AU2017332227A1

AU2017332227A1 - Method and device for detecting color fading on a value document, in particular a banknote, and value-document processing system

Info

Publication number: AU2017332227A1
Application number: AU2017332227A
Authority: AU
Inventors: Norbert Holl; Shanchuan Su
Original assignee: Giesecke and Devrient Currency Technology GmbH
Current assignee: Giesecke and Devrient Currency Technology GmbH
Priority date: 2016-09-22
Filing date: 2017-09-13
Publication date: 2019-04-18
Anticipated expiration: 2037-09-13
Also published as: EP3516634A1; WO2018054529A1; DE102016011417A1; EP3516634B1; AU2017332227B2; CA3035917A1; CA3035917C

Abstract

The invention relates to a method for detecting color fading on a value document, in particular a banknote, on the basis of a captured image (4) of the value document, which image has a plurality of pixels, each pixel being assigned at least one intensity value (P

Description

Method and device for detecting color fading on a value document, in particular a banknote, and value-document processing system [0001] The invention relates to a method and an apparatus for recognizing ink abrasions on a value document according to the independent claims as well as a value-document processing system.

[0002] For value documents in circulation, in particular bank notes, it can happen that the printed image imprinted on a value document or other color elements are removed at least in certain places or partly, or are changed, for example by fading, wearing away or washing out.

[0003] These so-called ink abrasions, which are also designated as ink wear, can occur in particular in value documents in which ink, in particular printing ink, is applied on a polymer substrate. If such a bank note is subjected to certain mechanical stresses, for example by folding or rubbing, it can happen that the color located on the bank note separates from these places at least partly.

[0004] Upon the automatic check of the state, the so-called fitness, of bank notes it is desirable to also check these with regard to possible ink abrasions.

[0005] The object of the invention is to state a method as well as a corresponding apparatus which makes a simple and reliable recognition of ink abrasions on a value document, in particular a bank note, possible.

[0006] In a method according to the invention, ink abrasions on a value document, in particular a bank note, are recognized by means of a captured image of the value document, which has a plurality of pixels which are associated respectively with at least one intensity value, wherein for each pixel of a selection of pixels and/or for each pixel in at least a partial region of the image the following steps are carried out: establishing an intensity value from intensity values which are associated with the pixel and pixels in the vicinity of the pixel; computing a contrast value by means of intensity values which are associated with the pixel and pixels in the vicinity of the pixel; and rating the respective pixel with regard to an ink abrasion of the value document in dependence on the respectively established intensity value and computed contrast value.

[0007] An apparatus according to the invention for recognizing ink abrasions on a value document, in particular a bank note, by means of a captured image of the value document, which has a plurality of pixels which are associated respectively with at least one intensity value, said apparatus having an evaluation unit which is devised to carry out the following steps for each pixel of a selection of pixels and/or for each pixel in at least a partial region of the captured image: establishing an intensity value from intensity values which are associated with the pixel and pixels in the vicinity of the pixel; computing a contrast value by means of intensity values which are associated with the pixel and pixels in the vicinity of the pixel; and rating the respective pixel with regard to an ink abrasion of the value document in dependence on the respectively established intensity value and computed contrast value.

[0008] A value-document processing system according to the invention has: a sensor apparatus for capturing an image of a value document, in particular of a bank note, an apparatus according to the invention for recognizing ink abrasions on the value document; and an apparatus for processing, in particular conveying and/or sorting and/ or counting, value documents in dependence on ink abrasions recognized on the respective value document.

[0009] An aspect of the invention is based on the approach of rating or classifying individual pixels of the captured image of the value document with consideration of a local brightness as well as a local contrast of the image in the region of the respective pixel with regard to a possible ink abrasion of the value document.

[0010] For this purpose, for each pixel which is located in at least a partial region of the image and/or belongs to at least a selection of pixels from the image, an intensity value representing the local brightness is established from the intensity values which are associated with the respective pixel as well as other pixels in the vicinity of the pixel. In the process, the established intensity value can be identical to an intensity value which is associated with the pixel or one of the pixels in the vicinity of the pixel. Alternatively, the established intensity value can also be a value which is derived from two or more intensity values which are associated with the pixel and/ or the pixels in the vicinity of the pixel.

[0011] In a corresponding manner, a contrast value characterizing the local contrast of the image in the region of the respective pixel is computed from the intensity values which are associated with the pixel and/or the pixels in the vicinity of the pixel.

[0012] Upon the classification of the respective pixel with regard to a possible ink abrasion of the value document, the intensity value established for this pixel as well as the contrast value computed for this pixel is then considered. Among other things, it is therefore achieved that e.g. pixels in relatively bright regions of the image with simultaneously relatively low contrast values as well as pixels in relatively dark regions of the image with simultaneously relatively high contrast values are reliably rated as pixels which suggest an ink abrasion at the corresponding places of the value document.

[0013] Advantageously, different regions of the value document, in particular IR regions, do not need to be treated separately in this connection. Further, a lower number of reference value documents is required for the adaptation of the method or the apparatus than for known methods of ink-abrasion recognition. The effort for any post-adaptation of the method or the apparatus is also lower. Not least, the influence of quality fluctuations of value documents on the reliability of the ink abrasion recognition is also reduced.

[0014] Altogether the invention therefore makes a simple and reliable recognition of ink abrasions on value documents, in particular bank notes, possible.

[0015] A selection of pixels of the image as intended by the invention can be a portion of all pixels of the image yet also all or substantially all pixels of the captured image.

[0016] The pixels for which respectively an intensity value is established, a contrast value is computed and finally a rating is performed with regard to an ink abrasion of the value document, can lie in a partial region yet also in the total area of the captured image. In the process, the method can be carried out for all or substantially all pixels lying in the partial region or in the total area of the image, but also only for respectively a selection of pixels from the partial region or the total area.

[0017] A selection of pixels is preferably acquired by a two-fold or n-fold subsampling of the partial region or the total area of the captured image, for which only every second or n-th pixel is selected in both dimensions of the value document. In this manner the computing time for the carrying out of the method can be reduced without the recognition or classification of the pixels being significantly impaired with regard to possible ink abrasions of the value document.

[0018] Preferably, the established intensity value corresponds to the respectively highest intensity value from the intensity values which are associated with the pixel and/ or the pixels in the vicinity of this pixel. The brightness considered upon the rating of the respective pixel is in this connection represented by the highest intensity value of the pixels to be found in the region of the respective pixel. In the process, the rating of the respective pixel with regard to a possible ink abrasion of the value document is especially reliable. Alternatively it can also be preferred, depending on the case of application, that the established intensity value corresponds to the respectively lowest intensity value from the intensity values which are associated with the pixel and/or pixels in the vicinity of this pixel.

[0019] In a particularly preferred execution the established intensity value corresponds to a mean value which is formed from the intensity values which are associated with the pixel and pixels in the vicinity of the pixel. The intensity value established in this connection therefore corresponds to a medium brightness of the image of the value document in the region of the respective pixel whose rating with regard to a possible ink abrasion of the value document is therefore especially reliable.

[0020] It is further preferred that the establishment of the intensity value and/or the computation of the contrast value are effected by means of intensity values which are associated with the pixel and/or the pixels immediately adjacent to this pixel. In case of rectangular, in particular quadratic, detector elements upon the capture of the image of the value document, the pixels of the image are likewise represented by rectangular or quadratic pixels. The pixels whose intensity values are considered upon the establishment of a maximum, minimum or medium intensity value and/or the computation of a contrast value form in this connection preferably a group of 3 x 3 pixels, wherein in the middle of the group lies respectively the pixel to be rated, with the remaining eight pixels bordering it. The latter thus form the immediate vicinity of the pixel.

[0021] Upon the establishment of the maximum, minimal or medium intensity value and/or the computation of the contrast value, preferably respectively a so-called local operator is employed by whom not only the intensity value associated with the respective pixel is taken into consideration, but also the intensity values of adjacent pixels or pixels from the vicinity. The local picture detail is hence also designated as operator window, which preferably comprises 3x3 pixels. However, in principle it is also possible to define greater operator windows having for example 5x5, 7x7, 9x9 or 11 x 11 pixels. The number of pixels in the operator window is preferably oddnumbered, so that the respectively viewed pixel lies in the middle of the operator window.

[0022] By considering the pixels immediately adjacent to the respectively viewed pixel or the pixels lying in a defined operator window upon the establishment of the maximum, minimal or medium intensity value and/or the computation of the contrast value, an especially reliable rating of the respective pixel can be performed in this connection by means of the obtained values with regard to an ink abrasion of the value document.

[0023] The computed contrast value is preferably a measure for a relative intensity change in the respectively viewed region of the image. Preferably, the contrast value is computed with consideration of a highest intensity value and a lowest intensity value, wherein the highest and lowest intensity value is established from the intensity values which are associated with the pixel and the pixels in the vicinity of the pixel. Preferably, the contrast value corresponds to a quotient from a difference between the maximum and minimal intensity value in the numerator and a further intensity value in the denominator, wherein the further intensity value preferably corresponds to the sum of the maximum and minimal intensity value. Alternatively, the further intensity value can also correspond to a mean value formed from the intensity values which are associated with the pixel and pixels in the vicinity of the pixel. Depending on the case of application, the respectively most informative contrast value can be computed in this connection and be considered together with the established intensity value, in particular the formed mean value, upon the rating of the respective pixel with regard to an ink abrasion of the value document, by which this has an especially high reliability.

[0024] Preferably, a rating parameter is computed by means of the established intensity value, in particular of the mean value, and the contrast value, and the rating of the respective pixel is effected in dependence on the result of a comparison of the rating parameter with a comparative value. In this way, the number of parameters to be compared upon the rating is reduced from originally two, namely the established intensity value or mean value and the computed contrast value, to only one rating parameter, so that a classification of the respective pixel is effected in a reliable and especially simple manner by a simple comparison of the computed rating parameter with a single comparative value.

[0025] Preferably, the rating parameter is computed by means of a distinguishing function with respectively a variable for the established intensity value, in particular the mean value, and the contrast value. The distinguishing function can preferably be a linear function, a step function yet also a polynomial, in particular a polynomial of the second degree. The respective parameters of the distinguishing function, which preferably are constant summands, subtrahends and/or factors, can preferably be fixed within the scope of a test effected in advance or an adaptation of the method or the apparatus. Depending on the case of application, such a distinguishing function allows an especially reliable distinction of pixels for which at the corresponding places of the value document an ink abrasion is present, and such pixels for which this is not the case.

[0026] Preferably it can be provided that from the pixels rated with regard to an ink abrasion of the value document those pixels are removed which lie in the region of an edge of the value document and/or an element characterizing the value document, in particular an element imprinted on the value document. Therefore, upon the classification of the value document pixels are reliably excluded for which the probability is comparatively high that these are falsely identified as pixels lying in the region of an ink abrasion. Accordingly, a classification of the value document on the basis of the pixels cleared in this manner is particularly reliable.

[0027] Preferably, there is established for the pixels rated with regard to an ink abrasion of the value document, where applicable after removing pixels lying in the region of an edge of the value document and/or pixels in the region of an element applied to the value document, respectively a measure for the ink abrasion at the corresponding place of the value document with consideration of at least one of the following parameters: region of the ink abrasion, brightness of the ink abrasion, contrast of the ink abrasion, location of the ink abrasion. Preferably, the measure of the ink abrasion is computed with the help of a linear function in which one or more of the parameters mentioned hereinabove form variables. The number of parameters as variables of the linear function as well as the appurtenant coefficients, in particular factors, can be fixed or established, for example upon test runs with value documents of known properties. The establishment of a measure characterizing the ink abrasion on the value document provides additionally information about the extent and/or the relevance of a recognized ink abrasion, so that an even more exact classification of the value document as fit or on account of ink abrasions as unfit is made possible.

[0028] It is preferred further that the captured image of the value document is a transmission image, in particular a dark-field transmission image. A rating of the respective pixel can be performed especially reliably by employment of the intensity values established by means of a transmission image or computed contrast values with regard to an ink abrasion of the value document. This holds particularly for dark-field transmission images.

[0029] Further advantages, features and application possibilities of the present invention result from the subsequent description in connection with the figures.

There are shown:

Fig. 1 an example of a value-document processing system having an apparatus for recognizing ink abrasions on value documents;

Fig. 2 an example of a transmission image of a bank note;

Fig. 3 a diagram with a first and second example of a distinguishing function;

and

Fig. 4 an example of an image derived from the transmission image shown in

Figure 2, having pixels which were rated as being pixels involving an ink abrasion of the bank note.

[0030] Fig. 1 shows an example of a value-document processing system having an apparatus 10 for recognizing ink abrasions on a value document 1, which is in particular a bank note, and an only schematically indicated apparatus 30 for processing, in particular conveying and/or sorting and/or counting, value documents, in particular in dependence on ink abrasions recognized on the respective value document 1.

[0031] The apparatus 10 has a radiation source 2 which is designed to generate electromagnetic radiation with which the bank note 1 is irradiated. Preferably the electromagnetic radiation is infrared radiation.

[0032] Further, an image sensor 3 is provided which is designed to capture in a spatially resolved manner electromagnetic radiation emanating from the bank note 1. In the arrangement shown in the present example of radiation source 2 and image sensor 3, the image of the bank note 1 captured by image sensor 3 is a so-called darkfield transmission image. Alternatively, it is however also possible to arrange the radiation source 2 relative to the image sensor 3 such that the latter captures a brightfield transmission image or a remission image.

[0033] The image of the bank note 1 captured by image sensor 3 has a plurality of pixels which respectively are associated with at least one intensity value which represents a measure for the intensity of the electromagnetic radiation transmitted or reflected by a plurality of places on the bank note 1 and captured by the image sensor

3.

[0034] The image data 4 generated by the image sensor 3, which contain the intensity values associated with the individual pixels, are relayed to an evaluation unit 5 and analyzed there in particular with regard to possible ink abrasions on the bank note 1 and are rated correspondingly.

[0035] Preferably, the analysis unit 5 is designed to control the apparatus 30 for processing bank notes such that bank notes 1 are outputted in different output pockets 31 and 32 in dependence on the result of the rating or classification with regard to an ink abrasion.

[0036] The analysis of the image data 4 with regard to possible ink abrasions of the bank note 1 and its corresponding rating or classification is explained in more detail hereinafter.

[0037] Fig. 2 shows an example of dark-field transmission image 4 of a polymer bank note which typically has a transparent substrate foil which is furnished with a primer of white lacquer. On the primer, a printed image, which is also designated as a background printed image, is imprinted for example using offset printing. Alternatively or additionally, for example digits and/or infrared-sensitive regions are imprinted using intaglio printing.

[0038] On account of the described typical construction of polymer bank notes, ink abrasions and other defects become apparent differently in dark-field transmission image 4. For example places 6, in which the white-lacquer primer is partly removed, appear very dark, whereas ink abrasions in printed regions appear as a rule as relatively bright places 7, because there the electromagnetic radiation striking obliquely from below is scattered in the white-lacquer primer. Holes and tears in the bank note usually become apparent in the dark-field transmission image 4 as relatively dark regions 8, because there no scattering of the electromagnetic radiation occurs and this is therefore not captured by the image sensor 3.

[0039] To now automatically establish the pixels lying in the regions 6 and 7 of image

4, which correspond to the places on the bank note having an ink abrasion, there is first established for each pixel of image 4 or at least for each pixel of a selection of pixels from image 4, for example in the region of a certain security feature, a mean value m as well as a contrast value c with consideration of the intensity values of the respective pixel as well as the pixels surrounding this pixel.

[0040] As is illustrated by means of an enlarged detail of a region 9 of image 4, altogether nine pixels are contained in region 9, wherein the pixel (hatched) for which the mean value m as well as the contrast value c is computed lies in the middle of region 9. Preferably the mean value m is the arithmetic mean, which is computed as follows:

m = (1) [0041] Po to Ps designate the intensity values of the observed pixel and the adjacent pixels.

[0042] It is further preferred that the contrast value c corresponds to the so-called Michelsen contrast, which is computed as follows:

maxi RI — minfPA

C =----^V ί ^G (0,8) max{Pj + minfPj (2) [0043] That is to say, the contrast value c corresponds to the quotient from the difference between the highest intensity value and the lowest intensity value on the one hand and the sum of the highest intensity value and the lowest intensity value on the other hand, namely respectively from the intensity values Po to Ps in the viewed region 9.

[0044] In principle it is possible to compute a corresponding mean value m and contrast value c for all pixels of image 4. To save computing time without losing relevant information at the same time, however, it can be advantageous to establish corresponding values m and c only for every second pixel using subsampling along the two dimensions of image 4, with consideration of the respective vicinity of the pixel, which is illustrated by way of example in Fig. 2 by means of the enlarged region 11 of image 4. Respectively a mean value m and a contrast value c are computed for the hatched pixels.

[0045] In a further step, a first rating, which can also be designated as a preliminary distinguishing, is carried out by means of the mean value m and contrast value c established per pixel to the effect whether the respective pixel suggests an ink abrasion at the corresponding place of the bank note.

[0046] This is carried out preferably by computing a rating parameter f with the help of a distinguishing function f (m, c) which is defined as follows:

f(rn, c')= c — a₀ + a_±X m¹ + a₂ X m²

[0047] In present example, the preliminary distinguishing function f(m, c) is a polynomial of the second degree in the variables m and c, wherein the parameters ai with i - 0,1 and 2 of the polynomial can be fixed or estimated by an adaptation carried out in advance of the method or the apparatus.

[0048] If the rating parameter f for a pixel computed by means of the distinguishing function f(m, c) is greater than or equal to a comparison value, which is zero in the present example, this pixel is rated as a so-called ink wear candidate, i.e. the pixel then possibly corresponds to a place on the bank note having an ink abrasion.

[0049] This is illustrated by means of a first example (dashed) shown in Fig. 3 of a course of the contrast c over the mean value m which forms the boundary between pixels which were not rated as an ink wear candidate (region below the dashed line) and pixels which were rated as an ink wear candidate (region above the dashed line). Ink wear candidates which correspond to an ink abrasion in the region of the intaglio printing typically have a relatively low medium intensity m and a relatively high contrast c and accordingly lie in the left region of the diagram above the dashed line. Ink wear candidates which correspond to an ink abrasion in the region of the water mark of the bank note have a relatively high medium intensity m and simultaneously a comparatively low contrast c and therefore lie in the right region of the diagram above the dashed line. Ink wear candidates which arise from an ink abrasion in the offset printing region of the bank note typically have a medium brightness as well as a medium contrast c and accordingly lie in the middle region of the diagram above the dashed line.

[0050] Depending on the case of application, other functions can in principle also come into consideration as a distinguishing function, for example, a linear function or a step function for which the second example (dot-dashed) shown in Fig. 3 of a course of the contrast c is obtained versus the mean value m. Pixels which on account of the respectively computed mean value m and contrast value c lie above the dotdashed course are rated as ink wear candidates, whereas pixels lying below the course are rated as non-ink wear candidates. Otherwise, the above explanations regarding the first example of the course hold accordingly.

[0051] In an optional further step it can be provided to check the pixels rated as ink wear candidates to the effect whether these lie in certain regions of the bank note, such as in the region of the edge of the bank note and/or in the region of certain print elements. Because pixels from corresponding image regions of image 4 can in some cases be falsely rated as ink wear candidates, it can be provided to remove these pixels from the group of the pixels rated as ink wear candidates. Because in this connection the ink wear candidates are checked or validated by means of their position in image 4 of the bank note as to their informative value, this step can also be viewed as a validation step or validation.

[0052] In a further optional step it can be provided to set the intensity value I(x, y) of the respective pixel (x, y) to zero, if the distinguishing function f(m, c) calculated from the appurtenant mean value m and contrast value c is smaller than zero, and to set to a certain intensity value, for example the highest intensity value I_max or max{Pi} established in the region of the respective pixel, if the distinguishing function is greater than or equal to zero:

!= (*- Υλ werm f(m, c) > Ο = 0 , we rm f (m, c) < 0 [0053] In this manner, all pixels rated as ink wear candidates are highlighted especially readily visibly vis-a-vis a black background.

[0054] In a further optional step, it can be provided to scale the image of ink abrasions obtained as described hereinabove with regard to brightness by associating a constant, in particular maximal, intensity value to all pixels having an intensity different from zero, so that the finally obtained image represents a black-and-white image in which the pixels which correspond to ink abrasions at the corresponding places of the bank note are represented as white regions in front of a black or very dark background.

[0055] Such an image is shown in Fig. 4 by way of example. The regions 6 and 7 drawn in Fig. 1, which correspond to places on the bank note having ink abrasion, are very well recognizable as white regions in front of a homogeneous dark background, whereas regions 8 (see Fig. 1) which stem from other defects in the bank note, such as tears or holes, do not appear. Likewise no more visible are printed regions and/or regions furnished with a watermark (see Fig. 1).

[0056] In an optional further step, which is also designated as a classification, it can be provided that for all pixels rated as ink wear candidates, where applicable after a validation with removal of pixels lying on the edge and/or in the region of certain print elements, a measure (the so-called ink wear level) is derived for the respective ink abrasion at the corresponding place of the bank note. This can preferably be effected by the fact that for each of these pixels an ink wear level (InkL) is computed by means of the following linear function:

InkL(i) = α-j X InkA + X InkB + c_t X InkC + d, X InkP (4) wherein ai states in which region of the image of the bank note the pixel rated as an ink wear candidate lies, and bi states a measure for the brightness, Ci a measure for the contrast and di the location of the pixel rated as ink wear candidates. The appurtenant coefficients InkA, InkB, InkC and InkP can be computed by test runs carried out in advance by means of bank notes with known properties. Likewise it can be fixed or configured which or how many of the hereinabove mentioned information items ai, bi, Ci or di are taken into consideration upon the computation of the ink wear levels.

[0057] Alternatively to the above-described determination of the respective contrast value c by considering the local vicinity of the respective pixel using a so-called local operation or a local operator, contrast values can also be established with the help of a histogram in which the quantity or frequency of pixels is plotted versus their intensity. Certain regions of the bank note in which an ink abrasion is present in the form of a wear or smearing of the printing ink can be identified by the fact that in the respective histogram a transition region occurs between a region of bright pixels and a region of dark pixels, in which only very few pixels can be found. This transition region depends on the respective state of the bank note. The respective transition region can be established by employment of a dynamically computed threshold in a histogram. By means of the respectively lowest and highest intensity value in this transition region, a corresponding contrast value c can then be established by means of equation (2).

[0058] Preferably, the method or the apparatus is also designed to analyze or classify bank notes 1 which, where applicable, are slanted or pass by the image sensor 3 obliquely, as to any ink abrasions. For this purpose preferably upon a subsampling a skew correction is carried out according to the Bresenham algorithm, thereby minimizing possible rounding errors which arise from a discretization of continuous coordinates.

Claims

Claims

1. A method for recognizing ink abrasions on a value document (1), in particular a bank note, by means of a captured image (4) of the value document (1), which has a plurality of pixels which are associated respectively with at least one intensity value (Pi), wherein for each pixel of a selection of pixels of image (4) the following steps are carried out:

a) establishing an intensity value (Pi, m) from intensity values (Po - Ps) which are associated with the pixel and pixels in the vicinity of the pixel,

b) computing a contrast value (c) by means of intensity values (Po - Ps) which are associated with the pixel and pixels in the vicinity of the pixel, and

c) rating the respective pixel with regard to an ink abrasion of the value document (1) in dependence on the respectively established intensity value (Pi, m) and computed contrast value (c).
2. The method according to claim 1, wherein the established intensity value (Pi) corresponds to the highest intensity value (max{Pi}) of the intensity values (Po - Ps) which are associated with the pixel and pixels in the vicinity of the pixel.
3. The apparatus according to claim 1, wherein the established intensity value (m) corresponds to a mean value (m) which is formed from the intensity values (Po - Ps) which are associated with the pixel and pixels in the vicinity of the pixel.
4. The method according to any of the preceding claims, wherein the establishment of the intensity value (Pi, m) and/or the computation of the contrast value (c) is effected by means of intensity values (Po - Ps) which are associated with the pixel and the pixels immediately adjacent to this pixel.
5. The method according to any of the preceding claims, wherein from the intensity values (Po - Ps), which are associated with the pixel and the pixels in the vicinity of the pixel, the highest intensity value (max{Pi}) and the lowest intensity value (min{Pi}) is established and the contrast value (c) is computed from the highest intensity value (max{Pi}) and lowest intensity value (min{Pi}).
6. The method according to any of the preceding claims, wherein by means of the established intensity value (Pi, m), in particular of the mean value (m), and the contrast value (c), a rating parameter (f) is computed and the rating of the respective pixel is effected in dependence on the result of a comparison of the rating parameter (f) with a comparative value.
7. The method according to claim 6, wherein the rating parameter (f) is computed by means of a distinguishing function (f(m, c)) with respectively one variable for the established intensity value (Pi, m), in particular the mean value (m), and the contrast value (c).
8. The method according to claim 7, wherein the distinguishing function (f(m, c)) is given by a linear function, a step function or a polynomial, in particular a polynomial of the second degree.
9. The method according to any of the preceding claims, wherein from the pixels rated with regard to an ink abrasion of the value document (1) those pixels are removed which lie in the region of an edge of the value document (1) and/or in the region of an element applied to, in particular imprinted on, the value document (1).
10. The method according to any of the preceding claims, wherein for the pixels rated with regard to an ink abrasion of the value document (1) a measure for the ink abrasion is established at the corresponding place of the value document (1) based on at least one of the following parameters: region of the ink abrasion (ai), brightness of the ink abrasion (bi), contrast of the ink abrasion (ci), location of the ink abrasion (di).
11. The method according to any of the preceding claims, wherein the captured image (4) of the value document (1) is a transmission image, in particular a dark-field transmission image.
12. An apparatus for recognizing ink abrasions on a value document (1), in particular a bank note, by means of a captured image of the value document (1), which has a plurality of pixels which are associated respectively with at least one intensity value (Pi), having an evaluation unit (5) which is devised to carry out the following steps for each pixel of a selection of pixels of image (4):

a) establishing an intensity value (Pi, m) from intensity values (Po - Ps) which are associated with the pixel and pixels in the vicinity of the pixel,

b) computing a contrast value (c) by means of intensity values (Po - Ps) which are associated with the pixel and pixels in the vicinity of the pixel, and

c) rating the respective pixel with regard to an ink abrasion of the value document (1) in dependence on the respectively established intensity value (Pi, m) and computed contrast value (c).
13. A value document processing system having

- a sensor apparatus (2, 3) for capturing an image (4) of a value document (1), in particular of a bank note,

- an apparatus (10) for recognizing ink abrasions on the value document (1) according to Claim 12 and

- an apparatus (30) for processing, in particular conveying and/or sorting and/or counting, value documents in dependence on ink abrasions recognized on the respective value document (1).