Method and apparatus for inspecting value documents and method and apparatus for producing a template

Abstract

Described is a method and an apparatus for generating templates for verifying value documents of a predetermined value document type, the value documents of the predetermined value document type having at least two predetermined manufacturing elements, using digital training images of training value documents of the predetermined value document type, the training images each having pixels, pixel data being respectively assigned to the pixels. The method has the following steps: determining, for the training images, respectively, a position dataset having position coordinates in a coordinate space, the position dataset respectively describing the position of the manufacturing elements on the value document at least relative to one another, and forming at least two position sub-areas of the coordinate space, which each contain at least a predetermined number of position datasets, the position sub-areas not containing a common position dataset, determining a template for the position sub-areas respectively by using the training images of those value documents whose position coordinates lie in the position sub-area, and storing the templates and position sub-area data, which describe the position and extent of the respective position sub-area. A method and an apparatus for verifying value documents are also described.

Description

Technical Field

[0001] This invention relates to a method and apparatus for examining valuable documents, and a method and apparatus for generating examination parameters for examining valuable documents. Background Technology

[0002] Here, a valuable document is understood to be, for example, a sheet of paper representing monetary value or rights and therefore should not be arbitrarily produced by unauthorized persons. Thus, it possesses the characteristics of being difficult to produce, and especially difficult to copy, the presence of which proves its authenticity, i.e., that it was produced by an authorized institution. Important examples of this type of valuable document are chip cards, coupons, vouchers, checks, and especially banknotes.

[0003] Regarding banknotes, further distinctions can be made between different types of valuable documents; within the scope of this invention, the type of valuable document can be determined by the currency and denomination or face value of the banknote, and, where necessary, by the date of its issuance or formal publication by a bank. The following description of banknotes also applies accordingly to any other type of valuable document.

[0004] Banknotes are typically made by printing on a substrate in multiple steps or manufacturing processes, whereby a printed layer is applied to the banknote substrate in each step. The applied printed layers should have a predetermined position relative to the banknote substrate and / or relative to each other; however, this is usually only approximate, as the printed layers may shift relative to each other due to manufacturing processes. Therefore, variations in the appearance of the banknotes occur. The printed layers may partially overlap or obscure each other.

[0005] When inspecting valuable documents of a predetermined type using machines, such as verifying authenticity or condition, especially when inspecting deterioration due to stains or ink loss, a digital image of the document to be inspected is typically captured and used for inspection. Within the scope of this disclosure, a digital image is understood to include pixels and pixel data associated with each pixel. A corresponding position in the image (or a corresponding position on the valuable document) corresponds to each pixel, and the pixel data is applied to that position. The pixel data preferably reflects the brightness or color value of the corresponding pixel. During inspection, the digital image is examined to determine whether and to what extent it conforms to at least one predetermined inspection criterion, according to the inspection method used.

[0006] In the verification process, verification parameters specific to the verification method are used. These parameters determine the preset characteristics of a valuable file of a preset valuable file type. In some verification methods, the verification criteria may, for example, verify whether a pixel or a group of pixels possesses the characteristics determined by the verification parameters. Verification parameters may include templates of the valuable file of a preset valuable file type. Within the scope of this application, a template is understood to include pixels of at least one segment of a digital image of a valuable file of a preset valuable file type, and template pixel data respectively assigned to the pixels. A corresponding position in the image corresponds to each pixel, and the pixel data is applied to that position. The resolution, i.e., the relationship between the number of pixels and their area, and the arrangement of pixels are preferably consistent with the resolution of the digital image of the valuable file of the preset valuable file type to be verified. Depending on the template type, the template pixel data may include at least one individual value and / or a range of values ​​that should be considered permissible for the pixel data of the valuable file image to be verified.

[0007] For example, when inspecting for deterioration such as stains or paint peeling, the inspection standard could be whether the pixel data or pixel value of the valuable file image is within a range of allowed values ​​preset by the template. If the value is below this range, it can be determined that there is staining; if the value is above this range, it can be determined that there is paint peeling. Otherwise, it is considered that there is no deterioration at the pixel level.

[0008] These verification parameters, and especially the templates, are adapted and generated for preset valuable document types, which in the case of banknotes are given by currency, value or denomination, and, if necessary, by the issuer.

[0009] For example, templates are typically generated by averaging images from a large number of training files of a predefined type of valuable file, using the average pixel data from the training files as template pixel data. For other templates, lower and upper limits can be set for pixel intensity as template pixel data, which can be given by the minimum and maximum intensity values ​​of the corresponding pixels in the set of training files.

[0010] Therefore, the same template is used for any valuable document of the same type. However, especially when using range limits, variations in the position of the print layer cause template inaccuracies.

[0011] When inspecting valuable documents of a predefined valuable document type, the aforementioned changes in the position of the printed layers mean that these methods can only work with low accuracy because the template is not very accurate and / or must allow tolerances to compensate for the displacement between printed layers relative to each other, thereby compensating for the changes in pixel data that occur with the displacement.

[0012] Similar problems arise when manufacturing-induced fluctuations occur in the relative positions of other manufacturing elements that at least partially determine the appearance of a valuable document. Within the scope of this application, a manufacturing element (or production element) is understood to be an element of a valuable document that at least partially determines and / or influences the appearance of the document in the visible or invisible (e.g., infrared and / or ultraviolet) spectral range, particularly the printed layer, and is manufactured or applied independently of other elements, particularly the manufacturing element, and / or in a manufacturing step separate from other elements, particularly the manufacturing element. Examples of manufacturing elements include: printed layers applied to a substrate of the valuable document by means of gravure or offset printing; elements applied to a substrate by means of screen printing, such as images or symbols whose color depends on the viewing angle; or anti-counterfeiting elements that can be optically acquired (captured), such as watermarks or embedded security threads applied to a substrate; or foil applied to a substrate, for example, by means of hot stamping, which may have a hologram. For ease of understanding, the following description may refer in part to printed layers. However, these descriptions also apply accordingly to any type of manufacturing element, particularly the above-described examples of manufacturing elements. Summary of the Invention

[0013] The technical problem to be solved by this invention is to provide a method for inspecting a valuable document of a preset valuable document type having at least two manufacturing elements, and a method for providing inspection parameters for the inspection method, said inspection parameters enabling simple and accurate inspection of such valuable documents. A further technical problem to be solved by this invention is to provide means for implementing said method.

[0014] The technical problem is solved by a method, particularly a method for generating or forming templates for verifying valuable documents of a predetermined valuable document type, particularly banknotes, wherein the predetermined valuable document type has at least two predetermined manufacturing elements, particularly printing layers and / or anti-counterfeiting elements, that may partially overlap each other, wherein training images of digital data of the predetermined valuable document type are used, the training images each having pixels, and pixel data respectively belonging to the pixels. The method includes the following steps: determining position datasets for each training image, each having position coordinates in coordinate space, the position datasets describing the positions of the manufacturing elements on the valuable document at least relative to each other, and forming at least two, preferably at least four position sub-regions in the coordinate space, each of these position sub-regions containing at least a predetermined number of position datasets, wherein these position sub-regions do not contain a common position dataset. Templates are then determined for each position sub-region using the training images of the valuable document, and the templates and position sub-region data, the position sub-region data describing the position and extension dimensions of the corresponding position sub-regions, are stored. The templates and position sub-region data are stored mutually associated. The method is preferably performed computer-aided. Because the method adapts the parameters of the actual testing method by forming a template, it is also referred to below as the adjustment method.

[0015] The aforementioned technical problem is also solved by a method, particularly a method for verifying a valuable document of a predetermined valuable document type, the valuable document having at least two predetermined, if necessary partially overlapping, manufacturing elements, particularly printed layers and / or anti-counterfeiting elements, the method being implemented using at least two templates, the templates being predetermined for predetermined position sub-regions of the positions of the manufacturing elements, preferably templates generated by means of an adjustment method or method according to the invention for generating or forming templates for verifying valuable documents of a predetermined valuable document type. The method, hereinafter also referred to as a verification method, includes the following steps: providing a digital valuable document image of a valuable document of a predetermined valuable document type to be verified, the valuable document image comprising pixels, pixel data respectively assigned to the pixels; determining at least the positions of the manufacturing elements in the provided valuable document relative to each other; determining a template for the digital valuable document image based on the determined positions of the manufacturing elements and the predetermined position sub-regions; and verifying the digital valuable document image using the determined template. This method is also referred to below as a verification method. The method, hereinafter also referred to as an image verification method, is used in the sub-step of verifying the digital valuable document image using the determined template. Preferably, the template is provided in the first step of the method.

[0016] Within the scope of this invention, at least two templates and position sub-regions or position sub-region data respectively associated with said templates are used. The position sub-regions or position sub-region data associated with the respective templates define which determined positions, or at least relative positions, or position data of manufactured elements in a pre-defined valuable document type to be inspected. For the valuable document to be inspected, a suitable template for inspecting the valuable document can be determined based on the determined positions, or at least relative positions, of the manufactured elements. The position sub-regions, or the position sub-region data defining these position sub-regions, can be stored in a storage device.

[0017] The adjustment method provides a template that can be used in the verification method. In the verification method, a template determined for a valuable document image can be used to verify a valuable document or a corresponding digital valuable document image. Therefore, the adjustment method and the verification method, especially the template type and the actual image verification method, must be compatible. A preset image verification method is preferred.

[0018] The adjustment method uses training images of digital training valuable files of a preset valuable file type. These training images may exist in the form of previously acquired and, for example, stored data. Alternatively, in other embodiments, the training images may be acquired and further processed directly using suitable means. The preset valuable file type of training valuable files consists of groups of preset valuable files, which preferably at least partially have different relative positions of manufacturing elements, typically resulting from manufacturing processes. Accordingly, the training images show the corresponding variations in the relative positions of the manufacturing elements. The number of training valuable files, and therefore the number of training images, is preferably greater than twice the preset number of the location dataset.

[0019] For each training image, at least one location dataset is determined, which describes the position of the manufacturing element on the corresponding training valuable document, at least relative to each other. The position of the manufacturing element relative to each other can be determined by any method. For example, at least one so-called anchor element and anchor point reflecting its position can be predefined for each manufacturing element. The anchor element is preferably an image segment representing the manufacturing element, such as a symbol or other unique printed image segment, which exists in the image of a valuable document of a predefined valuable document type, especially the training image. Such anchor elements can be selected automatically and / or manually. In the image, the position of the anchor element or anchor point can be determined by template matching or other related methods. To determine the position of the manufacturing element, it is preferable to predefine at least two anchor elements for each manufacturing element. If the positions of the manufacturing elements are determined independently, then their relative positions can always be determined. In this sense, determining the positions of the manufacturing elements independently is equivalent to determining at least their relative positions.

[0020] In this method, the position of the manufactured element is reflected by a position dataset with position coordinates in a predefined coordinate space. For each manufactured element, coordinates associated with a reference frame of the valuable document, such as the corners and two sides of the valuable document, can be used as the position dataset. However, it is also possible to predefine only the coordinates of displacement vectors, which are used to describe the offset of the manufactured elements relative to each other. This is particularly advantageous when the predefined manufactured elements are always in the same position in the valuable document image.

[0021] The method uses a sub-region of coordinate space where a location dataset exists or the sub-region includes the location dataset. This means that the corresponding location coordinates lie within the sub-region of coordinate space. The sub-region can be determined by sub-region data describing the location and extent of the sub-region. For example, the sub-region can be given by a set of explicitly represented sub-region data. The location region is preferably given by an interval or interval boundary of location coordinates, or by a function and function parameters that define the corresponding location region.

[0022] The adjustment method determines at least two location sub-regions in the coordinate space. For this purpose, location sub-region data is determined, which describes the position and extent of the corresponding location sub-region in the coordinate space.

[0023] On the one hand, the location sub-regions are determined such that each location dataset used for training images lies within one of the formed location sub-regions. In this respect, the location sub-regions preferably completely cover the location datasets of the training images. On the other hand, the location sub-regions are selected such that they do not contain common location datasets. This means that no location dataset lies simultaneously in at least two different location sub-regions. In this respect, the location sub-regions do not overlap.

[0024] Furthermore, each location sub-region includes at least a preset number of location datasets. This means that at least a preset number of location datasets exist in each location sub-region. The preset number can be preset according to the requirements when forming the template for the location sub-regions, but it is always greater than 10. If the template is generated using statistical methods, such as by averaging, then a certain minimum precision can be achieved by selecting the number based on the minimum required precision.

[0025] Preferably, at least four, and more preferably more, location sub-regions, particularly location sub-region data, and templates associated with said location sub-regions are formed and stored. Since changes in location coordinates and consequently, changes in the manufactured components within the location sub-regions are generally smaller than changes in the training images or training files used, more accurate templates can be created. As the number of location sub-regions increases and the resulting stretching size of the location sub-regions decreases, the templates, and therefore the verification, become increasingly accurate.

[0026] A template is created for each location sub-region from the training images in the training valuable file, where the location data coordinates of the training image are contained within or located in the corresponding location sub-region. Therefore, the corresponding location sub-region and the template determined for this purpose are mutually dependent. The corresponding template and the location sub-region data belonging to the template are stored mutually dependently.

[0027] In the inspection method according to the invention, these templates and associated location sub-region data are preferably used for valuable documents of a preset valuable document type. They are preferably provided at the start of the inspection method. The templates and associated location sub-regions can be stored in the inspection apparatus used to perform the method.

[0028] This testing method provides a digital image of the valuable document to be tested. Storing the image in any location is generally sufficient. However, it is preferable to acquire the image using a suitable device and process it in real time. The valuable document image preferably has the same resolution as the training valuable document and maps the same regions of the valuable document.

[0029] The location of the manufacturing element is then determined from the image of the valuable document, wherein the location is described by a location dataset having location coordinates in a coordinate space. The coordinate space is preferably the same during template generation and verification. Otherwise, the coordinates can be transformed accordingly. The location determination can be performed using any method. For example, at least one so-called anchor element and anchor point reflecting its location can be preset for each manufacturing element. The anchor element is preferably an image segment characterizing the manufacturing element, such as a symbol or other unique printed image segment, which exists in the image of a valuable document of a preset valuable document type, especially in the training image. Such anchor elements can be selected automatically and / or manually. In the image, the location of the anchor element or anchor point can be determined using template matching or other related methods. To be able to determine the location of the manufacturing element, it is preferable to preset at least two anchor elements for each manufacturing element. It is particularly preferred to use the same location determination method used when generating the template.

[0030] Then, based on the determined location, one template is selected, which is used for further verification. This is preferably implemented as follows: based on the location sub-region data of the template, it is determined which location sub-region contains the location coordinates, and then a template belonging to that location sub-region is determined. Then, preferably, this template is used for further verification using a previously known image verification method.

[0031] The verification method preferably further includes generating and outputting a verification signal representing the verification result. Based on this verification signal, for example, further processing of the valuable document mapped in the valuable document image can be controlled.

[0032] The method is preferably performed with computer assistance.

[0033] Therefore, the aforementioned technical problem is also solved by an apparatus, particularly an apparatus for generating or forming a template for verifying a valuable document of a predetermined valuable document type, particularly banknotes, wherein the predetermined valuable document type has at least two predetermined manufacturing elements, particularly printing layers and / or anti-counterfeiting elements, which may partially overlap each other, wherein training images of training valuable documents of the predetermined valuable document type, each having pixel numbers, are used, with pixel data respectively assigned to the pixels, and the apparatus has a storage device for storing the training images of the digital data of the predetermined valuable document type, wherein the apparatus is designed to perform the adjustment method according to the invention using the training images. The apparatus preferably also has an interface through which the generated template and the positional sub-regions associated with the template can be transmitted to another apparatus, preferably a remote apparatus, such as a storage device. The interface is preferably an interface for a data network. The apparatus is preferably designed to transmit the generated template and the positional sub-regions associated with the template to another apparatus. Additionally or alternatively, the apparatus itself is preferably designed to store the generated template and positional sub-region data in the storage device and / or the storage device.

[0034] The device may therefore include a computer, preferably having at least one processor connected to a storage device via a data link. A computer can be understood herein as any data processing device.

[0035] Therefore, the technical solution of the present invention is also a computer program, which has program code, and in particular program code tools, so that when the program is executed on a computer, it performs the adjustment method according to the present invention. In the device, the computer program is preferably stored in a storage device or the computer's memory.

[0036] Therefore, another technical solution of the present invention is a computer-readable data carrier having program code that can be executed by a computer, thereby enabling the computer to perform the adjustment method according to the present invention.

[0037] The aforementioned technical problem is also solved by an apparatus for verifying valuable documents, particularly banknotes, having at least two pre-defined manufacturing elements, particularly printing layers and / or security features, which may partially overlap. The apparatus performs the verification using a template, generated particularly by an adjustment method according to the invention, and positional sub-region data belonging to the template. The apparatus includes an analysis device comprising: at least one memory for storing the template and positional sub-region data respectively belonging to the template; and an interface for providing a digital image of the valuable document. The analysis device is designed to execute the verification method according to the invention. Such an apparatus is also referred to hereinafter as an verification device.

[0038] The analysis apparatus may in particular include a computer and have at least one processor connected to a memory via a data link. Program code may be stored in a program memory within the analysis apparatus, and when the program code is executed, the testing method according to the invention is performed by the computer or processor.

[0039] Therefore, the technical solution of the present invention is also a computer program having program code so that when the program is executed on a computer, the testing method according to the present invention is performed.

[0040] Another technical solution of the present invention is a computer-readable data carrier having program code, which executes the verification method according to the present invention when the program code is executed by a computer.

[0041] The inspection apparatus can, in principle, operate independently of the apparatus for acquiring images of valuable documents. However, the inspection apparatus preferably also includes an image acquisition device for acquiring digital images of the valuable documents to be inspected, which is connected via a signal connection to an interface for providing digital images of the valuable documents. The image acquisition device preferably has a position-resolved optical sensor, such as a camera. This has the advantage of enabling direct, real-time analysis of the valuable document images in relation to their acquisition.

[0042] The testing device may also have an interface through which it can output a test signal reflecting the result of the test performed.

[0043] Therefore, the technical solution of the present invention also includes providing an apparatus, particularly for processing, particularly for inspecting and / or counting and / or classifying and / or destroying valuable documents of a predetermined type, particularly banknotes, wherein the valuable documents each have at least two predetermined manufacturing elements, particularly printing layers and / or anti-counterfeiting elements, which may partially overlap each other if necessary. The apparatus comprises: an input device for inputting individual or separately separated valuable documents to be processed; an output device having at least one output section for receiving processed valuable documents; a transport device for transporting individual or separately separated valuable documents from the input device to the output device; and an inspection device according to the present invention. The image acquisition device of the inspection device is preferably arranged at the transport path and configured such that a digital image of the valuable document to be inspected, transported from the image acquisition device, is acquired during transport and provided for use in the inspection device. This apparatus for processing valuable documents is hereinafter also referred to as a processing device. The processing device is preferably designed to control further processing of the transported valuable documents based on the results of the inspection device's inspection of the digital image of the valuable document. Valuable documents that have been transported and inspected can be classified into different sections of the output device based on the inspection results, or they can be shredded if the processing device is designed accordingly.

[0044] These methods utilize at least two templates, each specifically designed for a defined region, or location sub-region, where the fabricated component is positioned. The extended dimensions and location of the location sub-region are described by location sub-region data.

[0045] In principle, positional sub-regions can be formed in any way during the adjustment process.

[0046] In a preferred embodiment of the adjustment method, the formation of location sub-regions may include multiple segmentation steps. In each segmentation step, an existing location sub-region containing more than twice the preset number of location datasets may be segmented into a preset number of newly formed location sub-regions, each containing at least the preset number of location datasets. The existing location sub-regions may be replaced by the newly formed location sub-regions. In the first segmentation step, the region containing the location coordinates of all training valuable files is used as the location sub-region. This segmentation step is performed frequently until all formed location sub-regions meet at least one preset termination criterion. This termination criterion may be checked at the beginning or end of the segmentation step.

[0047] The advantage of this approach is that it generates sub-regions containing at least a predetermined number of location datasets, but not an excessive number. Because the number of training images is predetermined, regions in the coordinate space containing multiple location datasets are segmented more finely than other regions.

[0048] Preferably, the number of sub-regions to be segmented, i.e., the number of sub-regions generated from the current sub-region, is selected such that the sub-regions formed in each segmentation step include at least a preset number of location datasets, and at most twice the preset number of location datasets. This selection produces a uniform segmentation of the relevant parts of the coordinate space relative to the number of training images. In regions with higher-density location datasets and corresponding training data, the formed sub-regions are smaller than in other regions. Therefore, the templates in these regions are more accurate, or more precise verification can be achieved.

[0049] Given the boundary conditions described so far, the segmentation of the current location sub-region can be performed in any manner. In a preferred embodiment of the adjustment method, the coordinate space can be n-dimensional, where n>1, and segmentation can be performed in only one dimension in each segmentation step. The successive segmentation of the location sub-regions and the resulting successive segmentation of the generated location sub-regions, particularly the direct successive segmentation of the location sub-regions and the direct successive segmentation of the generated location sub-regions, are preferably performed in different spatial dimensions. That is, the segmentation of the location sub-regions can be performed in the first dimension, and subsequent segmentation of the generated location sub-regions can be performed in another dimension. The advantage of this approach is that individual segmentations can be performed simply, while overall segmentation is performed in all dimensions.

[0050] The segmentation process continues until a preset termination criterion is met. The termination criterion implicitly includes sub-criteria, meaning the number of location datasets in the resulting location sub-regions must not be less than a preset number. However, the termination criterion may also include other sub-criteria; if at least one sub-criteria is met, preferably all sub-criteria, then the termination criterion is satisfied.

[0051] In a preferred embodiment of the method, the termination criterion may include the following criterion or sub-criterion: the location sub-region contains at most a preset number of location datasets, preferably twice the preset number. This generates a template for the location sub-region, the size of which is essentially determined by the preset number of location datasets: in areas with high location dataset density, the location sub-region is smaller; however, the template still meets the minimum accuracy criterion for the intervals of the contained location datasets.

[0052] However, in the method, the termination criterion may also include the following criteria or sub-criteria: the difference between the number of location datasets in the location sub-region and a reference value is less than a tolerance share, wherein the reference value is preferably the average number of location datasets in the location sub-region across all existing location sub-regions. The tolerance value can be, for example, 50%. After the segmentation step, the number of location datasets and their average values ​​in the existing location sub-regions can be generated. If the difference between the numbers is less than the tolerance value, no further segmentation is performed. Therefore, the location sub-regions are formed such that, after the method is completed, the difference between the number of location datasets in the location sub-region and the reference value is less than a tolerance share.

[0053] However, in the method, the termination criterion may also include a sub-criterion related to the extension size of the position sub-region in coordinate space. Therefore, in a preferred variation of the method, the termination criterion may include a criterion or sub-criterion that the extension size of the position sub-region along at least one direction in coordinate space is less than a preset maximum extension size for that at least one direction. This design avoids position sub-regions that extend over a large area of ​​the position dataset and lead to relatively inaccurate templates due to large offsets in the manufacturing elements.

[0054] However, in this method, the termination criterion may also include the following criteria or sub-criterion: the number of location sub-regions is less than or equal to a preset number of location sub-regions. This possibility is preferred when, in order to speed up and / or simplify the testing, only a small number of templates and location sub-regions respectively assigned to the templates should be generated for a large number of training images.

[0055] In the adjustment method, a template is formed for each location sub-region. Known methods for determining the template can be used for this purpose, preferably selected based on the verification method used for the template. In the simplest case, to determine the template for the location sub-region, only the following training images are used to determine the location datasets located within the location sub-regions. However, in other cases, it may be preferable in the method to use, when determining the template for the location sub-regions, training images of training valuable files whose location datasets are located within the respective location sub-regions, and training images of training valuable files whose location datasets are located within a predetermined distance of the boundary of the respective location sub-region, but outside the location sub-region. This approach particularly produces better templates in cases where the boundary between two location sub-regions extends through a cluster of location datasets of training valuable files. However, the extended size and position of the location sub-regions remain unchanged. Attached Figure Description

[0056] The invention is further illustrated below with reference to the accompanying drawings. In the drawings:

[0057] Figure 1 A schematic diagram of a valuable document processing device is shown, using a banknote sorting device as an example.

[0058] Figure 2 This is a rough schematic diagram of a device used to generate a template.

[0059] Figure 3A Figure B shows a rough schematic diagram of the valuable document, illustrating the different locations of the manufactured components.

[0060] Figure 3C It shows from Figure 3B A schematic diagram of digital images collected from valuable files.

[0061] Figure 4 A rough schematic flowchart illustrating an example of an adjustment method for generating templates is shown.

[0062] Figure 5 A schematic frequency diagram showing the positions of manufacturing elements in a preset training image is provided.

[0063] Figure 6 It shows having Figure 4 The flowchart of the sub-step S12 in the process,

[0064] Figure 7A -D indicates according to Figure 6 A view of the location dataset in coordinate space during different segmentation periods.

[0065] Figure 8 A rough schematic flowchart is shown as another example of an adjustment method for generating templates, and

[0066] Figure 9 A rough schematic flowchart of an example of a testing method is shown.

[0067] Figure 1 The valuable document processing device 10 is, for example, a device for processing valuable documents 12 in the form of banknotes of a preset valuable document type. It is designed to classify the valuable documents 12 according to the state determined by the valuable document processing device 10 and the authenticity verified by the valuable document processing device 10.

[0068] The valuable document processing apparatus has an input device 14 for inputting valuable documents 12, an output device 16 for outputting or receiving processed, i.e., classified valuable documents, and a transport device 18 for transporting individual valuable documents from the input device 14 to the output device 16.

[0069] In this example, the input device 14 has an input rack 20 for stacking valuable documents and a separate feeder 22 for separating the valuable documents 12 from the stack of valuable documents in the input rack 20 and inputting the separated valuable documents into the transport device 18.

[0070] In this example, the output device 16 includes three output sections 24, 25, and 26. Based on the intermediate processing results, or in this example, the acquisition results, the processed valuable documents can be categorized into these three output sections. In this example, each section has a stacking rack and stacking wheels (not shown), which allow the input valuable documents to be placed into the stacking rack. In other embodiments, the output sections can be replaced by a device for destroying banknotes.

[0071] The transport device 18 has at least two, in this example three, branches 28, 29 and 30, with output sections 24, 25 or 26 arranged at the end of each branch, and switches 32 and 34 on the branches that can be controlled by adjustment signals. With the help of switches 32 and 34, valuable documents can be input into branches 28 to 30 and from there into output sections 24 to 26 according to the adjustment signals.

[0072] Sensor device 38 is arranged on transport path 36 defined by transport device 18, which is between input device 14, more precisely, individual feeder 22 in this example, and first switch 32 located after individual feeder 22 along the transport direction T. Sensor device 38 acquires physical characteristics of the valuable document as it is transported and generates sensor signals reflecting the acquisition results, which represent sensor data. In this example, sensor device 38 includes: image acquisition device 40 with an optical diffuse reflection sensor capable of acquiring diffuse color images of the valuable document; and other sensors 42 for other physical characteristics of the valuable document, indicated only by box symbols.

[0073] The control and analysis device 46 is connected via signal connections to the sensor device 38 and the transport device 18, particularly the switches 32 and 34. This control and analysis device, in conjunction with the sensor device 38, classifies valuable documents into one of preset classification categories based on signals or sensor data from the sensor device 38. These classification levels can be preset, for example, based on state values ​​determined by sensor data and true values ​​also determined by sensor data. For example, "suitable for circulation" or "unsuitable for circulation" can be used as state values, and "forged," "suspected forged," or "authentic" as true values. Based on the determined classification level, the control and analysis device controls the transport device 18, more specifically, the switches 32 or 34, by issuing control signals, such that the valuable documents are classified into the output section of the output device 16 corresponding to the classification level determined during classification. The assignment to one of the preset classification levels or grades is implemented according to preset criteria used to evaluate state and authenticity, which depend on at least a portion of the sensor data.

[0074] The control and analysis device 46, in particular, includes, in addition to at least one corresponding interface 44 for the sensor device 38 or its sensors, especially the image acquisition device 40, a processor 50 and a memory 48 connected to the processor, which stores at least one computer program with program code. When the computer program is executed, the processor 50 controls the device and analyzes the sensor signals of the sensor device 38, in particular to determine the classification level of the processed valuable documents. Furthermore, the control and analysis device also stores program code, which, when executed, controls the device and, corresponding to the analysis, the transport device 18.

[0075] Interface 44, processor 50, and memory 48, or a segment of memory 48 storing corresponding computer programs and method parameters, are part of a computer and constitute the analysis apparatus 47 in the sense of this disclosure. In this example, the analysis apparatus 47 analyzes the signal from diffuse reflection sensor 40 separately from signals from other sensors. Furthermore, other segments of processor 50 and memory 48 may perform other functions, such as controlling the valuable document processing apparatus 10.

[0076] The diffuse reflection sensor 40 is designed to acquire RGB diffuse reflection images of valuable documents as they are transported by the transport device 18 through the diffuse reflection sensor 40, thereby generating a digital image, which is then analyzed by the analysis device 47.

[0077] The control and analysis device 46, more precisely the analysis device 47, uses sensor data from different sensors in a partial analysis to determine whether the determined characteristics of the valuable document reflect its condition or authenticity, based on the characteristics of the valuable document. The corresponding data can then be stored in the control and analysis device 46, for example, in a memory 48, for later use. Based on the partial analysis, the control and analysis device 46 then determines a classification level as the overall result of the inspection according to a preset overall standard, and generates a classification or control signal for the transport device 18 based on the determined classification level.

[0078] To process the valuable document 12, the valuable document 12, which is either stacked or placed individually in the input rack 20, is separated by the individual feeder 22 and individually fed into the transport device 18, which transports the separated valuable document 12 through the sensor device 38. The sensor device collects the characteristics of the valuable document 12, thereby forming sensor signals reflecting the characteristics of the corresponding valuable document. The control and analysis device 46 collects the sensor signals or sensor data, determines the corresponding classification level of the valuable document based on these sensor signals or sensor data, i.e., in this example, a combination of authenticity level and status level, and controls the switch according to the result, so that the valuable document is transported to the output section corresponding to the determined classification level.

[0079] The analysis device 47 and the image acquisition device 40 together constitute an example of an inspection device for inspecting valuable documents of a preset valuable document type, wherein the valuable document has two preset manufacturing elements, particularly a printing layer and / or anti-counterfeiting elements. Correspondingly, the computer program contains instructions for performing a method for inspecting valuable documents of a preset valuable document type, particularly banknotes, wherein the banknotes have two preset manufacturing elements, particularly a printing layer and / or anti-counterfeiting elements, the inspection being performed using a template, particularly a template generated using the adjustment method described below, which is preset for the manufacturing elements or the pre-defined positions of the manufacturing elements. In the inspection method, a digital valuable document image of the valuable document to be inspected is acquired by means of a diffuse reflection sensor 40 and provided in a corresponding segment of the memory 48 of the analysis device 47. For the provided valuable document image, the relative positions of the manufacturing elements to each other can be determined. Then, based on the positions of the manufacturing elements determined by using position sub-region data, a template for a digital valuable document image or a template for inspecting a digital valuable document image is determined in real time. The digital valuable document image is then inspected using the determined template.

[0080] Figure 2The adjustment device, roughly schematically shown, namely the device 70 for generating a valuable document for verifying a preset valuable document type, is used to provide a template, wherein the valuable document of the preset valuable document type has at least two preset manufacturing elements, particularly printing layers and / or anti-counterfeiting elements, that overlap if necessary. The device 70 is a data processing device having a storage device 72 for storing training images of the preset valuable document type and preferably for storing the generated template. The adjustment device 70 is designed to perform the adjustment method described below using the training images, and to store the generated template and associated positional sub-region data in the storage device 72. For this purpose, the device may include at least one processor 74 connected to the storage device 72 via a data link, and a program memory 76 connected to the processor 74 via a data link, wherein program code is stored, and when the program code is executed, the device, with the aid of the processor 74, performs the adjustment method described below using the training images stored in the storage device 72. In other embodiments, the program memory 76 may also be composed of segments of the storage device 72. The adjustment device 70 may also have a data interface, such as a network card or a local area network card, which can transmit the generated component template stored in the storage device 72 to another device.

[0081] Figure 3A An example of a pre-defined valuable document 12 is roughly illustrated, which has two pre-defined manufacturing elements 62 and 64 in the form of printed layers.

[0082] Figure 3B It shows the relationship with Figure 3A Another valuable file of the same valuable file type, wherein the relative positions of manufacturing elements 62' and 64' corresponding to manufacturing elements 62 and 64 are... Figure 3A The valuable documents in the document are different. Figure 3A The position of manufacturing component 62 in Figure 3B The difference in relative position is represented by a dashed line. Figure 3A and Figure 3B The vector V or V' represents the direction of the manufacturing element 62 or 62', pointing from a predetermined feature element 62A or 62A' of the manufacturing element 62 or 62' to another predetermined feature segment or element 64A or 64A' of the manufacturing element 64 or 64'. In, for example, a two-dimensional Cartesian coordinate system with x and y axes, the vector is represented by two corresponding components.

[0083] exist Figure 3C The text again schematically shows... Figure 3BThe digital image 60 of the valuable document is shown. This digital image contains pixels 66, which in this example are arranged on a square grid and correspond to their positions in the digital image, and therefore also to their positions on the mapped valuable document. The image is preprocessed such that it represents the valuable document 12 only in planar form, i.e., the edges of the valuable document in the image extend along the corresponding edge pixels. In this example, preprocessing is performed for all images so that the image contains the corresponding view. The vector V or V' can then be represented by the corresponding pixel coordinates, which are preferably non-negative integers. The x-axis and y-axis extend parallel to the long and short sides of the valuable document in the image.

[0084] The adjustment method uses training images of digital data from training valuable files of a preset valuable file type. These training images are provided for the adjustment method in a first step. In this example, for this purpose, a finished, clean, preferably freshly printed valuable file of the preset valuable file type is used, the valuable file preferably exhibiting variations in the position of the manufacturing elements. The valuable files preferably also include valuable files with significant differences in the position of the manufacturing elements. The valuable files are preferably selected such that they contain corresponding valuable files with different relative positions of manufacturing elements 62 and 64, wherein, particularly preferably, the frequency of valuable files with a given relative position at least approximately corresponds to the frequency of actual valuable files of the preset valuable file type.

[0085] To generate the template, the following exemplary adjustment method is used, which adjusts... Figure 4 The diagram is roughly schematic.

[0086] For example, training images can be acquired by the processing device 10, particularly the diffuse reflection sensor 40, and stored as digital images in the input analysis device 47. These digital images can be transmitted to the adjustment device 70 via a storage medium or a data connection (not shown), where they are stored in the storage device 72 and thus provided. The digital images each have the same number of pixels and pixel arrangement structure, and display the entire valuable file.

[0087] In step S10, a location dataset with position coordinates in coordinate space is determined for each training image. The location dataset, or the position coordinates therein, describes the relative positions of the manufacturing elements to each other on the valuable file. In this example, two manufacturing elements are used, so the coordinate space is two-dimensional.

[0088] These positions are determined with reference to the same positional reference system, which is given by the edges of the valuable document in the image, or by the edges of the image or the corresponding axis, since the image only shows the entire valuable document.

[0089] To determine the location, anchor regions 62A and 64A are used in this example. These anchor regions are pre-determined for valuable files of a preset valuable file type and characterize manufacturing elements, particularly those that are always visible. In this embodiment, only one anchor region is used for simplicity; in other embodiments, it is preferable to use at least two or more anchor regions per manufacturing element. Anchor regions can be searched in digital images using methods known per se. In this example, the average position of the pattern can be used as the position of the anchor region. For each training valuable file or training image, the relative position is stored associated with the training image in the form of the previously mentioned vector, using corresponding vector components, which are also considered as position coordinates. Each vector in the example is directed from anchor region 62A to anchor region 64A. The result of this step is a set of training images and vectors or vector components Δx associated with the training images in mutually orthogonal directions or dimensions in coordinate space. AB and △y AB Therefore, each training image is assigned a vector.

[0090] Figure 5 An exemplary distribution of a vector or vector components is shown. This is due to the coordinates and therefore the vector components Δx. AB and △y AB Since nB is an integer, the frequency or number of relative positions or vectors determined in the training images can be represented by bars in a two-dimensional histogram, with the height of the bars corresponding to the number of training files. Relative positions that are close to the predefined relative positions usually occur very frequently, while relative positions that deviate significantly from the predefined relative positions are relatively rare.

[0091] In step S12, at least two location sub-regions in the coordinate space are formed. Each location sub-region includes at least a preset number of location datasets. The location sub-regions, i.e., different location sub-regions, do not contain a common location dataset.

[0092] The location sub-regions can, in principle, have any preset shape; however, it is preferable that all location sub-regions have the same shape, such as a rectangle. However, location sub-regions can differ at least in location, and if necessary, in size, such as side length. The location, shape, and size or dimensions of the location sub-regions are described by suitable location sub-region data, which, in the described embodiment, is preferably given by the coordinates of suitable points in coordinate space describing the location and at least the size. Taking a rectangle as an example, as long as a rectangle is selected as the shape in the method, the coordinates of the diagonally opposite corner points in the corresponding rectangular location sub-region can be used.

[0093] like Figure 6Roughly illustrated, step S12 involves performing sub-steps, and repeating them as necessary, to form location sub-regions until a preset termination criterion is met. In this step, if possible, the current location sub-region is divided into N newly formed location sub-regions, each of which contains at least a preset number of location datasets. In this example, N=2.

[0094] In the first implementation, the following region is used as the location sub-region of the first segmentation step, which contains the location coordinates of all training valuable files, i.e., the smallest rectangular region in which all location datasets are located.

[0095] In the embodiment described above, the corresponding location sub-region is segmented only in one dimension, preferably one dimension or direction of the coordinate space.

[0096] In the example, the successive division of the positional sub-regions and the resulting successive division of the positional sub-regions are performed in different spatial dimensions or spatial directions of the coordinate space.

[0097] More precisely, in the adjustment method of the example described, the division is performed along one of two mutually orthogonal division directions. In this example, these are the coordinate axes in the coordinate space.

[0098] If a location sub-region is segmented via a first segmentation direction, then the segmentation of that location sub-region is performed along a second segmentation direction orthogonal to the first segmentation direction. This is implemented such that for each location sub-region generated during segmentation, information is stored regarding which segmentation direction was used during the segmentation, or which direction the next segmentation will follow. During the first segmentation, one of the segmentation directions is given and used. The appropriate segmentation direction to use is then determined for the next segmentation.

[0099] More precisely, in step S12.1, a location sub-region that should be processed next is selected from the existing current location sub-regions.

[0100] In the next step S12.2, it is checked whether the number of location datasets in the selected current location sub-region is greater than N times the minimum value. If this is not the case, it is impossible to segment into location sub-regions, each with at least the minimum number of location datasets. The method then continues with step S12.5, in which the termination criteria are checked. This will be described in more detail below.

[0101] If the number of location datasets in the selected current location sub-region is greater than N times the minimum number, then in step S12.3, the current location sub-region is divided into N location sub-regions. If the current location sub-region was generated by segmentation along a first segmentation direction, then in step S12.3, a second segmentation direction orthogonal to the first segmentation direction is used. The segmentation is performed such that in the resulting location sub-regions, there exists at least a minimum number of location datasets.

[0102] There are typically several possibilities for this. In the example described, the segmentation is performed such that the number of location datasets contained within a rectangular location sub-region is determined. If this number is less than three times a preset number, the location sub-region is divided into two newly formed location sub-regions along a preset direction, the newly formed sub-regions containing approximately the same number of location datasets. Otherwise, the segmentation along the preset direction creates a location sub-region containing a preset number of location datasets and another location sub-region containing at least twice the preset number of location datasets.

[0103] In step S12.4, the current location sub-region, or the location sub-region data defining the current location sub-region, is replaced by the newly formed location sub-region, or the location sub-region data defining the newly formed location sub-region. For the newly formed location sub-region, location sub-region data belonging to the newly formed location sub-region is stored, and these location sub-region data respectively give the location and shape of the location sub-region. Then, step S12.5 is executed, in which the termination criteria mentioned above are checked.

[0104] The segmentation steps S12.1 to S12.4 are performed frequently until all formed location sub-regions meet at least one preset termination criterion, which is checked in the previously mentioned step S12.5. If the termination criterion is met, no further segmentation is performed, and the method continues with step S14. Otherwise, a re-segmentation is attempted, for which step S12.1 is re-executed based on the existing location sub-regions.

[0105] Termination criteria can be a single criterion or can include multiple sub-criteria, which must be met cumulatively or alternatively in order to be considered as having met the termination criteria.

[0106] In the current embodiment, the criterion is only to check whether at least one location sub-region existing at the current time contains less than N times the minimum number of location datasets or N times the minimum number of location datasets. If this is the case, the termination criterion is met, step S12 terminates, and the method continues with step S14. Otherwise, the method continues with step S12.1.

[0107] Figures 7A to 7DSome segmentation steps are roughly and simplified for specific examples. Figure 7A The initial state before the first division is shown: in coordinate space, that is, with coordinate axes Δx respectively. AB and △y AB In the Cartesian coordinate system, the relative positions of all manufacturing elements of the training images used are represented by a cross. The cross is located within a rectangular sub-region T0, which is selected as the first sub-region in step S12.1. The position and size of the sub-region are described by two diagonally opposite corner points A0 and B0 (top left and bottom right corners) or their coordinates as sub-region data. For simplicity, the minimum size of the position dataset is chosen to be 5.

[0108] After checking the number of location datasets in region T0 in step S12.2, Δx is selected in step S12.3. AB The direction serves as the dividing direction for the selected location sub-region. Figure 7B In the diagram, the corresponding dividing lines are represented by dashed lines. The division produces location sub-regions T1 and T2, each containing at least a minimum number of location datasets.

[0109] In the step corresponding to S12.4, the location sub-region T0 is replaced by the newly formed location sub-regions T1 and T2. Furthermore, the location sub-region data A0 and B0 are replaced by the corresponding location sub-region data A1 and B1 or A2 and B2.

[0110] After verifying the termination criterion in the step corresponding to step S12.5, in other loops, the location sub-region T1 is selected as the next location sub-region to be segmented in the step corresponding to step S12.1 (see also...). Figure 7C ).

[0111] like Figure 7C As shown, now along the direction orthogonal to the dividing direction in the previous step S12.3, that is, the second axis of the coordinate space or coordinate system, i.e., Δy AB The axis is used for segmentation. This segmentation is performed such that the two newly formed location sub-regions T3 and T4 contain at least a minimum number of location datasets. In the example, region T3 is chosen such that one of the resulting location sub-regions contains the minimum number of location datasets possible, yet this number is greater than or equal to the minimum number. The dividing line is still represented by a dashed line.

[0112] In the step corresponding to S12.4, the location sub-region T1 is replaced by the newly formed location sub-regions T3 and T4. Furthermore, the location sub-region data A1 and B1 are replaced by the corresponding location sub-region data A3 and B3 or A4 and B4.

[0113] After verifying the termination criterion in the step corresponding to step S12.5, in other loops, in the step corresponding to step S12.1, a position sub-region is selected from the currently existing position sub-regions T2, T3, and T4, for example, position sub-region T4, as the next position sub-region to be segmented (see also...). Figure 7D The number of location datasets in location sub-region T3 is less than twice the preset minimum number, therefore further segmentation is not possible.

[0114] like Figure 7D As shown, now along the direction orthogonal to the dividing direction in the previous step S12.3, that is, the first axis of the coordinate space or coordinate system, i.e., Δx AB The axis is used for segmentation. This segmentation is performed such that the two newly formed location sub-regions T5 and T6 contain at least a minimum number of location datasets. In the example, region T5 is chosen such that one of the resulting location sub-regions contains the fewest possible number of location datasets, yet this number is greater than or equal to the minimum number. The dividing line is still represented by a dashed line.

[0115] In the step corresponding to S12.4, the location sub-region T4 is replaced by newly formed location sub-regions T5 and T6. Furthermore, location sub-region data A4 and B5 are replaced by corresponding location sub-region data A5 and B5 or A6 and B6.

[0116] After verifying the termination criterion in the step corresponding to step S12.5, in other loops, in the step corresponding to step S12.1, T2 is selected from the currently existing unique location sub-region as the next location sub-region to be segmented (see also...). Figure 7C Then, the segmentation is performed in a similar manner to the segmentation described earlier, until the termination criterion is met.

[0117] After step S12 is completed and the segmentation is thus finished, in step S14, templates are determined for each of the formed positional sub-regions using training images whose positional coordinates are located within the corresponding positional sub-regions from the valuable files. Furthermore, for each formed positional sub-region, the determined templates and positional sub-region data belonging to the templates are stored, the positional sub-region data describing the position and extent of the corresponding positional sub-region.

[0118] The pixel data of the template pixels is determined based on the corresponding pixel data of those training images whose position coordinates lie within the corresponding position sub-regions. For contamination detection, for example, lower and upper limits for allowed pixel data values ​​can be determined as pixel data, which are respectively derived from the minimum and maximum values ​​of the corresponding pixel data of the corresponding pixels in the training images.

[0119] The location sub-region data preferably provides the location and extended dimensions of the sub-region given its shape, or provides the location, extended dimensions, and shape of the corresponding sub-region. In the current example, a rectangle is preset as the shape, and the diagonally opposite corners of the rectangle in the used coordinate space are used as the location sub-region data. This location sub-region data reflects the location and extended dimensions of the corresponding sub-region.

[0120] Figure 8 Other embodiments of the adjustment method are roughly illustrated, differing from the previous embodiments only in that step S14 is replaced by step S14'.

[0121] The difference between step S14' and step S14 is that, in order to determine the template for the location sub-region, not only are the training images of those training valuable files whose location datasets are located within the corresponding location sub-regions used, but also the training images of those training valuable files whose location datasets are located within a preset maximum distance from the boundary of the corresponding location sub-region but outside the location sub-region. The actual determination is performed similarly to that in the first example.

[0122] In the example, the maximum distance is 1 pixel.

[0123] Figure 9 A rough illustration shows an example of an inspection method for inspecting valuable documents of a preset valuable document type, wherein a template is used, the template being preset for a predetermined location of a manufactured component. In particular, template and associated location sub-region data determined by one of the examples of the adjustment method can be used. The inspection method can be performed by means of a valuable document processing device 10. The template and associated location sub-region data can, for example, be stored in an analysis device 47.

[0124] In the inspection method, in step S20, during the transport of the valuable document, a digital image of the valuable document to be inspected is acquired using a sensor device 38, particularly an image acquisition device 40, as the document passes through or is transported via the sensor device 38. The digital image of the valuable document comprises pixels, each assigned pixel data. The resolution of the valuable document image corresponds to the resolution of the training image. This means that the valuable document image and the training image have substantially the same number of pixels and arrangement structure. The view of this valuable document image is similar to... Figure 3C The view corresponds to the training valuable files in the template. This also applies to the template accordingly.

[0125] In step S22, the position of the manufacturing element is determined for the acquired valuable file image, wherein the same adjustment method as in the first example is used in this example. More specifically, the position coordinates are determined in the same coordinate space as used in the adjustment method.

[0126] In step S24, a template for the digital valuable document image is determined based on the determined position coordinates, thereby determining a template for further verification. This verification checks which stored position sub-region the determined position coordinates for the current valuable document fall within, using position sub-region data respectively assigned to the template. In this example, the determined position coordinates are more precisely determined to be within which position sub-region, i.e., the rectangle, defined by the position sub-region data. The template corresponding to the position sub-region data and therefore to the position sub-region is used as the template for the subsequent step S26.

[0127] In step S26, the determined template is then used to inspect the digital valuable document image using a preset image inspection method. In the example of damage inspection, the image inspection method, in its simplest example, checks whether the pixel data is between the minimum and maximum values ​​preset by the template used for the pixel. If the pixel data is not between the minimum and maximum values ​​preset by the template used for the pixel, and the number of pixels exceeds a predetermined number, it is identified as damaged; otherwise, it is identified as in a sufficiently good state.

[0128] Based on the test results, a corresponding signal is generated and output in step S28, which reflects the test results. Based on this signal, the classification signal described at the beginning can then be formed and output.

[0129] Other embodiments of the adjustment method differ from the previous two embodiments in that the termination criterion does not include a criterion for the number of location datasets in the reference location sub-region, but rather includes a criterion that the difference between the number of location datasets in the location sub-region and the reference value is less than a tolerance share. As the reference value, in this example, the arithmetic mean of the number of location datasets in the currently existing location sub-regions is used. As the tolerance share, for example, a value of 20% can be chosen. This enables a more uniform segmentation. However, the segmentation step must be adapted for this if necessary.

[0130] Another embodiment differs from the previous two in that the termination criterion does not include the number of location datasets in the reference location sub-region, but rather includes the criterion that the stretch size of the location sub-region in at least one direction in the coordinate space is different from the preset maximum stretch size for at least one direction. In the example, the stretch size can be preset along two coordinate directions. This avoids excessively large location sub-regions.

[0131] Another implementation scheme differs from the previous two in that the termination criterion includes the following: the number of location sub-regions is less than or equal to a preset number of location sub-regions. This number can be selected based on the number of available training images and the speed required during testing. Preferably, the criteria used in the previous paragraphs can also be used. Termination only occurs when both criteria are met simultaneously.

[0132] Other embodiments of the adjustment method differ from the previously described embodiments in that N>2 is chosen, for example, N=3. In the segmentation step, three new location sub-regions are formed, wherein these new location sub-regions must contain at least a minimum number of location datasets.

[0133] Other embodiments may differ from the previously described embodiments in that the template is determined from the associated training images in a different manner. For example, the average value of the corresponding pixel data in the training images can be determined as pixel data for the template pixels.

Claims

1. A method for generating a template, the template being used to verify valuable documents of a preset valuable document type, wherein, A valuable document of a preset valuable document type has at least two preset manufacturing elements, which are a printed layer and / or an anti-counterfeiting element. In the method, training images of digital data from training valuable files of a preset valuable file type are used. Each training image has pixels, and pixel data is assigned to each pixel. The method includes the following steps: - For each of the training images, a location dataset is determined, each having position coordinates in coordinate space, wherein the location datasets describe the positions of the manufacturing elements on the valuable document, at least relative to each other, and - Forming at least two location sub-regions in the coordinate space, each location sub-region containing at least a predetermined number of location datasets, wherein different location sub-regions do not contain common location datasets. - Templates are determined for each location subregion by using training images of valuable files whose location coordinates are located within the location subregion, and the templates and location subregion data are stored, the location subregion data describing the position and extension size of the corresponding location subregion.

2. The method according to claim 1, wherein, The valuable document in question is banknotes.

3. The method according to claim 1, wherein, The pre-designed manufacturing elements partially overlap each other.

4. The method according to claim 1, wherein, Form at least four positional sub-regions in the coordinate space.

5. The method according to claim 1, wherein, The formation of the location sub-region includes multiple segmentation steps. In each segmentation step, the current location sub-region, which at the beginning of the segmentation step contains more than twice the preset number of location datasets, is segmented into a preset number of newly formed location sub-regions. Each of the newly formed location sub-regions contains at least the preset number of location datasets, and the current location sub-regions in the existing location sub-regions are replaced by the newly formed location sub-regions. In the first segmentation step, the region containing the location coordinates of all training valuable files is used as the location sub-region, and This process involves frequently performing segmentation steps until all resulting location sub-regions meet at least one preset termination criterion.

6. The method according to claim 5, wherein, The coordinate space is n-dimensional, where n > 1, and the segmentation is performed in only one dimension in each segmentation step. The successive segmentation of the positional sub-regions and the resulting successive segmentation of the positional sub-regions are performed in different spatial dimensions.

7. The method according to claim 5 or 6, wherein, The termination criteria include the following: the location sub-region contains at most twice the preset number of location datasets.

8. The method according to claim 5, wherein, The termination criteria include the following: the difference between the number of location datasets in the location sub-region and the reference value is less than the tolerance share.

9. The method according to claim 5, wherein, The termination criteria include the following criteria: the extension dimension of the location sub-region along at least one direction in the coordinate space is less than a preset maximum extension dimension for the at least one direction.

10. The method according to claim 5, wherein, The termination criteria include the following criteria: the number of location sub-regions is less than or equal to the preset number of location sub-regions.

11. The method according to any one of claims 1 to 6, wherein, When determining the template for the location sub-region, training images of training valuable files whose location datasets are located in the corresponding location sub-regions are used, as well as training images of training valuable files whose location datasets are located within a preset distance from the boundary of the corresponding location sub-region but outside the location sub-region.

12. An apparatus for generating a template, the template being used to verify a valuable document of a preset valuable document type, wherein, A valuable document of a preset valuable document type has at least two preset manufacturing elements, wherein the manufacturing elements are a printed layer and / or an anti-counterfeiting element, wherein training images of digital data of a training valuable document of the preset valuable document type are used, wherein each training image has pixels and pixel data is respectively assigned to the pixels, the device has a storage device for storing the training images of digital data of a valuable document of the preset valuable document type, wherein the device is designed to perform the method according to any one of claims 1 to 11 by using the training images.

13. The apparatus according to claim 12, wherein, The device has an interface through which the generated template and location sub-region data can be transmitted to another device.

14. The apparatus according to claim 13, wherein, The device is designed to transmit the generated template and location sub-region data to the other device, and / or The device is designed to store the generated template and location sub-region data in a storage device.

15. The apparatus according to claim 12, wherein, The valuable document in question is banknotes.

16. The apparatus according to claim 12, wherein, The pre-designed manufacturing elements partially overlap each other.

17. A computer program product having program code so that, when the program is executed in a computer, it performs the steps of the method according to any one of claims 1 to 11.

18. A computer-readable storage medium having program code that can be executed by a computer, thereby causing the computer to perform the method according to any one of claims 1 to 11.

19. A method for verifying a valuable document of a preset valuable document type, the valuable document having at least two preset manufacturing elements, wherein, The manufacturing element is a printed layer and / or an anti-counterfeiting element. The method is implemented using a template generated according to any one of claims 1 to 11, wherein the template is preset for a preset position sub-region of the manufacturing element. The method comprises the following steps: - Provides a digital image of a valuable file of a preset valuable file type to be verified, wherein the valuable file image includes pixels, and pixel data is assigned to each pixel. - Determine the at least relative positions of the manufactured components in the provided valuable documents. - Based on the determined position of the manufacturing element and the preset position sub-region, a template for a valuable digital file image is determined, and - Use the determined template to examine valuable document images of numbers.

20. The method according to claim 19, wherein, The pre-designed manufacturing elements partially overlap each other.

21. An apparatus for verifying valuable documents, said valuable documents having at least two preset manufacturing elements, wherein, The manufacturing element is a printed layer and / or an anti-counterfeiting element, and the apparatus performs the inspection using a template generated according to any one of claims 1 to 11 and location sub-region data belonging to the template. The device includes an analysis device, which has at least one memory storing the template and location sub-region data respectively associated with the template. and an interface for providing digital, valuable document images, wherein the analysis device is designed to perform the method according to claim 19.

22. The apparatus according to claim 21, wherein, The valuable document in question is banknotes.

23. The apparatus according to claim 21, wherein, The pre-designed manufacturing elements partially overlap each other.

24. The apparatus of claim 21, further comprising an image acquisition device for acquiring digital images of a valuable document to be examined, the image acquisition device being connected via a signal connection to an interface for providing digital images of the valuable document.

25. A computer program product having program code so that, when the program is executed in a computer, it performs the method according to claim 19.

26. A computer-readable storage medium having program code that, when executed by a computer, performs the method of claim 19.

27. An apparatus for processing valuable documents of a preset valuable document type, wherein each valuable document has at least two preset manufacturing elements, wherein, The manufacturing element is a printed layer and / or an anti-counterfeiting element, and the device includes: Input device for inputting individual or separate valuable documents to be processed. The output device has at least one output section for receiving processed valuable files. A transport device for transporting individual or separately valuable documents from the input device to the output device, and The apparatus for verifying valuable documents according to claim 24, The image acquisition device of the apparatus for examining valuable documents is arranged at the transport path and configured such that digital images of the valuable documents to be examined, which are transported from the image acquisition device, are acquired during transport and provided for use in the apparatus for examining valuable documents.

28. The apparatus according to claim 27, wherein, The device is used to inspect and / or count and / or classify and / or destroy valuable documents of a preset valuable document type.

29. The apparatus according to claim 27, wherein, The valuable document in question is banknotes.

30. The apparatus according to claim 27, wherein, The pre-designed manufacturing elements partially overlap each other.

Images