels.

100591 The evaluation device 34 then, in step S 12, establishes from the digital colored image the value-document type of the value document, i.e. in the case of bank notes for example the denomination or face value and currency, and the orientation of the value document, for example front side or back side of the bank note and upright alignment or not, and stores corresponding values. The colored image is not subsequently required for the method to be described here, so that further employment of the colored image will not be described.

100601 In step 514, the evaluation device establishes the pixels in the examination regions of the IR image that are stipulated for the value-document type and the recognized orientation, and, for said examination regions, error pixels among the pixels in the respective examination region. For this purpose, it checks for each of the pixels in the respective examination region whether the pixel data of the pixels lie within or outside a stipulated pixel-and/or location-dependent authenticity region.

Pixels are recognized as error pixels when their pixel data lie outside the stipulated pixel-and/or location-dependent authenticity region. In the present example, all pixels can be run through successively for this purpose, with the check only being carried out for the pixels in the examination regions. However, in other exemplary embodiments it is also possible to establish for each of the examination regions a list of those pixels that lie within the respective examination region; thereafter the check can then be effected for each of the examination regions while employing the respective list.

100611 In the present example, the image is an IR image whose pixels only have lighthess values as pixel data. The pixel data are thus only one-dimensional.

Accordingly, the authenticity region for a respective pixel is likewise only one-dimensional, in the present example an interval. The limits of the respective interval depend on the respective pixel of the image of the value document or on the location on the value document.

100621 The intervals for the pixels of the examination regions can be obtained by capturing IR images for a stipulated, preferably great, number of value documents of the stipulated value-document type in the respective orientation and in the freshly printed as well as used states, making the JR images as congruent as possible, and then establishing from the distribution of the pixel data, in this example of the lightness values, the limits of the respective interval. The limits can be established for example with statistical methods or by estimation. There thus result for the examination regions two threshold images: an upper threshold image is given by the upper interval limits as pixel data, and a lower threshold image by the lower interval limits as pixel data.

100631 Fig. 4 shows for illustration a diagram in which lightnesses H are shown for pixels n along a straight line parallel to a longitudinal edge of a value document, for example the value document in Fig. 2. The lines 00 and UG shown as continuous for simplicity's sake represent the upper and lower limits of the authenticity region for the pixels n. The authenticity region for a pixel lies between the lines UG and 00 on a straight line corresponding to the pixel, cutting the n-axis at the place colTesponding to the pixel and parallel to the H-axis. By way of example there are shown pixel data of a captured JR image, which are represented by crosses. Some of the pixel data lie within the authenticity region, i.e. between the upper and lower limits. The pixel data marked by o lie above the authenticity region, thus corresponding to upper error pixels, while the pixel data marked by u lie below the corresponding authenticity region and thus belong to lower error pixels.

100641 In this exemplary embodiment, in the step of establishing the error pixels for the examination regions, the error pixels are divided into upper error pixels whose n -, Li pixel data lie outside the authenticity region in a first direction, and lower error pixels whose pixel data lie outside the authenticity region in a second direction opposite to the first direction. Since the authenticity region is one-dimensional in this example, being an interval for a respective pixel, the upper error pixels are error pixels whose pixel data lie above the authenticity region, i.e. here the upper interval limit, and the lower error pixels are error pixels whose pixel data lie below the authenticity region, i.e. here the lower interval limit. Upon the check, the evaluation device 34 writes for the examination regions the pixels that it has recognized as upper error pixels to a first list assigned to the respective examination region, and the pixels that it has recognized as lower error pixels to a second list assigned to the respective examination region.

That the pixels are written to the list can be understood in particular to mean that colTesponding data identifying the pixels are written to the list. However, in oilier exemplary embodiments it is also possible to assign identifiers to the pixels or pixel data, which identify a respective pixel as an upper or lower error pixel.

100651 In the following step S16, the evaluation device 34 checks whether error pixels were recognized at all. if this is not the case, the authenticity class "recognized as authentic" is assigned to the value document or its image in a step 520.

100661 If this is not the case, the evaluation device 34 checks, in the following step Si 8, whether or not the error pixels in the examination regions correspond to soiling, for example a stain, on the value document. In the present exemplary embodiment, the evaluation device 34 for this purpose carries out in the checking step for each of the examination regions a first region check S l8A for the upper error pixels and a second region check 518B for the lower error pixels (cf. Fig. 5). A region check is only carried out and subsequently taken into consideration when corresponding upper or lower error pixels were found for the respective examination region. Said region checks for the examination regions can be executed successively, in interlaced fashion, or in parallel.

10067] The evaluation device 34 then ascertains that the upper or lower error pixels in the examination regions correspond to at least one instance of soiling, for example at least one stain, on the value document when a stipulated minimum number, in the example all, region checks for the respective examination regions yield that the upper or lower error pixels in the respective examination region correspond to at least one instance of soiling, for example at least one stain, on the value document.

10068] 011 the oilier hand, the evaluation device 34 recognizes that en-or pixels do not correspond to soiling when at least one of the region checks yields that the error pixels employed in the respective region check do not correspond to soiling, for example to at least one stain, on the value document.

100691 In step S20, the evaluation device 34 assigns the authenticity class "suspect" to the value document or image when an indication of no soiling was recognized, while otherwise the authenticity class "recognized as authentic" is assigned. In this step S20, the evaluation device 34 stores a corresponding value and forms a signal representing the assigned authenticity class.

10070] These results for the authenticity classes are then, in subsequent, unspecified method steps, linked with results of an authenticity check on the basis of the colored image. These are represented in Fig. 3 as only step S22. In dependence on the result of the check, the control device then actuates the gates. For example, a value document can be classified as "recognized as authentic" when all authenticity checks have yielded this authenticity class. In the other cases the authenticity class "suspect" can be assigned.

10071] In die present exemplary embodiment, the evaluation device 34 compares in the first or second region check for a respective one of the examination regions as a region criterion whether the number of the upper error pixels in the respective exammation regioll exceeds a first maximnmn value stipulated for the respective examination region, or the number of the lower error pixels exceeds a second maximum value stipulated for the respective examination region. In the present exemplary embodiment, the upper and lower maximum values for a respective examination region are equally great amid equal to die smallest integer that is greater than 90% of the number of all pixels in the respective examination region. The evaluation device 34 recognizes an indication of soiling in the examination region on the value document when the numbers of the error pixels, i.e. the number of the upper error pixels and that of the lower error pixels, undershoot the respective maximum value. These checks are symbolized hi Fig. 6A and Fig. 6B by blocks B lOa and B lob, respectively.

100721 A second exemplary embodiment hi Fig. 7 differs from the preceding exemplary embodiment in that the first and second region checks S 18A and S 18B are replaced by first and second region checks S18A' and 5l8B' (cf Fig. 8). Step S14 is also replaced by a step S 14'. The other steps remain unchanged. Step S14' differs from step S14 in that upon the establishment of the error pixels for the upper and lower error pixels, distances of the pixel data from the authenticity regions are established. For the upper error pixels there is hence formed and stored as the distance the difference between pixel data and upper region boundary or interval limit, and for the lower error pixels as the distance the difference between lower region boundary or interval limit and pixel data. This is illustrated in Fig. 4, where the distance A' represents the difference between the lightness, i.e. the pixel data, and the upper limit, for an upper error pixel. Accordingly, the distance A" represents the difference between the lower limit and the lightness, i.e. the pixel data, for a lower error pixel.

100731 The step S18 is replaced by a step S18', represented very schematically in Fig. 8, which differs from the step S18, on the one hand, in that, instead of the region checks Sl 8A and S 1 8B with the region checks B 1 Oa and B lOb for the examination regions, there are now carried out region checks Si 8A' and SI 8B' with subchecks B I 2a and B 14a or B I 2b and B I 4b for the examination regions.

100741 When checking whether error pixels in the examination regions correspond to soiling, the evaluation device 34 carries out the subchecks B12a and B12b for the upper and lower error pixels of a respective examination region of whether or not a distribution of the distances of the upper error pixels satisfies an upper distribution criterion stipulated for the respective examination region for the presence of soiling, and a distribution of the distances of the lower error pixels a lower distribution criterion stipulated for the respective examination region.

100751 This is illustrated by way of example in Fig. 10. The two representations (a) and (b) show histograms which show -in arbitrary units -the frequency H of the found distances as a function of the distances A. In soiled value documents (cf Fig. 10 (a)) a wide distribution is found, since deviations due to soiling are not systematic. In a forgery (cfi Fig. 10 (b)), on the other hand, the distances are distributed much more sharply, i.e. with less width, since the deviations due to forgery are very frequently systematic and in particular very similar.

100761 In the present exemplary embodiment, the evaluation device 34 checks for this purpose (cf. block B l2a and B 12b in Fig. 9A, Fig. 9B) when checking the upper or lower distribution criterion respectively whether the width of the distribution exceeds or undershoots a stipulated upper or lower threshold value. As a measure of the width of the distribution there is established in the present exemplary embodiment the coefficient of variance, i.e. the standard deviation of the distances of the upper error pixels divided by the average of the distances of the upper error pixels or the standard deviation of the distances of the lower error pixels divided by the average of the distances of the lower error pixels. The threshold values are stipulated separately for each of the examination regions and can have been established by analyzing stipulated value documents of the stipulated value-document type for example by estimation. The threshold values can in particular be greater than the corresponding coefficients of variance of freshly printed value documents of the stipulated type.

100771 When the respective threshold value is exceeded by the respectively established width, i.e. here the distribution coefficient, an indication of the presence of soiling is recognized.

100781 Altogether, an indication of soiling is only recognized in this region check when the upper and lower coefficients of variance exceed the stipulated threshold value.

100791 Further, the evaluation device 34, when checking whether error pixels in the examination regions or each of the examination regions correspond to soiling, carries out for each of the examination regions iii the region checks the subchecks B 14a or B14b (di Figs. 9A and 9B) of whether or not a local correlation of the upper error pixels in the respective examination region with a stipulated reference image of the examination region or a correlation of the lower error pixels in the examination region with a stipulated reference image of the examination region exceeds an upper or lower correlation maximum value stipulated for the region. An undershooting is recognized as an indication of the presence of soiling. In this embodiment, there is employed as a reference image a printed image for value documents of the recognized value-document type, which can be obtained by averaging over a stipulated number of colored printed images of freshly printed value documents of the recognized value-document type in the recognized orientation. If there is a high local correlation of the deviations with the reference image, one must assume systematic deviations, as typically occur in forgeries. Soiling or stains are frequently present on a value document randomly, however, and show a non-uniform, frequently likewise random course, which then leads to deviations that do not correlate well with the reference image. In this exemplary embodiment, the correlation is established for different displacement vectors on the grid of the pixels of the image, and there is employed as a correlation the maximum correlation. The displacement vectors v here have a length smaller than a stipulated maximum length, in the example than a stipulated factor M multiplied by the length of the pixels or the distance of nearest neighboring pixels from each other. In the example, the natural number M is equal to 5. The correlation is represented by the maximum normalized cross-correlation between the distances of the error pixels and reference-image pixels corresponding in their position relative to each other. When {i} and {j} respectively designate the sets of the upper and lower error pixels, and A1 and A the deviations of the pixel data, i.e. lightnesses of the error pixels, from the authenticity region boundaries at locations i and j in the JR image, and R1 and R the pixel data for a suitable color channel of the reference image, for example the red chaimel, for pixels at the locations i andj, there can be employed as the normalized cross-correlation as a function of the displacement vector v the function >J4R1+V 1JAI2 JR/+V (I) (I) for the correlation of the upper error pixels, and the function {j) A32 / R ti V {J} for the correlation of the lower error pixels.

100801 R+ designates the value for the pixel at the location that results by addition of the vector v to the location of the error pixel i. The same holds accordingly for the lower error pixels.

100811 Only when the two subchecks B l2a and B 14a or the two subchecks B l2b and B 14b for a respective one of the examination regions yield that indications of the presence of soiling were respectively recognized, it is recognized for the corresponding first or second region check S 18k or S 18ff that the respective error pixels in the examination region correspond to at least one instance of soiling.

100821 In step S18 it is then ascertained that the error pixels in the examination regions correspond to at least one instance of soiling when this was established for the upper and lower error pixels in the region checks S 18k and S 18ff for all examination regions.

100831 The assigmnent of the authenticity class in step S20 is effected as in the first exemplary embodiment.

100841 A third exemplary embodiment differs from the preceding exemplary' embodiment in that the first and second region checks Si 8A' and SI 8ff are carried out hi an interlaced manner. In so doing, the subchecks B 12a, B 14a, B 12b. B 14b are first carried out for each of the examination regions. For each of the examination regions it is established whether or not the error pixels therein correspond to at least one instance of soiling. The evaluation device decides that the error pixels in the examination regions correspond to at least one instance of soiling when this was established for each of the examination regions.

100851 A further exemplary embodiment differs from the preceding exemplary embodiments hi that in step Si 8 or step Si 8' it is already decided that error pixels do not correspond to soiling when for at least one of the examination regions the fir St or second region check has yielded that the upper or lower error pixels of said examination region do not correspond to soiling, and then no further region checks are carried out.

100861 Yet further exemplary embodiments differ from the preceding exemplary embodiment or the second exemplary embodiment in that the first and second region checks have three subchecks, namely respectively the checks B lOa or B 1 Ob of the first exemplary embodiment and the subchecks B12a and B14a or Bl2b and Bi4b. The evaluation device 34 then ascertains in the step corresponding to step S18 that the first region check for the upper error pixels or the second region check for the lower error pixels yields that the upper or lower error pixels in the respective examination region correspond to at least one instance of soiling, for example at least one stain, on the value document when a stipulated minimum number, in the example two, of the subchecks for the respective region check yield that the upper or lower error pixels in the respective examination region correspond to at least one instance of soiling, for example at least one stain, on the value document.

100871 Yet further exemplary embodiments differ from the preceding exemplary embodiments in that the special examination region is additionally drawn on for the check.

100881 Fig. I! illustrates one of these exemplary embodiments which corresponds to the first exemplary embodiment but includes the check of the special examination region. Unless otherwise mentioned hereinafter, the steps with the reference signs of the first exemplary embodiment correspond to the steps of the first exemplary embodiment and the explanations thereon also hold accordingly here.

10089] As illustrated in Fig. Ii, a step S14" is carried out instead of the step S14.

The step S 14" differs from the step S 14 in that the evaluation device additionally establishes for the special examination region among the pixels therein authenticity pixels whose pixel data lie above a minimum threshold value.

10090] In the case of the special examination region containing the security thread, there is fixed as a minimum threshold value a minimum lightness value which corresponds to a transparent value document and can be chosen for example in dependence on the noise level of the employed sensor. This can also be understood to mean that here, too, an interval is fixed whose lower limit is zero and whose upper limit is the minimum value. For establishing the authenticity pixels, the evaluation device 34 checks whether the pixel data of the pixels lie above the minimum threshold value. If this is the case, the pixels are identified as authenticity pixels similarly to the error pixels.

100911 In a manner corresponding to the check in the examination regions, the evaluation device 34 carries out a check for the special examination region. After the step S 14" it compares, in a step S30', the number of the authenticity pixels with the minimum number stipulated for the portion or special examination region. In this example, there is fixed as a minimum number the integer that is greater than 95% of the number of the pixels in the special examination region or the corresponding portion of the value document that results for an authentic, freshly printed value document.

10092] In step S32, it then assigns an authenticity class selected from the stipulated authenticity classes to the value document or image in dependence on the result of comparison of the number of the authenticity pixels with the minimum number. More precisely, the authenticity class "suspect" is assigned when the established number of the authenticity pixels is smaller than the minimum number. Otherwise, the authenticity class "recognized as authentic" is assigned. As in step S20, the evaluation device 34 stores a corresponding value and, in the step 522", fonns a signal representing the assigned authenticity class, with the assigned authenticity class "recognized as authentic" only being assigned when both step 520 and step S32 have yielded this.

100931 Other exemplary embodiments differ from the preceding exemplary embodiments in that for at least one of the examination regions the region checks SI 8A or SI 8B are carried out, and for at least one other of the examination regions the region checks 518A' or 518B'.

100941 Further exemplary embodiments differ from the preceding exemplary embodiments in that there is employed as a reference image an IR image which can be obtained for example by averaging over a stipulated number of freshly printed authentic value documents in the recognized orientation.

100951 Yet further exemplary embodiments differ from the preceding exemplary' embodiments in that not the IR image is employed, but rather the colored image. In the color space in which the pixel data are stated, a direction is then fixed. Instead of the IR data, the projections of the pixel data onto said direction are then employed. The authenticity regions are then stated with respect to said direction.

100961 In yet further exemplary embodiments there can be employed, instead of the normalized cross-correlation, the absolute or square error, with the highest correlation occurring upon minimal error.

100971 Further exemplary embodiments can differ from the above-described exemplary embodiment in that only one examination region is employed.

Claims

Claims A method for checking value documents, in particular also soiled value documents, for authenticity, having the steps of -capturing pixel data of pixels of an image of the value document, -establishing pixels in one or at least two stipulated examination regions of the image and establishing for the one or the at least two examination regions error pixels among the pixels in the one or a respective one of the at least two examination regions, wherein the pixel data of the error pixels respectively lie outside a stipulated pixel-and/or location-dependent authenticity region, -checking whether or not the error pixels in the examination region or regions correspond to soiling, for example a stain, on the value document, and -assigning an authenticity class selected from stipulated authenticity classes to the value docmnent in dependence on the result of the check.
2. The method according to claim 1, vherein, hi the step of establishing the error pixels or in the checking step, the error pixels are divided into upper error pixels, whose pixel data lie outside the authenticity region in a first direction, and lower error pixels, whose pixel data lie outside the authenticity region hi a second direction opposite to the first direction.
3. The method according to claim 2, wherein the authenticity region is one-dimensional, and the upper error pixels are error pixels whose pixel data lie above the authenticity region, and the lower error pixels are error pixels whose pixel data lie below the authenticity region.
4. The method according to claim 2 or 3, wherein, in the checking step for the examination region or regions, it is checked respectively in a first region check whether die upper error pixels correspond to at least one instance of soiling, for example at least one stain, and/or it is checked respectively in a second region check whether the lower error pixels correspond to at least one instance of soiling, and -, -, ii it is decided that the error pixels in the examination region or regions correspond to soiling when at least a stipulated minimum number of the region checks has yielded that the upper or lower error pixels in the respective examination region correspond to at least one instance of soiling, for example at least one stain, on the value document, or it is decided that error pixels in the examination region or regions do not correspond to at least one instance of soiling, for example at least one stain, on the value docmnent when at least a stipulated inhthnuin nmnber of the region checks has yielded that the upper or lower error pixels in the respective examination region do not correspond to at least one instance of soiling, for example at least one stain, on the value document.
5. The method according to claim 4, wherein in the first region check for the examination region or one of the examination regions the number of the upper error pixels in the respective examination region is compared with a stipulated first maximum value and/or in the second region check for the examination region or one of the examination regions the number of the lower error pixels is compared with a stipulated second maxhnum value, and an indication of soiling in the respective examination region is recognized when the numbers do not exceed the respective maximum value, or an indication of no soiling on the value document hi the respective examination region is recognized when the first and/or the second number exceeds the first or second maximum value.
6. The method according to any of claims 2 to 5, wherein, upon establishing or upon checking, distances of the pixel data from the authenticity regions are established for the upper and lower error pixels.
7. The method according to claim 6, wherein in the first region check for the examination region or one of the at least two examination regions it is checked whether a distribution of the distances of the upper error pixels satisfies an upper distribution criterion stipulated for the examination region, and/or in the second region check for the examination region or one of the at least two examination regions whether a distribution of the distances of the lower error pixels satisfies a lower distribution criterion stipulated for the examination region for the presence of soiling.
8. The method according to claim 7, wherein when checking the distribution criterion it is respectively checked whether the width of the distribution exceeds or midershoots a stipulated threshold value, and an indication of the presence of soiling is preferably recognized when the threshold value is exceeded.
9. The method according to any of claims 2 to 8, wherein in the first or second region check for the examination region or one of the at least two stipulated examination regions a correlation of the upper or lower error pixels in the examination region with a stipulated reference image of the examination region is established, and an indication of the presence of soiling is preferably recognized when the correlation of the upper or lower error pixels undershoots a stipulated maximum colTelation value.
10. The method according to claim 1, wherein in the step of checking whether the error pixels in the examination region or at least two examination regions correspond to soiling, it is checked whether the number of the error pixels is smaller than a threshold value stipulated for the respective examination region, and when this is the case a presence of soiling is recognized or an indication of the presence of soiling is recognized.
11. The method according to any of the preceding claims, wherein the image is an JR image of the value document.
12. A method for checking value documents, in particular soiled value documents, having a portion with an IR-absorbent security feature, preferably a security thread, for authenticity, having the steps of -capturing pixel data of pixels of an JR image of at least the portion of the value document, -establishing authenticity pixels among the pixels in the IR image of the portion, wherein the pixel data of the authenticity pixels lie above a minimum threshold value, -companng the number of the authenticity pixels with a minimum number stipulated for the portion, and -assigning an authenticity class selected from stipulated authenticity classes to the value document hi dependence on the result of comparison of the number of the authenticity pixels with the ininhnum number.
13. A device for checking value documents, in particular soiled value documents, for authenticity, having an optical sensor for capturing an image of the value document, wherein the image has pixels having pixel data assigned thereto, and an evaluation device which is arranged for carrying out a method according to any of the preceding claims.
14. An apparatus for processing value documents, in particular also soiled value documents, having -a feeding device for value documents to be processed, -an output device for value documents which has at least two output portions for receiving processed value documents, -a transport device for transporting singled value documents from the feeding device along a transport path to die output device, -a checking device according to claim 13 which is so disposed that the transport path extends through the capture region thereof, and -a control device which actuates the transport device in dependence on an authenticity signal of the checking apparatus for a value document transported by the transport device, such that the value document is transported into an output portion of die at least two output portions that corresponds to die authenticity signal.
15. A computer program for executioll by means of a data processing device having a processor, containing program code upon whose execution the processor carries out the method according to any of claims 1 to 12.
16. A computer-readable data carrier on which a computer program according to claim 15 is stored.