SENSOR AND METHOD FOR VERIFYING VALUE DOCUMENTS

Abstract

The invention relates to a sensor for verifying value documents, which has a plurality of fibres that are distributed over the value document and have a characteristic optical or magnetic signal. The sensor has an image capture device for spatially resolved detection of optical or magnetic signals of the value document, which device is designed to capture a value document image of the value document. Furthermore, the sensor has an analysis device which, for the purpose of analysing the value document, is designed to locate the fibres in the value document image and determine at least one local fibre characteristic value for one or more different locations of the value document image in each case and classify the value document as suspected counterfeit on the basis of the local fibre characteristic value. If the value documents are processed automatically, the suspected counterfeit value document can then be automatically rejected.

Claims

1-17. (canceled)

18. A sensor for verifying a value document which comprises a multiplicity of fibers distributed over the value document, and which is brought into a capture region of the sensor for the purpose of verifying said value document, wherein the sensor comprises: an image capture device configured to capture a value document image of the value document, wherein the value document image shows a characteristic optical or magnetic signal of the fibers, and an evaluation device, which, for the purpose of evaluating the value document image, is configured, to localize the fibers contained in the respective value document image, and to ascertain for one or more different locations of the value document image in each case at least one local fiber characteristic value which applies to the respective location and optionally to surroundings of the respective location on the value document image, and to classify the value document as suspected counterfeit depending on the at least one local fiber characteristic value.

19. The sensor according to claim 18, wherein the evaluation device, for the purpose of evaluating the value document image, is configured to scan the value document image point by point using grid points, wherein a local fiber characteristic value is ascertained for each grid point of the value document image, and to classify the value document as suspected counterfeit depending on the local fiber characteristic values ascertained for the grid points.

20. The sensor according to claim 19, wherein the evaluation device, for the purpose of evaluating the value document image, is configured: to identify one or more conspicuous grid points which are conspicuous with regard to the local fiber characteristic value ascertained at the respective grid point, to identify one or more conspicuous regions of the value document image in which conspicuous grid points are situated, and to classify the value document as suspected counterfeit depending on one or more properties of the respective conspicuous region and/or depending on the local fiber characteristic values which were ascertained for the grid points of the respective conspicuous region.

21. The sensor according to claim 19, wherein the evaluation device, for the purpose of evaluating the value document image, is configured to ascertain the local fiber characteristic value ascertained for the respective grid point of the value document image for a surrounding region assigned to the respective grid point, or, for the purpose of evaluating the value document image, is configured to ascertain the respective local fiber characteristic value ascertained for the respective grid point of the value document image pointwise for the respective grid point.

22. The sensor according to claim 21, wherein the evaluation device, for the purpose of evaluating the value document image, is configured: to determine one or more partial images of the value document image, each of which contains a segment of the value document image, wherein the respective partial image of the value document image is a surrounding region around a respective one of the grid points, to determine the local fiber characteristic value for the respective partial image, to classify the value document as suspected counterfeit depending on the local fiber characteristic value(s) of one or more of the partial images.

23. The sensor according to claim 18, wherein the local fiber characteristic value used is a measure of local density of the fibers, wherein the number of fibers situated in the surroundings of the respective location, the respective partial image or in the respective surrounding region around a grid point, or the fiber density in the surroundings of the respective location, in the respective partial image or in the respective surrounding region, or a distance measure characteristic of the distance between the fibers of the fibers in the surroundings of the respective location, including a mean distance of the fibers of the partial image or of the surrounding region from the fiber that is the most closely adjacent to the respective fiber.

24. The sensor according to claim 22, wherein the evaluation device is configured to determine at least two partial images of the value document image which together cover at least 50% of the area of a value document side of the value document, wherein the partial images are arranged in rows and/or in columns on the value document, wherein the partial images are arranged in a grid composed of at least two rows and/or at least two columns on the value document.

25. The sensor according to claim 18, wherein the local fiber characteristic value used is a measure of the local constitution of the fibers at the respective location and optionally in the surroundings of the respective location, which concerns the shape or size or color or brightness of the fibers at the respective location and optionally in the surroundings of the respective location, wherein the local fiber characteristic value used is a representative value of the local constitution of the fibers at the respective location and optionally in the surroundings of the respective location.

26. The sensor according to claim 18, which is configured, on the basis of one or more properties of the fibers localized in the value document image and/or on the basis of one or more of the local fiber characteristic values of the respective value document which were ascertained for the different location(s) of the value document image, to determine individual fingerprint data of the respective value document, to assign the individual fingerprint data to processing data of the respective value document and to store them together with the processing data of the respective value document, optionally in a list for a plurality of value documents, in each case in a manner assigned to one another.

27. The sensor according to claim 18, wherein the evaluation device has information about the value document type of the value document to be verified and is configured to determine those locations of the value document image for which the respective local fiber characteristic value is determined depending on the value document type of the value document to be verified, to carry out the determining of the partial images depending on the value document type of the value document to be verified, wherein a plurality of partial images are fixedly predefined in the evaluation device and the evaluation device is configured to determine specific partial images for evaluation purposes from these fixedly predefined partial images depending on the value document type of the value document to be verified.

28. The sensor according to claim 18, wherein the sensor is configured for verifying value documents of one or more specific value document types, the value documents of which comprise in each case at least one disturbing security element which adversely affects the capture of the characteristic optical or magnetic signal of the fibers, and the evaluation device of the sensor is configured to classify the value documents of the specific value document type as suspected counterfeit depending on the at least one local fiber characteristic value of only one or a plurality of such locations of the value document image at which the capture of the optical or magnetic signal of the fibers is not adversely affected by the at least one disturbing security element.

29. The sensor according to claim 18, wherein the evaluation device, for the purpose of evaluating the value document image, is configured: to assign an effect region to each of the localized fibers, for a plurality of different locations of the value document image to analyze in each case locally the effect regions of the fibers present at the respective location and optionally in the surroundings thereof, with regard to the area covered or not covered by the effect regions or with regard to the overlap of adjacent effect regions, and to ascertain the respective local fiber characteristic value of the respective location on the basis of the local analysis of the effect regions which lie at the respective location and optionally in the surroundings thereof, wherein the evaluation device, for the purpose of evaluating the value document image, is configured to choose the size of the effect regions in each case depending on an expected density value indicating the area density of the fibers that is expected for the value document image and optionally for the respective location of the value document image.

30. The sensor according to claim 27, wherein the evaluation device, for the purpose of evaluating the value document image, is configured: to assign an effect region to each of the localized fibers of the value document image or of the respective partial image or of the respective surrounding region and to identify coverage regions which belong to the effect region of at least one of the fibers, and/or to identify free regions which do not belong to any effect region of at least one of the fibers, and/or to identify overlap regions in which the effect regions of at least two adjacent fibers overlap, and to identify one or more conspicuous regions of the value document image which are conspicuous with regard to their coverage regions and/or free regions and/or overlap regions, and to classify the value document as suspected counterfeit depending on one or more properties of the coverage regions and/or the free regions and/or the overlap regions in/at one or more conspicuous regions of the value document image.

31. The sensor according to claim 18, wherein the evaluation device, for the purpose of evaluating the value document image, is configured: for one or a plurality of locations of the value document image to compare the local fiber characteristic value of the respective location with an expected value, wherein the expected value, for the respective value document, and optionally for the respective location of the respective value document, is predetermined or the evaluation device is configured to determine the expected value with which the local fiber characteristic value of the respective location is compared on the basis of the local fiber characteristic values which are ascertained for one or more other locations of the same value document image, and to classify the value document as suspected counterfeit depending on the comparison results obtained for one or more locations of the value document image, depending on a difference established during the comparison between the local fiber characteristic value of the respective location and the expected value.

32. The sensor according to claim 18, wherein the value document image is a first value document image, which is captured from a first value document side of a value document to be verified, and the evaluation device has a second value document image, which is or was captured from the second sideopposite the first sideof the same value document, and the evaluation device is configured: to determine in each case at least one local fiber characteristic value for at least one first partial image of the first value document image and for at least one second partial image of the second value document image, and optionally to combine the local fiber characteristic value of the first partial image and the local fiber characteristic value of the second partial image with one another or to compare these with one another, and to classify the value document as a suspected counterfeit depending on the local fiber characteristic value of the first partial image and depending on the local fiber characteristic value of the second partial image, depending on a result of the combination or depending on a result of the comparison, wherein the second partial image is a second partial image which corresponds to the first partial image, and which is a segment of the second value document image, wherein the first partial image and the corresponding second partial image contain the optical or magnetic signals of the front and rear sides of the same value document section of the value document.

33. A value document processing apparatus comprising: a sensor according to claim 18, and a transport device for introducing the value documents into the capture region of the sensor.

34. A method for verifying a value document with the aid of a sensor, with the aid of the sensor according to claim 18, wherein the value document comprises a multiplicity of fibers distributed over the value document, and wherein the following steps are carried out in the method: introducing the value document into the capture region of the sensor and capturing a value document image of the value document with the aid of an image capture device of the sensor, wherein the value document image contains a characteristic optical or magnetic signal of the fibers, evaluating the value document image with the aid of an evaluation device of the sensor, which evaluation device for the purpose of evaluating the value document image, is configured: to localize the fibers contained in the respective value document image, and to ascertain for one or more different locations of the value document image in each case at least one local fiber characteristic value which applies to the respective location and optionally to surroundings of the respective location, and to classify the value document as suspected counterfeit depending on the at least one local fiber characteristic value.

Description

[0083] FIG. 1 shows a schematic diagram regarding the setup of a value document processing apparatus comprising an optical sensor for verifying the value documents,

[0084] FIGS. 2a-c show a fiber distribution and a possible partial image in the case of an authentic banknote (FIGS. 2a,b) and an associated counterfeit (FIG. 2c),

[0085] FIGS. 3a-c show a fiber distribution and a grid of predefined partial images in the case of an authentic banknote (FIGS. 3a, 3d) and in the case of an associated counterfeit (FIGS. 3b, 3c, 3e, 3f),

[0086] FIGS. 4a-c show a fiber distribution and a grid of predefined partial images in the case of an authentic banknote (FIGS. 4a, 4c) and in the case of an associated counterfeit (FIGS. 4b, 4d),

[0087] FIGS. 5a-b show a fiber distribution and mutually corresponding partial images of both opposite sides of an authentic banknote, and

[0088] FIGS. 6a-c show a luminescence image which is scanned systematically in a grid of grid points (FIG. 6a), in the case of which an effect region is assigned to each localized fiber (FIG. 6b), in the case of which a surrounding region around each grid point is defined (FIG. 6c).

[0089] In the exemplary embodiments, a banknote having luminescent (e.g. fluorescent or phosphorescent) fibers is considered as value document to be verified, and an optical sensor that records and evaluates a luminescence image of the banknote is considered. However, the invention equally relates to the verification of banknotes having reflective or magnetic fibers or the verification of other value documents having luminescent, reflective or magnetic fibers, wherein a magnetic sensor is used for verification purposes in regard to magnetic fibers, which correspondingly records and correspondingly evaluates a magnetic signal image of the value document.

[0090] FIG. 1 shows by way of example the schematic setup of a value document processing apparatus 1 comprising an introduction compartment 2, in which a stack of banknotes 3 to be processed is provided, and a separator 8, which successively detects a respective banknote of the introduced stack and transfers it to amerely schematically representedtransport device 10 (transport belts and/or transport rollers), which transports the banknote past an optical sensor 25 in the transport direction x for the verification of said banknote.

[0091] The optical sensor 25 has an optical image sensor 20, which converts the luminescence intensities emitted by the banknote transported past into corresponding sensor signals. The optical excitation of the luminescence of the banknote is effected e.g. by means of excitation light sources 27, 28 arranged on both sides of the image sensor 20. However, it is also possible for only one of the light sources to be used. The image sensor 20 has e.g. one sensor linear array or a plurality of sensor linear arrays, e.g. for different spectral components of the luminescence light. The sensor linear array(s) is/are arranged transversely with respect to the transport direction x of the banknotes. The image sensor 20 is controlled by a control device (not shown) in such a way that it detects the luminescence of the banknote at a plurality of detection times in order to optically scan the banknote transported past. In this case, detection regions of the banknote that are arranged adjacently along the transport direction are captured successively over time. The successively captured detection regions of the banknote each correspond to an image point of the luminescence image.

[0092] The optical sensor 25 is arranged on the left-hand side of the transport pathas viewed in the transport direction x of the banknote. A further optical sensor 29 can be arranged opposite the optical sensor 25, on the right-hand side of the transport path, and likewise has an optical image sensor (not shown) and optionally illumination devices for the optical excitation of the banknote luminescence and optionally a dedicated evaluation device. The luminescence image recorded by the optical image sensor of the optical sensor 29 situated opposite is optionally transmitted to the evaluation device 19 of the optical sensor 25 in order to enable joint evaluation of both luminescence images of the same banknote.

[0093] The image sensor 20 forwards the recorded luminescence image to the evaluation device 19 of the optical sensor 25. The evaluation device 19 can be contained in the housing of the optical sensor 25 or else outside that, e.g. in the value document processing apparatus 1. The evaluation device 19 determines the respective local fiber characteristic value of one or more locations, e.g. of one or more partial images, of the luminescence image recorded by the image sensor 20. On the basis of the local fiber characteristic values, the authenticity of the banknote is verified and the banknote is possibly classified as suspected counterfeit.

[0094] For one or more value document types, information about the grid points or partial images and optionally about expected values can be stored in a data memory 26 of the evaluation device 19. Information about the value document type to be verified in each case and optionally about the transport position of the value documents 3 can be communicated to the evaluation device 19 by the control device 50 of the apparatus 1.

[0095] Depending on the authenticity of the respective banknote ascertained by the evaluation device 19, the transport device 10 and also the diverters 4 and 5 along the transport path are controlled by the control device 50 in such a way that the banknote is fed to one of a plurality of dispensing compartments 30 and 31 and is placed there. By way of example, banknotes which were recognized as authentic are placed in a first dispensing compartment 30, while banknotes categorized as suspected counterfeit are placed in a second dispensing compartment 31. At the end of the illustrated transport path (reference numeral 6), further dispensing compartments and/or other devices can be provided, for example for storing or for destroying banknotes and/or a reject compartment, into which banknotes are placed for a separate treatment, for example by an operator.

[0096] In the example illustrated, the value document processing apparatus 1 furthermore comprises an input/output device 40 for the input of data and/or control commands by an operator, for example by means of a keyboard or a touchscreen, and for the output or display of data and/or information concerning the processing process, in particular concerning the banknotes processed in each case.

[0097] 1.sup.st exemplary embodiment

[0098] FIG. 3a shows by way of example the fiber distribution of luminescent fibers of an authentic banknote 80 of currency A, denomination 10. Moreover FIG. 3b shows an example of the fiber distribution of a composed counterfeit 66 with respect to this banknote 80. The composed counterfeit 66 is a composition of a part 82 of an authentic banknote, which has a mean fiber density of 1/cm.sup.2, and a counterfeit part 62, which has no fibers at all.

[0099] For evaluation purposes, firstly the fibers are localized in the luminescence image recorded by the optical sensor 25. The luminescence image is then scanned systematically in a grid of e.g. 24 grid points. In the first exemplary embodiment, the fiber density is used as a local fiber characteristic value. For each grid point, the local fiber density in a surrounding region lying around the respective grid point is determined in each case, said region corresponding to a partial image of the banknote image. In the example in FIGS. 3d,e,f, the surrounding regions/partial images 11-14, 21-24 are distributed over the banknote in a grid comprising 2 rows and 4 columns. The 8 grid points lie e.g. in the respective center of the respective surrounding region or partial image. In the evaluation device 19 of the optical sensor 25, the grid of the partial images 11-14, 21-24 is fixedly predefined for a plurality of banknote types. The area content of the partial images is e.g. F=8 cm.sup.2. It is assumed that the optical sensor 25 is configured for verifying banknotes of currencies A and D in which all denominations have fibers with a distribution that is uniform, but random over the banknote, with a specific mean area density (expected density value).

[0100] The optical sensor 25 acquires information about the banknote type to be verified from the value document processing apparatus 1 or from another sensor of the value document processing apparatus 1. A table stored in the data memory 26 of the evaluation device 19 stipulates which of the predefined partial images 11-14, 21-24 are actually intended to be examined for the respective banknote type, cf. table 1. In accordance with table 1, all partial images of the P t and 2n d columns are intended to be verified for currency A, denomination 10, and only 7 of the possible 8 partial images (partial image 24 is omitted) for currency A, denomination 50. With the aid of the information made available to the optical sensor 25 regarding the banknote type to be verified, the evaluation device 19 can select one or more of the possible partial images for evaluation purposes depending on the banknote type.

[0101] The expected density value DE of the fibers is also indicated for the respective banknote type in stored table 1.

TABLE-US-00001 TABLE 1 for currencies A and D, denominations 10 and 50 in each case Expected density Partial images to value DE be verified Currency A, 1/cm.sup.2 11-14, 21-24 denomination 10 Currency A, 1/cm.sup.2 11-14, 21-23 denomination 50 Currency D, 5/cm.sup.2 11-14, 21-24 denomination 10 Currency D, 2/cm.sup.2 11-14, 21-24 denomination 50

[0102] The acceptable fluctuation range S for the fiber density can be fixedly predefined or be calculated depending on the expected density value and depending on the area content, e.g. S=0.5/cm.sup.2.

[0103] On the basis of the recorded value document image, the evaluation device 19 of the optical sensor 25 determines the fiber number N in each of the partial images to be verified and with the aid of the area content F calculates therefrom the respective fiber density D=N/F. It then compares the latter with the respective expected density value DE, with DE=1/cm.sup.2 in the case of currency A, for each of the partial images to be verified. This involves verifying whether the fiber density D is in the range DE+/S. For the case of currency A, denominations 10 and 50, in accordance with table 1, this involves e.g. verifying whether the fiber density in the partial images is in the range E+/S=1+/0.5/cm.sup.2, i.e. between 0.5/cm.sup.2 and 1.5/cm.sup.2. If this is the case, the respective partial image or the banknote is categorized as not suspicious. If the fiber density D is greater or less than DE+/S, the respective partial image or the banknote is classified as suspected counterfeit.

[0104] The composed counterfeit 66 from FIGS. 3b, 3e is a composed counterfeit with respect to currency A, denomination 10. The latter means verifying all the partial images 11-14 and 21-24 according to table 1. No fibers are found (fiber density D=0) in the two partial images 11 and 21 and the latter are therefore categorized as suspicious. In the case of the partial images 12 and 22, the fiber density D is still in the acceptable range DE+/S, and so these partial images are not categorized as suspicious. Moreover in the case of the partial images 13, 14, 23, 24, too, a fiber density D in the range DE+/S is found in each case and these partial images are categorized as not suspicious. On account of the two partial images 11 and 21 categorized as suspicious, the composed counterfeit 66 from FIGS. 3b, 3e is classified as suspected counterfeit.

[0105] The composed counterfeit 67 from FIGS. 3c, 3f is a counterfeit with respect to currency A, denomination 50. Both the authentic banknote and the counterfeit 67 have a luminescent security element in the region 68. Said security element outshines the luminescence of the luminescent fibers in the luminescence image. Table 1 reveals the evaluation device 19 of the optical sensor 25 that for currency A, denomination 50in contrast to denomination 10 of currency Athe fiber density is verified only in the partial images 11-14 and 21-23, but is not verified in the partial image 24 (which lies in the region of the luminescence element 68). The verification results of the partial images 11-14 and 21-23 correspond to those of the counterfeit 66 from FIGS. 3b,3e.

[0106] If the transport position of the banknotes is variable, the sensor can acquire information about the transport position of the banknote and, on the basis of the transport position, determine the partial images in which the fiber density is intended to be verified. For this purpose, in table 1 the information about the partial images to be verified can be stored for up to four transport positions of the respective banknote type. In the course of sensor adaptation, in order to create table 1 extended in this way, it is possible to ascertain in which of the partial images 11-14, 21-24 the luminescence element 68 is situated in the context of the respective transport position and must therefore be omitted in the course of the verification. By way of example, in the context of a different transport position of the composed counterfeit 67 where the luminescence element 68 is at the bottom left in the luminescence image, the partial image 21 would be omitted instead of the partial image 24.

[0107] 2.sup.nd exemplary embodiment

[0108] FIG. 2a shows by way of example the fiber distribution of luminescent mottled fibers of an authentic banknote 70 of currency B. The banknote 70 has mottled fibers only in the left-hand half, specifically with a mean area density of 1/cm.sup.2. The right-hand half of the banknote 70 does not have any mottled fibers. In this example, the sensor does not acquire information about the banknote type to be verified, but rather is fixedly set to currency B, the denominations of which all have such a fiber distribution.

[0109] Independently of the banknote type, in the evaluation device 19, a single partial image 11 in the left-hand half of the banknote is always defined for all denominations of currency B, cf. FIG. 2b. The area content of said partial image can be fixedly predefined, e.g. as 10 cm.sup.2. However, it is also possible to choose the area content or the size depending on the expected density value (given by the known mean area density) which is known for currency B and which is stored in the evaluation device 19. Assuming that the expected value E should be 10, then a partial image 11 with an area content of 10 cm.sup.2 should be chosen given an expected density value of 1/cm.sup.2.

[0110] For evaluation purposes, firstly the fibers are localized in the recorded luminescence image. In the second exemplary embodiment, the number of fibers localized in a partial image 11 is used as a local fiber characteristic value. FIG. 2c shows a counterfeit 64 with respect to a banknote of currency B, which has no mottled fibers at all. The fiber number N=0 is then determined in the partial image 11. However, since a fiber number of more than zero (E>0) is expected in the partial image 11 for currency B, the counterfeit 64 from FIG. 2c is classified as suspected counterfeit.

[0111] In order also to find counterfeits with an incorrect number of fibers, a more accurate verification can be carried out in which the fiber number N of the respective partial image is compared with the expected value E for the number: in the case of a banknote to be verified, the number N of mottled fibers in the partial image 11 is then determined and compared with the predetermined expected value, e.g. E=10. Preferably, in the context of the comparison, an exact correspondence to the expected value is not demanded, rather an acceptable fluctuation range S around the expected value E is permitted, e.g. S=5. The comparison involves verifying whether or not the fiber number N is in the range E+/S. If it is not, the banknote is classified as suspected counterfeit, and if it is, the banknote is classified as not suspected counterfeit.

[0112] In the concrete example, where E=10 and S=5, the banknote is classified as follows depending on the fiber number N: [0113] if no or up to 4 mottled fibers are contained (as in the case of counterfeit 64 from FIG. 2c): suspected counterfeit [0114] if 5 to 15 mottled fibers are counted (as in the case of the authentic banknote 70 from FIGS. 2a,b): not suspected counterfeit [0115] if more than 15 mottled fibers are counted: suspected counterfeit.

[0116] If the transport position of the banknotes is variable in the case of the value document processing apparatus, information about the transport position can be communicated to the optical sensor 25, on the basis of which said optical sensor defines the position of partial image 11 in such a way that the partial image 11 lies in a banknote section in which the fibers are present in the case of the authentic banknote 70. In the context of the transport position from FIGS. 2a,b, the partial image 11 is therefore put into the left-hand half of the value document image of the banknote 70. In the context of a different transport position of the banknote 70 where the fibers are on the right in the value document image, the partial image 11 would be positioned in the right-hand half of the value document image.

[0117] 3.sup.rd exemplary embodiment

[0118] FIG. 4a shows an authentic banknote 90 of currency C, denomination 50, which has fibers distributed approximately uniformly over the banknote with a mean area density (expected density value) of 0.66/cm.sup.2, which applies to all denominations of currency C.

[0119] Moreover FIG. 4b shows an example of the fiber distribution of a composed counterfeit 65 with respect to the banknote 90, in the case of which a part 92 of an authentic banknote is combined with a counterfeit part 63, in which there is a lower fiber density than in the case of the authentic banknote. The authentic part 92 has e.g. a mean fiber density of 0.67/cm.sup.2, with that of the counterfeit part 63 being 0.1/cm.sup.2.

[0120] Owing to the mean area density of the fibers known to be lower, the evaluation device 19 defines partial images with a larger area content for currency C than for currency A.

[0121] In the course of sensor adaptation, a table 2 with a partial image definition (position and dimensions), an expected value E and an acceptable fluctuation range S for each partial image was determined for each banknote type to be verified by the sensor, cf. in table 2. The respective area content is found by multiplying the partial image width dx and the partial image height dy. The partial images 11, 12, 21, 22 from table 2 all have the same shape and size, cf. FIG. 4c.

TABLE-US-00002 TABLE 2 for currency C, all denominations Partial image Partial Partial coordinates Partial image Fluctu- image (x, y) (top image width height Expected ation number left corner) dx in mm dy in mm value E range S 11 5, 5 60 40 16 5 12 5, 50 60 40 16 5 21 70, 5 60 40 16 5 22 70, 50 60 40 16 5

[0122] By virtue of the explicit partial image definitions in table 2, however, different partial image sizes on the same banknote are also possible. The expected values E and fluctuation ranges S are identical here for all the partial images 11, 12, 21, 22 since the area content thereof is identical. However, they can also be different, e.g. in the case of partial images having different area contents. With a larger area content, a larger fluctuation range S can also be permitted. Corresponding tables for other denominations or currencies to which other numerical values generally apply are also stored in the optical sensor 25. In the case of nonuniformly distributed fibers, such as e.g. in the case of the banknote 70 of currency B from FIG. 2a, the corresponding table would contain different expected values E for different partial images.

[0123] The acceptable fluctuation range S can be indicated in the table. It can also be calculated only depending on the expected value E, e.g. 40% of the expected value E, or be settable for the operator of the value document processing apparatus.

[0124] The optical sensor 25 acquires information about the banknote type to be verified, i.e. currency C and optionally the denomination, from the value document processing apparatus 1 or from another sensor. With the aid of the information about the banknote type to be verified, the evaluation device 19 obtains from table 2 the information about the partial images that are predetermined for the banknote type to be verified.

[0125] For evaluation purposes, once again the fibers are firstly localized in the luminescence image recorded by the optical sensor 25. On the basis of the recorded luminescence image of the banknote to be verified, e.g. the number N of fibers in each of these partial images 11,12,21,22 can be determined and compared with the respective expected value E+/S. In the example of currency C, denomination 10, a fiber number of E+/S in the range of 11 to 21 is expected in each of the partial images 11, 12, 21, 22. In the case of the composed counterfeit 65 in FIGS. 4b, 4d, 18 mottled fibers are actually found in the partial images 11 and 21, and so these partial images are not categorized as suspicious. For the partial images 12 and 22, by contrast, in each case a low number of mottled fibers of 8 and 10, respectively, is found and these partial images are therefore categorized as suspicious. Since at least one partial image is categorized as suspicious (12 and 22), the composed counterfeit 65 from FIGS. 4b, 4d is classified as suspected counterfeit with regard to a composed counterfeit.

[0126] In addition or as an alternative to the number of fibers in the respective partial image, it is also possible to use a constitution of the fibers in the respective partial image as a local fiber characteristic value, such as e.g. the length of the fibers, their aspect ratio (width-length ratio), the shape or the color of the fibers, or the respective mean value of said constitution over the respective partial image, or the dispersion of said constitution around said mean value in the respective partial image. Conspicuous partial images can then be identified, the constitution or mean value or dispersion of which deviates from a reference value (by more than an acceptable fluctuation range). Moreover the value document is classified as suspected counterfeit if one or more partial images that are conspicuous in this regard were found.

[0127] 4.sup.th exemplary embodiment

[0128] The fourth exemplary embodiment considers a banknote with fibers distributed uniformly over the banknote, such as e.g. the banknote 90 of currency C in FIG. 4a.

[0129] For evaluation purposes, once again the fibers are firstly localized in the luminescence image recorded by the optical sensor 25 and the number N of fibers in each of the partial images 11,12,21,22 is determined.

[0130] In contrast to the third exemplary embodiment, however, the expected value E for the number of fibers is not stored in the sensor, i.e. is not predefined, but rather is determined only in the course of verification of the respective banknote. A relative verification of the number of fibers of at least two different partial images is carried out, wherein the number of fibers of at least one specific partial image defines the expected value E for the other partial image(s).

[0131] For example, in the case of the banknote 90 of currency C from FIG. 4c, the fiber number of the partial image 11 (e.g. N=18) can predefine the expected value E for the other partial images 12, 21 and 22, i.e. E=18. The fiber number N of the partial images 12, 21, 22 is compared with E=18. In the case of a difference of more than the acceptable fluctuation range S (with S=4, e.g. a fiber number in the range 14-22 is expected), the banknote to be verified is classified as suspected counterfeit. This then results in the following for the counterfeit 65 from FIGS. 4b, 4d: [0132] Partial image 11: N=18, is defined as expected value E, [0133] Partial image 12: 10 fibers, suspected counterfeit, [0134] Partial image 21: 17 fibers, not suspected counterfeit, [0135] Partial image 22: 9 fibers, suspected counterfeit.

[0136] In the case of uniformly distributed fibers and partial images having identical areas, such as the partial images 12, 12, 21 22 from FIGS. 4c, 4d, the expected value used for one of the partial images can also be the average fiber number of the other partial images 12, 21 and 22: in this regard, e.g. the average fiber number of the partial images 12, 21 and 22 can be used as expected value for partial image 11, which results in E=12 in the above example. With a fluctuation range of S=4, a fiber number of 8-16 is expected. The deviation of the fiber number in partial image 11 (N=18) is then categorized as suspicious.

[0137] In an additional verification, the average fiber number of the partial images 11, 12 and 21 can be used as expected value for the partial image 22, which results in E=15 in the above example. With a fluctuation range S=4, a fiber number of 11-19 is expected and the deviation of partial image 22 (N=9) is categorized as suspicious.

[0138] 5.sup.th Exemplary Embodiment

[0139] A banknote having luminescent fibers is considered in the fifth exemplary embodiment. In the value document processing apparatus 1, two luminescence images of the banknote are recorded from opposite sides and the fibers imaged there are localized.

[0140] As an example, the banknote 70 of currency B from FIG. 2a is considered, which has luminescent fibers only in its left-hand half. In the course of the verification in the value document processing apparatus 1, the optical sensor 25 records a luminescence image of the banknote front side of the banknote 70, cf. FIG. 5a, and the optical sensor 29 situated opposite simultaneously records a luminescence image of the banknote rear side of the banknote 70, cf. FIG. 5b.

[0141] Since the luminescence light is usually greatly attenuated by the value document substrate, on the respective side normally only the luminescence light of the fibers lying on or directly beneath the value document surface is capturable, but not the luminescence light of the fibers located deeper in the substrate or on the other side of the value document. The luminescent fibers of the front side are therefore recognizable only in the luminescence image of the banknote front side, but not in the luminescence image of the banknote rear side, and vice versa.

[0142] The verification of both opposite luminescence images enables a particularly accurate verification of the banknote on both sides, which involves determining the local fiber characteristic value for two mutually corresponding partial images of the opposite sides of one and the same banknote section.

[0143] In the front side image of the banknote 70, e.g. the number N1 of luminescent fibers is determined in a partial image 11, cf. FIG. 5a. In the rear side image of the banknote 70, a partial image 11 is defined which is congruent with the partial image 11 of the front side image and covers exactly the same banknote section as the partial image 11 of the front side. The partial images 11 and 11 thus show the front and rear sides of one and the same banknote section. For example, the two partial images 11, 11 lie in the white field of the banknote 70. A number N2 of luminescent fibers is determined for the partial image 11 of the rear side.

[0144] The sum of luminescent fibers of both partial images 11, 11 yields the total number (N.sub.tot=N1+N2) of fibers contained in total in the respective first partial image 11 and in the respectively corresponding second partial image 11. In the case of the authentic banknote 70, e.g. a sum of 20 fibers is expected (i.e. expected value E=20) for the sum of the luminescent fibers. The banknote that is verified is then classified as suspected counterfeit or not suspected counterfeit, depending on the total number N.sub.tot.

[0145] If, in the case of a composed counterfeit, for example, in the left-hand part of the banknote 70, for example, the authentic banknote was replaced by a piece of paper without luminescent fibers, then a total number of N tot=0 luminescent fibers would be correspondingly determined both in the partial image 11 and in the partial image 11 and the composed counterfeit would thus be segregated. The total number N tot of fibers can be compared with the expected value E=20 and can be classified as suspected counterfeit in the case of an excessively large deviation from E (more than the acceptable fluctuation range S), and as not suspected counterfeit in the case of only little deviation. However, it is also possible to use other local fiber characteristic values for verifying the partial images 11, 11 corresponding to one another.

[0146] 6.sup.th Exemplary Embodiment

[0147] The sixth exemplary embodiment considers the composed counterfeit 65 from FIG. 4b, which is a counterfeit of the banknote 90 from FIG. 4a, wherein one part 92 of an authentic banknote is combined with a counterfeit part 63 with a lower fiber density.

[0148] For evaluation purposes, once again the fibers are firstly localized in the luminescence image recorded by the optical sensor 25. The luminescence image is then scanned systematically in a grid of grid points R. In the example in FIG. 6a, the grid points R are distributed over the banknote in a grid comprising 3 rows and 9 columns.

[0149] As a local fiber characteristic value of the respective grid point, the distance a from the respective grid point R to the closest fiber in each case is used, cf. FIG. 6a, and is compared with a reference distance chosen depending on the expected density value DE of the fibers. Those grid points whose distance a is greater than the reference distance are categorized as conspicuous. The value document is then categorized as suspected counterfeit e.g. depending on whether one or more conspicuous grid points are found in the banknote image, or e.g. depending on whether one or more conspicuous regions having a plurality of conspicuous (e.g. a minimum proportion of conspicuous) grid points are found. The classification of the value document as suspected counterfeit can also be effected depending on the position of the conspicuous grid points or regions on the value document.

[0150] In the example of the composed counterfeit 65 from FIG. 6a, the two columns of grid points on the right each have many grid points that are conspicuous in this regard, and form a conspicuous region. The composed counterfeit 65 is therefore classified as suspected counterfeit.

[0151] 7.sup.th exemplary embodiment

[0152] The seventh exemplary embodiment is also based on the grid points R described in the sixth exemplary embodiment, cf. FIG. 6b.

[0153] For the purpose of evaluating the luminescence image, each localized fiber is assigned an (areal) effect region W, cf. the areas marked in gray in FIG. 6b. The effect region W is a region around the respective fiber which is defined only for the purpose of the evaluation, and which is not physically present on the value document, however. In the example shown, circular effect regions W are defined around the respective fiber, which is situated within (e.g. in the center of) the effect region W assigned to it. The size or area of the effect region W is preferably defined depending on an expected density value DE of the fibers. By way of example, the area F of the effect regions W is chosen by means of the formula F=K/DE, where K is a numerical factor. For example, with a numerical factor K=1, the effect region area can be chosen with a magnitude such that, in the case of a fiber density that is equal to the expected density value, the sum of the areas of all the effect regions corresponds to the banknote area.

[0154] By way of example, the association (yes/no) of the respective grid point R with one or more of the effect regions W around the fibers is used as a local fiber characteristic value. For each grid point, a check is thus made to establish whether or not said grid point lies in one or more of the effect regions W. If it does, the respective grid point is classified as associated with an effect region, and if it does not, the respective grid point is classified as not associated with an effect region.

[0155] This is then followed by searching for one or more conspicuous regions of the value document image in which the number or the proportion of grid points which were classified as not associated with an effect region is greater than expected. In the example from FIG. 6a, in the two columns on the right that lie in the counterfeit part 63 of the composed counterfeit 65, the proportion of grid points which were classified as not associated with an effect region is in each case 3 out of 4 grid points, corresponding to 75%. This proportion is compared e.g. with a reference proportion (e.g. 50%) and, in the event of the reference proportion being exceeded, the respective column of the grid is categorized as conspicuous. The reference proportion can be predetermined or settable or can be determined depending on the proportions of grid points not associated with an effect region in other banknote regions. For example, the proportion of the grid points not associated with an effect region in the respective right-hand columns can be compared with the mean proportion of grid points not associated with an effect region that is ascertained for a plurality of the left-hand columns of the grid.

[0156] The value document can then be classified as suspected counterfeit depending on whether one or more conspicuous columns are found. In this regard, the composed counterfeit 65 is classified as suspected counterfeit owing to the two conspicuous right-hand columns.

[0157] 8.sup.th exemplary embodiment

[0158] The eighth exemplary embodiment is also based on the grid points R described in the sixth exemplary embodiment and on the effect regions W used in the seventh exemplary embodiment.

[0159] Each of the grid points R is assigned an (areal) surrounding region 111-119, 121-129, 131-139, 141-149 around the respective grid point, said surrounding region corresponding to a partial image of the banknote image, cf. FIG. 6c. The surrounding regions 111-119, 121-129, 131-139, 141-149 overlap one another. The grid points R lie e.g. in the center of the respective surrounding region.

[0160] That proportion of the respective surrounding region around the respective grid point which is constituted by the area (gray or dark gray) covered by at least one effect region W or which is constituted by the area (white) not covered by at least one effect region is used as a local fiber characteristic value. For this purpose, in the respective surrounding region, it is possible to identify coverage regions (gray or dark gray in FIG. 6c) which belong to the effect region of at least one of the fibers (these may be fibers within or outside the respective surrounding region, the effect region of which fibers lies partly in the surrounding region). Alternatively or additionally, in the respective surrounding region, it is possible to identify free regions (white) which do not belong to an effect region of at least one of the fibers. Alternatively or additionally, in the respective surrounding region, it is also possible to identify overlap regions in which the effect regions of (adjacent) fibers overlap (dark gray).

[0161] On the basis of the size of the respective coverage region and/or free region and/or overlap region, a local fiber characteristic value of the respective surrounding region is determined. The local fiber characteristic value used can be e.g. [0162] that area proportion of the respective surrounding region which is constituted by the (gray) coverage regions (e.g. percentage indication), or [0163] that area proportion of the respective surrounding region which is constituted by the (white) free regions (e.g. percentage indication), or [0164] that area proportion of the respective surrounding region which is constituted by the (dark gray) overlap regions (e.g. percentage indication), or [0165] the area content of the coverage regions and/or the area content of the free regions and/or the area content of the overlap regions.

[0166] The value document is then classified as suspected counterfeit depending on one or more of these proportions or area contents which were ascertained for the examined surrounding regions of the value document.

[0167] For example, the respective proportion or area content of the respective surrounding region can be compared with a reference value and, in the event of the reference value being exceeded, the respective surrounding region can be categorized as conspicuous. In the case of the composed counterfeit 65, e.g. the surrounding regions 118, 119, 128, 129, 138, 139, 147, 148, 149 are conspicuous in this regard. The value document can then be classified as suspected counterfeit depending on whether one or more conspicuous surrounding regions are found.

[0168] Alternatively, it is also possible to determine the mean value of the proportions or area contents of all the surrounding regions of the value document image and to evaluate the dispersion of the proportions or area contents around said mean value. In the event of excessively great dispersion, the value document would be segregated as suspected counterfeit.