Methods, data processing system and data processing terminal for authentication of marker, and trained artificial neural network, computer programs and data carrier for use in the method and system

Abstract

In a method, performed in a data processing system, an object marker is authenticated. The method comprises receiving object marker image data of the object marker. From an original marker image database, original marker image data of an original marker are retrieved. Marker comparison parameters based on a comparison of the object marker image data to the retrieved original marker image data are determined. The marker comparison parameters are input to a plurality of input nodes of an artificial neural network, ANN. The ANN has been trained to discriminate between object marker image data of the original marker, and a copy or clone of the original marker. The ANN provides output data, representative of the extent of authenticity of the object marker, at its output node. An authenticity signal based on the output data of the ANN is output.

Description

[0001] P36913PC00 / ME

[0002] Methods, data processing system and data processing terminal for authentication of marker, and trained artificial neural network, computer programs and data carrier for use in the method and system

[0003] FIELD OF THE INVENTION

[0004] The invention relates to the field of authentication, and more specifically to methods, a data processing system, a data processing terminal and a trained artificial neural network for authentication of an object marker associated with an object. The invention also relates to associated computer programs and a data carrier.

[0005] BACKGROUND OF THE INVENTION

[0006] A two-dimensional marker may be applied to a product for brand protection and supply chain security solutions. The marker helps companies fight issues of counterfeit, diversion and grey market trading of their products by authenticating the marker when it is encountered on the market.

[0007] The marker is physical and optically detectable. The marker may be a matrix code, such as a Quick Response, QR, code, a data matrix code, a barcode or any other two-dimensional code, such as the iTrace 2DMI® code, associated with an object.

[0008] Natively, the marker is sensitive to replication or copying, and not immune to sophisticated copy or clone attempts.

[0009] Herein, a copy of an original object marker is to be understood as any reproduction of an original object marker, e.g. a photocopy or a scan of an original object marker used to produce a new object marker resembling the original object marker. A copy or reproduction marker is designed to be indistinguishable from an original object marker that has been applied to an original object. Copying is the most common form of attack as it does not require access to authentic marking data and can be executed by a variety of readily available technological means. Counterfeiters typically make copies by photographing or scanning a mark from an original marker and then duplicating it, either from the image taken or reconstructing it from the scanned data. While copying is common, even high-fidelity copies of original object markers can often be distinguished from the original object marker as micro deviations in the creation of the copy marker are detectable. Copying an original object marker only gets the counterfeiter a single marker, and for some markers, such as iTrace 2DMI® markers, counterfeiters may be unable to create a range of different and / or unique markers based on that single copy. To copy a large number of original object markers, a counterfeiter would need to obtain the same number of original objects.

[0010] Further herein, a clone of an original object marker is to be understood as a marker created on the same machine, from the same batch of material, using the same source information as the original object marker, e.g. a reprint of an original document for a digital source, or additional unauthorized production at a factory (also known as production overrun, or 3rdshift production). In other words, cloning is the generation of a marker which is identical to the original object marker using the same hardware, software and parameters that are used to create the original, authentic marker. For example, counterfeiters are able to create new source data for QR codes by scanning a standard QR code and reproducing a new one from the scanned data. This will likely be hardly distinguishable from the original and can be reapplied many times as a new source. For other markers, cloning is less common than copying because it requires insider help and access to the original machines to be executed successfully. If this attack is successful, however, it can be difficult to detect.

[0011] Hence, dedicated systems developed to authenticate the marker in the field may be fooled by marker copies and clones. This has traditionally been attempted to overcome by securing the production of objects with the original object markers, using dedicated sophisticated software for authentication, and checking geo locations of authentications, where a same authentication from different geo locations indicates the presence of copies and / or clones.

[0012] For example, US 2023 / 0222775 A1 discloses a counterfeit and imaging detection system including a processor, a counterfeit product detection app, and a steganographic imaging model, electronically accessible by the counterfeit product detection app trained using image data and configured to cause the processor to obtain a digital image of a physical product of a product line, the digital image captured by an imaging device and the digital image comprising pixel data, analyze the digital image to detect within the pixel data a batch code uniquely identifying a batch of the physical product of the product line, analyze the pixel data of the digital image to determine that the batch code is counterfeit, and augment a counterfeit list of batch codes to include the batch code, wherein the counterfeit list of batch codes remains electronically accessible to the counterfeit product detection app for one or more further counterfeit detection iterations. It is desirable to overcome the perceived weakness of being able to copy or clone an original object marker.

[0013] SUMMARY OF THE INVENTION

[0014] It would be desirable to provide a method and system for authenticating an object marker associated with an object with high reliability. It would also be desirable to provide a method and system for authenticating an object marker associated with an object, being able to recognize an original marker, a copy of an original marker, and a clone of an original marker, so as to discard objects associated with a copy or a clone of an original marker, and take appropriate measures to fight copying and cloning.

[0015] To better address one or more of these concerns, in a first aspect of the invention a method, performed in a data processing system, of authenticating an object marker associated with an object is provided. The method comprises: receiving object marker image data of the object marker; retrieving, from an original marker image database comprising original marker image data of a plurality of original marker images, original marker image data of an original marker; determining marker comparison parameters based on a comparison of the object marker image data to the retrieved original marker image data; inputting each one of the marker comparison parameters to a respective one of a plurality of input nodes of an artificial neural network, ANN, wherein the ANN further comprises a hidden layer and an output node, and wherein the ANN has been trained to discriminate between object marker image data of the original marker, and object marker image data of a copy or clone of the original marker, wherein the ANN is configured to provide output data, representative of the extent of authenticity of the object marker, at the output node; and outputting an authenticity signal indicating the authenticity of the object marker based on the output data of the ANN.

[0016] In conjunction with the method, unique original object markers have been created, i.e. materialized in association with an original object. This materialization may e.g. consist of printing or engraving, such as laser engraving, the original object marker on or in the object. Then, the original object markers have been imaged, i.e. an image is taken from each unique original object marker after the creation of the original object marker. Each original marker image is made up of original marker image data, and the original marker image data are stored in an original marker image database. Thus, the original marker image database contains original marker image data of a plurality of different, unique original markers. The imaging of the original markers is performed in a secure environment, and the original marker image database is managed in a secure environment, to provide optimum protection against unauthorized marker creation and unauthorized database access.

[0017] Now, according to the method, if it is desired to authenticate an object marker found in the field, object marker image data are generated and supplied to a data processing system. Accordingly, the data processing system receives object marker image data of the object marker.

[0018] Then, the data processing system retrieves original marker image data of an original marker from the original marker image database to allow a comparison of the object marker image data to the original marker image data. The comparison of the object marker image data to the original marker image data involves determining marker comparison parameters based on the comparison of the object marker image data to the original marker image data.

[0019] In the data processing system, each one of the marker comparison parameters is input to respective ones of a plurality of input nodes of an artificial neural network, ANN. The ANN further comprises a hidden layer and an output node. The ANN has been trained to discriminate between object marker image data of the original marker, and object marker image data of a copy or clone of the original marker, wherein the ANN is configured to provide output data representative of the extent of authenticity of the object marker at the output node. For example, the output data may represent a number having a value between a minimum value and a maximum value, wherein the maximum value indicates certain authenticity (i.e., the object marker is an original object marker) and the minimum value indicates certain non-authenticity (i.e., the object marker is a copy or clone), and values in between the minimum value and the maximum value indicate a probability of authenticity or non-authenticity.

[0020] The data processing system is configured to output an authenticity signal indicating the authenticity of the object marker based on the output data of the ANN. The authenticity signal may be processed to an indication which may be produced on a display screen, such as an alphanumeric indication, and / or an audible indication produced by a loudspeaker and / or a tactile indication produced by vibration device. For such purposes, the data processing system may be provided with, or connected to suitable output devices. The indication may be a binary indication, representing “authentic”, meaning that the object marker is deemed to be an original marker, or representing “non-authentic”, meaning that the object marker is deemed to be a copy or clone of an original marker. The binary indication representing “authentic” may be provided, for example, if the value of the output data of the ANN is in a predetermined range adjacent to, and including said maximum value, whereas values of the output data outside said range lead to the binary indication representing “non-authentic”. The binary indication may include two different colors (such as green and red), two different symbols (such as the check symbol and the cross symbol X) or characters or words, two different sounds or audio fragments, and / or two different tactile outputs. In exceptional circumstances, if the value of the output data of the ANN is in a predetermined range not including said maximum value and said minimum value, where a low probability of authenticity and a low probability of non-authenticity exists, an indication “new object marker image required” may be produced, in an attempt to obtain improved object marker image data which can lead to one of said binary results.

[0021] The ANN has shown to provide a near 100% accuracy in discriminating between object marker images taken from original markers, and object marker images taken from high quality copies or clones of original markers, even if they were made later on the same machine at about the same time and on the same substrate material. The ANN can be trained with different kinds of substrates for a specific class of objects, such as paper objects, metal objects, glass objects, etc., and / or for a specific way of applying the marker, such as printing, engraving (mechanical engraving, laser engraving, etc.) or any other means of application.

[0022] The training of the ANN can be performed fast, in particular in a timeframe of minutes or less.

[0023] If it is not known of which specific original marker image the original marker image data are to be compared to the object marker image data, and the output data of the ANN indicate that the object marker is not authentic, then original marker image data of further original marker images are to be retrieved from the original marker image database to be compared to the object marker image data, until either the output data of the ANN indicate that the object marker is authentic (i.e. the object marker image data are sufficiently the same as the original marker image data of one of the original marker images), or no output data of the ANN indicate that the object marker is authentic, after all original marker image data have been compared to the object marker image data. In some situations, a great amount of original marker image data is to be processed, which may be time-consuming. To alleviate this problem, parallel computing or quantum computing may be applied.

[0024] In an embodiment of the method, each original marker image carries unique original marker identifier data, and the step of retrieving the original marker image data comprises: deriving object marker identifier data from the object marker image data; and retrieving, from the original marker image database, the original marker image data associated with the original marker identifier data being identical to the object marker identifier data. Different from the situation described above, the comparison of the object marker image data to the original marker image data can be performed directly and efficiently, since only the original marker image data of one specific original marker image, identified by the original marker identifier, needs to be retrieved from the original marker image database for such comparison. In an embodiment, the object marker identifier data may e.g. be contained in the object marker image data as an alphanumeric code which can be read through optical character recognition, OCR, to indicate the original marker identifier of the original marker to compare to the object marker. In another embodiment, the object marker identifier data may be obtained by analyzing the imaged object marker structure which, through application of an algorithm, leads to an original marker identifier of the original marker to compare to the object marker.

[0025] The data processing system is operated in a secure environment, and comprises software loaded into a processor of the data processing system, the software comprising computer instructions for performing the receiving, retrieving, determining, inputting and outputting step of the method. The original marker image database may be included in the data processing system, or may be separately stored in another system configured to communicate with the data processing system. Object marker image data may be generated in the data processing system, e.g. by a camera, or may be generated in another system configured to communicate with the data processing system, in particular in a data processing terminal.

[0026] In particular, in a second aspect of the invention, a method, performed in a data processing terminal, of authenticating an object marker associated with an object is provided, wherein a data processing system may be made to interact with a data processing terminal for authenticating an object marker associated with an object. This method comprises, in the data processing terminal: providing object marker image data of the object marker; transmitting the object marker image data to a data processing system, wherein the data processing system is configured to: retrieve, from an original marker image database comprising original marker image data of a plurality of original marker images, original marker image data of an original marker; determine marker comparison parameters based on a comparison of the object marker image data to the retrieved original marker image data; input each one of the marker comparison parameters to a respective one of a plurality of input nodes of an artificial neural network, ANN, wherein the ANN further comprises a hidden layer and an output node, and wherein the ANN has been trained to discriminate between object marker image data of the original marker, and object marker image data of a copy or clone of the original marker, wherein the ANN is configured to provide output data representative of the extent of authenticity of the object marker at the output node, output an authenticity signal indicating the authenticity of the object marker based on the output data of the ANN, and receiving, from the data processing system, an authentication result signal based on the authenticity signal generated in the data processing system.

[0027] The data processing terminal may comprise software, in particular an app(lication), configured to be loaded into a processor of the data processing terminal, the software comprising computer instructions for performing the generating, transmitting and receiving step of the method. The data processing terminal may further comprise a camera to take an image of the object marker. However, an image of the object marker may also be made available in the data processing terminal by transfer of data through a port of the data processing terminal, or through receipt of a message comprising object marker image data in the data processing terminal. The data processing terminal may further comprise a telecommunication system configured for performing the transmitting and receiving step of the method, e.g. via a wireless telecommunication network. In some embodiments, the data processing terminal is a laptop computer, tablet computer, smartphone, or any other portable data processing device.

[0028] In an embodiment, the data processing system and / or the original marker image database may be included in the data processing terminal, wherein the transmitting and receiving steps of the method are performed internally in the data processing terminal.

[0029] In an embodiment of the method, the comparison of the object marker image data to the original marker image data comprises the steps of: processing the object marker image data and / or the original marker image data to align the object marker image to the original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the object marker image and the original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the object marker image and the original marker image as a marker comparison parameter.

[0030] In an embodiment, the method further comprises the training of the ANN before the step of receiving object marker image data of the object marker, wherein the training of the ANN comprises the steps of: generating reference original marker image data of a reference original marker image of an original marker; generating additional original marker image data of a plurality of additional original marker images of the original marker, the plurality of additional original marker images being generated under varying circumstances; generating copy marker image data of a plurality of copy marker images of copies of the original marker; generating clone marker image data of a plurality of clone marker images of clones of the original marker;

[0031] (a) determining at least one first training marker comparison parameter based on a comparison of additional original marker image data of an additional original marker image to the reference original marker image data, inputting each one of the first training marker comparison parameters to a respective one of a plurality of input nodes of the ANN, setting the output data at the output node of the ANN to be representative of the authenticity of the additional original marker image, and back-propagating the set output data to the hidden layer of the ANN;

[0032] (b) determining at least one second training marker comparison parameter based on a comparison of copy marker image data of a copy marker image to the reference original marker image data, inputting each one of the second training marker comparison parameters to a respective one of a plurality of input nodes of the ANN, setting the output data at the output node of the ANN to be representative of non-authenticity of the copy marker image, and back-propagating the set output data to the hidden layer of the ANN;

[0033] (c) determining at least one third training marker comparison parameter based on a comparison of clone marker image data of a clone marker image to the reference original marker image data, inputting each one of the third training marker comparison parameters to a respective one of a plurality of input nodes of the ANN, setting the output data at the output node of the ANN to be representative of non-authenticity of the clone marker image, and back-propagating the set output data to the hidden layer of the ANN;

[0034] (d) for completing one training epoch, repeating step (a) for different combinations of additional original marker image data of one of the plurality of additional original marker images and the reference original marker image data, repeating step (b) for different combinations of copy marker image data of one of the plurality of copy marker images and the reference original marker image data, and repeating step (c) for different combinations of clone marker image data of one of the plurality of clone marker images and the reference original marker image data; and

[0035] (e) repeating steps (a) to (d) in multiple epochs until the mean squared error of the output data of the ANN approaches or reaches a minimum.

[0036] In the training of the ANN, the comparison of additional original marker image data of an additional original marker image to the reference original marker image data comprises the steps of: processing the additional original marker image data and / or the reference original marker image data to align the additional original marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the additional original marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the additional original marker image and the reference original marker image as a marker comparison parameter.

[0037] In the training of the ANN, the comparison of copy marker image data of a copy marker image to the reference original marker image data comprises the steps of: processing the copy marker image data and / or the reference original marker image data to align the copy marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the copy marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the copy marker image and the reference original marker image as a marker comparison parameter.

[0038] In the training of the ANN, the comparison of clone marker image data of a clone marker image to the reference original marker image data comprises the steps of: processing the clone marker image data and / or the reference original marker image data to align the clone marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the clone marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the clone marker image and the reference original marker image as a marker comparison parameter.

[0039] Herein, aligning may involve rotating, shifting and / or scaling such that the object marker image overlays the original marker image in an optimum manner.

[0040] In an embodiment of the method, the marker is a two-dimensional marker. The two- dimensional marker may be embodied as a QR code marker or barcode marker, or as an iTRACE 2DMI® marker, or a combination thereof.

[0041] An iTRACE 2DMI® marker comprises a plurality of line segments each extending between a first end point and a second end point thereof, wherein the first end points and the second end points of the line segments are located on an edge line comprising at least one edge line segment. Line segments may be straight lines, or may e.g. comprise a part of a waveform pattern. Width of different line segments in the marker may differ.

[0042] In a third aspect of the invention, a data processing system for authenticating an object marker associated with an object is provided. The data processing system comprises: a receiving component for receiving object marker image data of the object marker; a retrieving component for retrieving, from an original marker image database comprising original marker image data of a plurality of original marker images, original marker image data of an original marker; a comparison component for determining marker comparison parameters based on a comparison of the object marker image data to the original marker image data; an input component for inputting each one of the marker comparison parameters to a respective one of a plurality of input nodes of an artificial neural network, ANN, wherein the ANN further comprises a hidden layer and an output node, and wherein the ANN has been trained to discriminate between object marker image data of the original marker, and object marker image data of a copy or clone of the original marker, wherein the ANN is configured to provide output data representative of the extent of authenticity of the object marker at the output node; and an output component for outputting an authenticity signal indicating the authenticity of the object marker based on the output data of the ANN.

[0043] Herein, the term ‘component’ refers to combined hardware and software functionality, i.e. a hardware processing component (such as a processor) configured to load and / or store software comprising instructions for performing a specific function or method step as described.

[0044] In an embodiment of the data processing system, the original marker image database is comprised in the data processing system. The data processing system may be part of a portable computer, tablet computer, smartphone or other portable computer device having a memory storing the original marker image database.

[0045] In an embodiment, the data processing system further comprises a camera coupled to the receiving component. The data processing system may be part of a portable computer, tablet computer, smartphone or other portable computer device, and having a camera for imaging an object marker, wherein the object marker image data may be transferred to the receiving component of the data processing system.

[0046] In a fourth aspect of the invention, a data processing terminal for authenticating an object marker associated with an object is provided. The data processing terminal comprises: an image input component for obtaining object marker image data of the object marker; a transmission component for transmitting the object marker image data to a data processing system of the invention; and a receiving component for receiving, from the data processing system, an authentication result signal based on the authenticity signal generated in the data processing system.

[0047] The data processing terminal may be a portable computer, a tablet computer, a smartphone, or any other portable computing device remote from a data processing system. The transmitting and receiving between the data processing terminal and the data processing system are performed using a wireless telecommunication network.

[0048] In an embodiment of the data processing terminal, the original marker image database is comprised in the data processing terminal.

[0049] In an embodiment of the data processing terminal, the image input component comprises a camera for imaging an object marker, wherein the object marker image data may be transmitted from the data processing terminal to the data processing system.

[0050] In a fifth aspect of the invention, a trained artificial neural network, ANN, for use in the method of the invention is provided. The ANN comprises a plurality of input nodes, a hidden layer and an output node, wherein the ANN has been trained to discriminate between object marker image data of the original marker, and object marker image data of a copy or clone of the original marker, wherein the ANN is configured to provide output data representative of the extent of authenticity of the object marker at the output node, and wherein the ANN has been trained by the steps of: generating reference original marker image data of a reference original marker image of an original marker; generating additional original marker image data of a plurality of additional original marker images of the original marker, the plurality of additional original marker images being generated under varying circumstances; generating copy marker image data of a plurality of copy marker images of copies of the original marker; generating clone marker image data of a plurality of clone marker images of clones of the original marker;

[0051] (a) determining at least one first training marker comparison parameter based on a comparison of additional original marker image data of an additional original marker image to the reference original marker image data, inputting each one of the first training marker comparison parameters to a respective one of a plurality of input nodes of the ANN, setting the output data at the output node of the ANN to be representative of the authenticity of the additional original marker image, and back-propagating the set output data to the hidden layer of the ANN;

[0052] (b) determining at least one second training marker comparison parameter based on a comparison of copy marker image data of a copy marker image to the reference original marker image data, inputting each one of the second training marker comparison parameters to a respective one of a plurality of input nodes of the ANN, setting the output data at the output node of the ANN to be representative of non-authenticity of the copy marker image, and back-propagating the set output data to the hidden layer of the ANN;

[0053] (c) determining at least one third training marker comparison parameter based on a comparison of clone marker image data of a clone marker image to the reference original marker image data, inputting each one of the third training marker comparison parameters to a respective one of a plurality of input nodes of the ANN, setting the output data at the output node of the ANN to be representative of non-authenticity of the clone marker image, and back-propagating the set output data to the hidden layer of the ANN;

[0054] (d) for completing one training epoch, repeating step (a) for different combinations of additional original marker image data of one of the plurality of additional original marker images and the reference original marker image data, repeating step (b) for different combinations of copy marker image data of one of the plurality of copy marker images and the reference original marker image data, and repeating step (c) for different combinations of clone marker image data of one of the plurality of clone marker images and the reference original marker image data; and

[0055] (e) repeating steps (a) to (d) in multiple epochs until the mean squared error of the output data of the ANN approaches or reaches a minimum.

[0056] In an embodiment of the ANN, the comparison of additional original marker image data of an additional original marker image to the reference original marker image data comprises the steps of: processing the additional original marker image data and / or the reference original marker image data to align the additional original marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the additional original marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the additional original marker image and the reference original marker image as a marker comparison parameter.

[0057] In an embodiment of the ANN, the comparison of copy marker image data of a copy marker image to the reference original marker image data comprises the steps of: processing the copy marker image data and / or the reference original marker image data to align the copy marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the copy marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the copy marker image and the reference original marker image as a marker comparison parameter.

[0058] In an embodiment of the ANN, the comparison of clone marker image data of a clone marker image to the reference original marker image data comprises the steps of: processing the clone marker image data and / or the reference original marker image data to align the clone marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the clone marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the clone marker image and the reference original marker image as a marker comparison parameter.

[0059] In a sixth aspect of the invention, a computer program is provided, the computer program comprising computer instructions which, when processed in a data processing system, cause the data processing system to carry out the method of the invention.

[0060] In a seventh aspect of the invention, a computer program is provided, the computer program comprising computer instructions which, when processed in a data processing terminal, cause the data processing terminal to carry out the method of the invention.

[0061] In an eighth aspect of the invention, a data carrier comprising computer instructions for the computer program of the invention is provided.

[0062] These and other aspects of the invention will be more readily appreciated as the same becomes better understood by reference to the following detailed description and considered in connection with the accompanying drawings in which like reference symbols designate like parts.

[0063] BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 depicts a first example of a marker, in particular an embodiment of an iTRACE 2DMI® marker, of which the authenticity can be checked according to the present invention.

[0065] Figure 2 depicts a second example of a marker, in particular another embodiment of an iTRACE 2DMI® marker, of which the authenticity can be checked according to the present invention.

[0066] Figure 3 depicts a third example of a marker, in particular a QR code marker, of which the authenticity can be checked according to the present invention.

[0067] Figure 4 depicts a fourth example of a marker, in particular a barcode marker, of which the authenticity can be checked according to the present invention.

[0068] Figure 5 diagrammatically illustrates a first method for authentication of object markers according to the invention. Figure 6 diagrammatically illustrates a second method for authentication of object markers according to the invention.

[0069] Figure 7 schematically illustrates the structure of an ANN according to the invention.

[0070] Figure 8 depicts a flow diagram illustrating the training of the ANN.

[0071] Figure 9 illustrates SSIM pixels found in an alignment of an object marker image and an original marker image, wherein the object marker may be a copy marker or clone marker.

[0072] Figure 10 illustrates an XOR comparison of an object marker image and an original marker image, wherein the object marker is an original marker.

[0073] Figure 11 illustrates an XOR comparison of an object marker image and an original marker image, wherein the object marker is a copy marker or clone marker.

[0074] DETAILED DESCRIPTION OF EMBODIMENTS

[0075] Figure 1 depicts an example embodiment of an iTRACE 2DMI® marker 10, as applied to an object 20, e.g. printed on a label, printed on a part of the object surface, engraved on an object, or otherwise applied to an object such that an image of it can be taken. The marker 10 comprises a plurality of line segments each extending between a first end point and a second end point thereof, wherein the first end points and the second end points of the line segments are located on an edge line comprising at least one edge line segment. In the example embodiment as shown in Figure 1, the edge line 30 comprises four edge line segments, i.e. an upper edge line segment 32, a lower edge line segment 34, a left edge line segment 36 and a right edge line segment 38, together forming a rectangle, in particular a square. Eight straight line segments extend between a respective first end point on the upper edge line segment 32 and a respective second end point on the lower edge line segment 34, and another eight straight line segments extend between a respective first end point on the left edge line segment 36 and a respective second end point on the right edge line segment 38. It is noted that an iTRACE 2DMI® marker may also comprise another number (e.g. less or more than eight) of line segments extending between the upper edge line segment 32 and the lower edge line segment 34, and / or another number (e.g. less or more than eight) of line segments extending between the left edge line segment 36 and right edge line segment 38. The marker 10 is associated with a marker identifier, which may e.g. be generated through an algorithm from the locations of the first and second end points of the straight lines on the edge line segments 32, 34, 36 and 38. Imaging the marker 10 and applying the algorithm will provide the marker identifier. After the marker 10 has been applied to the object 20, it is considered to be an original marker, and image data generated from an image of the original marker are stored in an original marker image database, in association with the original marker identifier, where there is a one-to-one relationship between the original marker image data and the original marker identifier by ensuring that each original marker identifier is unique, which is realized by a unique combination of the first and second end points of the straight line segments on the edge line 30.

[0076] When the marker 10 is imaged again at a later time, it is considered an object marker. Said algorithm allows to regenerate the original marker identifier from the object marker image data, whereby the original marker image data associated with the original marker identifier can be retrieved from the original marker image database for a comparison between the object marker image data and the original marker image data.

[0077] Figure 2 depicts another embodiment of an iTRACE 2DMI® marker 50. Like the marker 10, the marker 50 comprises a plurality of line segments each extending between a first end point and a second end point thereof, wherein the first end points and the second end points of the line segments are located on an edge line comprising at least one edge line segment. In the embodiment as shown in Figure 2, the edge line 60 comprises four edge line segments, i.e. an upper edge line segment 62, a lower edge line segment 64, a left edge line segment 66 and a right edge line segment 68, together forming a rectangle. Two line segments 70, 72, wherein line segment 70 has a larger width than line segment 72, extend between a respective first end point on the upper edge line segment 62 and a respective second end point on the lower edge line segment 64, and two line segments 74, 76 extend between a respective first end point on the left edge line segment 66 and a respective second end point on the right edge line segment 68. Line segment 70 is a straight line segment, whereas line segments 72, 74 and 76 each depict a part of a waveform, each with a different frequency.

[0078] The marker 50 is associated with a marker identifier, which may e.g. be generated through an algorithm from the locations of the first and second end points of the straight lines on the edge line segments 62, 64, 66 and 68, the widths of the line segments 70, 72, 74, 76 and the frequencies of the waveforms of the line segments 70, 72, 74, 76. Imaging the marker 50 and applying the algorithm will provide the marker identifier. After the marker 50 has been applied to an object, it is considered to be an original marker, and image data generated from an image of the original marker are stored in an original marker image database, in association with the original marker identifier, where there is a one-to-one relationship between the original marker image data and the original marker identifier by ensuring that each original marker identifier is unique, which is realized by a unique combination of the first and second end points of the line segments on the edge line 60, the widths of the line segments 70, 72, 74, 76 and the frequencies of the waveforms of the line segments 70, 72, 74, 76.

[0079] Figure 3 depicts a marker embodied as a QR code marker 80. The marker 80 comprises three alignment features 82 and a plurality of code blocks 84 which may be arranged to represent a marker identifier. Imaging the marker 80 and applying a predetermined algorithm will provide the marker identifier. After the marker 80 has been applied to an object, it is considered to be an original marker, and image data generated from an image of the original marker are stored in an original marker image database, in association with the original marker identifier, where there is a one-to-one relationship between the original marker image data and the original marker identifier by ensuring that each original marker identifier is unique, which is realized by a unique combination of code blocks 84.

[0080] Figure 4 depicts a marker embodied as a barcode marker 90. The marker 90 comprises a series of combinations of wide and narrow line segments 92. Each combination of wide and narrow line segments 92 represents a numerical value 94, where the series of numerical values 94 may be a marker identifier. Imaging just the marker 90 and applying a predetermined algorithm will provide the marker identifier. After the marker 90 has been applied to an object, it is considered to be an original marker, and image data generated from an image of the original marker are stored in an original marker image database, in association with the original marker identifier, where there is a one-to-one relationship between the original marker image data and the original marker identifier by ensuring that each original marker identifier is unique, which is realized by a unique combination of numerical values 94, and consequently by a unique combination of wide and narrow line segments 92.

[0081] Figure 5 diagrammatically illustrates steps of a first method for authentication of an object marker associated with an object. The object marker e.g. is printed on the object or engraved on or in the object. The method is performed in a data processing system.

[0082] In a first step 510 of the method, a receiving component of the data processing system receives object marker image data of the object marker. The object marker image data is obtained by imaging the object marker. In a second step 520 of the method, a retrieving component of the data processing system retrieves, from an original marker image database comprising original marker image data of a plurality of original marker images, original marker image data of an original marker, for comparison of the object marker to the original marker, to determine whether the object marker is authentic or not. The original marker image database has been prepared before, and contains original marker image data of a collection of original markers. The original marker image database may be part of the data processing system or not.

[0083] In a third step 530 of the method, a comparison component of the data processing system determines marker comparison parameters based on a comparison of the object marker image data to the retrieved original marker image data.

[0084] In a fourth step 540 of the method, an input component of the data processing system inputs each one of the marker comparison parameters to a respective one of a plurality of input nodes of an artificial neural network, ANN, wherein the ANN further comprises a hidden layer and an output node, wherein the ANN has been trained to discriminate between object marker image data of the original marker, and object marker image data of a copy or clone of the original marker, and wherein the ANN is configured to provide output data representative of the extent of authenticity of the object marker at the output node. The output data may e.g. provide a value in a predetermined range between a first value and a second value, wherein, if the value is close the first value, the output data indicate that the object marker has been found authentic by the ANN, whereas, if the value is close to the second value, the output value indicates that the object marker has been found non-authentic by the ANN.

[0085] In a fifth step 550 of the method, an output component of the data processing system outputs an authenticity signal indicating authenticity of the object marker based on the output data of the ANN. The authenticity signal may be presented to a user of the data processing system, and may e.g. be presented visually (e.g. on a display of the data processing system) and / or audibly (e.g. through a loudspeaker of the data processing system) and / or in tactile form, such as by a vibration or a series of vibrations.

[0086] According to the above, the method illustrated in Figure 5 is executable in a data processing system 686 (see Figure 6) for authenticating an object marker associated with an object, the data processing system comprising: a receiving component for receiving object marker image data of the object marker; a retrieving component for retrieving, from an original marker image database comprising original marker image data of a plurality of original marker images, original marker image data of an original marker; a comparison component for determining marker comparison parameters based on a comparison of the object marker image data to the original marker image data; an input component for inputting each one of the marker comparison parameters to a respective one of a plurality of input nodes of an artificial neural network, ANN, wherein the ANN further comprises a hidden layer and an output node, wherein the ANN has been trained to discriminate between object marker image data of the original marker, and object marker image data of a copy or clone of the original marker, and wherein the ANN is configured to provide output data representative of the extent of authenticity of the object marker at the output node; and an output component for outputting an authenticity signal indicating the authenticity of the object marker based on the output data of the ANN.

[0087] It is reiterated that the term ‘component’ refers to combined hardware and software functionality, i.e. a hardware processing component (such as a processor) configured to load and / or store software comprising instructions for performing a specific function or method step as described.

[0088] Figure 6 diagrammatically illustrates steps of a second method for authentication of an object marker associated with an object. The object marker e.g. is printed on the object or engraved on or in the object. The method is performed in a data processing terminal, schematically indicated by a dashed line box 684, which is in communication with a data processing system, schematically indicated by a dashed line box 686.

[0089] In a first step 610 of the method executable in the data processing terminal, an image input component of the data processing terminal 684 obtains object marker image data of the object marker. The object marker image data may be obtained by imaging the object marker with a camera, which may or may not be part of the data processing terminal. In the latter case, object marker image data imaged with an external camera may be obtained by the data processing terminal through a data input port or data communication port thereof.

[0090] In a second step 620 of the method executable in the data processing terminal, a transmission component of the data processing terminal 684 transmits the object marker image data to a data processing system 686 as described and discussed above by reference to Figure 5. The transmission may be through a wireless telecommunication network. Next, the data processing system 686 performs the steps 510, 520, 530, 540 and 550. In step 520, original marker image data are retrieved from original marker image database 682. The database 682 may be part of a separate processing system 688, or may be included in the data processing system 686. After step 550, the data processing system 686 transmits an authentication result signal based on the authenticity signal to the data processing terminal 684. The transmission may be through a wireless telecommunication network.

[0091] In a further step 630 of the method executable in the data processing terminal, a receiving component of the data processing terminal 684 receives, from the data processing system 686, the authentication result signal based on the authenticity signal generated in the data processing system 686. The authentication result signal may be presented to a user of the data processing terminal 684, and may e.g. be presented visually (e.g. on a display of the data processing terminal) and / or audibly (e.g. through a loudspeaker of the data processing terminal) and / or in tactile form, such as by a vibration or a series of vibrations.

[0092] According to the above, the method illustrated in Figure 6 is executable in a data processing terminal 684 for authenticating an object marker associated with an object, the data processing terminal 684 comprising: an image input component for obtaining object marker image data of the object marker; a transmission component for transmitting the object marker image data to a data processing system; and a receiving component for receiving, from the data processing system, an authentication result signal based on the authenticity signal generated in the data processing system.

[0093] As indicated by dashed line box 690, the method may alternatively be performed in one data processing device 690, whereby the transmission steps performed in the method take place internally in the data processing device 690, without using a wireless telecommunication network.

[0094] As indicated by dashed line box 692, the method may alternatively be performed in one data processing device 692, whereby transmission steps performed in the method take place internally in the data processing device 692, without using a wireless telecommunication network. Here, the database 682 is part of the data processing device 692. The data processing terminal 684 or data processing device 690 or data processing device 692 may be a smartphone or other portable device allowing to take an image of an object marker in the field, to allow an authentication of the object marker and the related object.

[0095] Figure 7 schematically represents a structure of an artificial neural network, ANN, 700 used in the methods and systems described above. The ANN 700 comprises an input layer 710, a hidden layer 720 and an output layer 730. The input layer 710 comprises a plurality of input nodes 740. The hidden layer 720 comprises a plurality of hidden nodes 750. The output layer comprises one output node 760.

[0096] For training purposes, in an embodiment the ANN activation function for the hidden layer 720 is a hyperbolic tangent, tanh, activation function. Values input to the hidden nodes 750 are transformed by the tanh activation function to values between -1 and 1 output from the hidden nodes 750. In an embodiment, the ANN 700 has a learning rate of 0.05, used in the back propagation of the training of the ANN 700.

[0097] As illustrated in the flow diagram of Figure 8, a training of the ANN 700 comprises a step 770 of generating reference original marker image data of a reference original marker image of an original marker, a step 772 of generating additional original marker image data of a plurality of additional original marker images of the original marker, the plurality of additional original marker images being generated under varying circumstances, a step 774 of generating copy marker image data of a plurality of copy marker images of copies of the original marker, and a step 776 of generating clone marker image data of a plurality of clone marker images of clones of the original marker.

[0098] The plurality of additional original marker images are generated by imaging an original marker from slightly different viewpoints and / or under slightly varying light circumstances to produce additional original marker image data. The plurality of copy marker images are generated by first providing copies of the original marker made by different techniques, such as scanning and printing, and then imaging the copy markers from slightly different angles and / or under varying light circumstances to produce copy marker image data. The plurality of clone marker images are generated by first cloning the original marker, i.e. reproducing the original marker using the same software, hardware and substrate type, and then imaging the clone markers from slightly different angles and / or under varying light circumstances to produce clone marker image data. Next in the training of the ANN 700, the generated reference original marker image data, additional original marker image data, copy marker image data and clone marker image data are processed, performing the following steps in series or in parallel:

[0099] (a) a step 778 of determining at least one first training marker comparison parameter based on a comparison of additional original marker image data of an additional original marker image to the reference original marker image data, inputting each one of the first training marker comparison parameters to a respective one of a plurality of input nodes 740 of the ANN 700, setting the output data at the output node 760 of the ANN 700 to be representative of the authenticity of the additional original marker image, and back- propagating the set output data to the hidden layer 720 of the ANN 700;

[0100] (b) a step 780 of determining at least one second training marker comparison parameter based on a comparison of copy marker image data of a copy marker image to the reference original marker image data, inputting each one of the second training marker comparison parameters to a respective one of a plurality of input nodes 740 of the ANN 700, setting the output data at the output node 760 of the ANN 700 to be representative of nonauthenticity of the copy marker image, and back-propagating the set output data to the hidden layer 720 of the ANN 700;

[0101] (c) a step 782 of determining at least one third training marker comparison parameter based on a comparison of clone marker image data of a clone marker image to the reference original marker image data, inputting each one of the third training marker comparison parameters to a respective one of a plurality of input nodes 740 of the ANN 700, setting the output data at the output node 760 of the ANN 700 to be representative of non-authenticity of the clone marker image, and back-propagating the set output data to the hidden layer 720 of the ANN 700;

[0102] (d) for completing one training epoch, a step 784 of repeating step (a) for different combinations of additional original marker image data of one of the plurality of additional original marker images and the reference original marker image data, a step 786 of repeating step (b) for different combinations of copy marker image data of one of the plurality of copy marker images and the reference original marker image data, and a step 788 of repeating step (c) for different combinations of clone marker image data of one of the plurality of clone marker images and the reference original marker image data.

[0103] In each one of the steps (a), (b) and (c), in a step 790 an error of the output data of the ANN is determined.

[0104] Further in the training of the ANN 700, in a step 792 a mean squared error, MSE, of the errors determined in respective steps 790 is determined. The steps (a) to (d) are repeated in multiple epochs, as indicated by arrow 794. In an analysis of the multiple epochs, as represented by diamond box 796, the development of the MSE is monitored, e.g. by producing a graph of the MSE against the number of epochs. If it is observed that the MSE of the output data of the ANN 700 approaches or reaches a minimum, the number of training epochs may be considered optimal, and the training may be considered as completed. In Figure 8, the symbol N indicates that a minimum has not yet been reached, whereas the symbol Y indicates optimal training of the ANN 700, avoiding overtraining of the ANN 700.

[0105] Both in the use of the ANN 700 and in the training thereof, specific marker comparison parameters are determined from comparisons of marker image data. In the use of the ANN 700, object marker image data are compared to original marker image data. In the training of the ANN 700, additional original marker image data are compared to reference original marker image data, copy marker image data are compared to reference original marker image data, and clone marker image data are compared to reference original marker image data.

[0106] In particular, in the use of the ANN, the comparison of the object marker image data to the original marker image data comprises the steps of: processing the object marker image data and / or the original marker image data to align the object marker image to the original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the object marker image and the original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the object marker image and the original marker image as a marker comparison parameter.

[0107] Further in particular, in the training of the ANN, the comparison of additional original marker image data of an additional original marker image to the reference original marker image data comprises the steps of: processing the additional original marker image data and / or the reference original marker image data to align the additional original marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the additional original marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the additional original marker image and the reference original marker image as a marker comparison parameter.

[0108] Further in particular, in the training of the ANN, the comparison of copy marker image data of a copy marker image to the reference original marker image data comprises the steps of: processing the copy marker image data and / or the reference original marker image data to align the copy marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the copy marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the copy marker image and the reference original marker image as a marker comparison parameter.

[0109] Further in particular, in the training of the ANN, the comparison of clone marker image data of a clone marker image to the reference original marker image data comprises the steps of: processing the clone marker image data and / or the reference original marker image data to align the clone marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the clone marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the clone marker image and the reference original marker image as a marker comparison parameter.

[0110] In the training of the ANN, each of the first, second and third training marker comparison parameters may comprise at least one of the SSIM score marker comparison parameter and the XOR score marker comparison parameter.

[0111] Figure 9 illustrates a comparison of an iTRACE 2DMI® object marker to an original marker. The marker comprises a plurality of straight line segments each extending between a first end point and a second end point thereof, wherein the first end points and the second end points of the line segments are located on an edge line comprising at least one edge line segment. After processing the object marker image data and / or the original marker image data to align the object marker image to the original marker image, an SSIM comparison is performed to measure the similarity between the object marker image and the original marker image. In the SSIM comparison, (areas of) pixels 900 indicating differences between the object marker image and the original marker image may be identified (dark grey areas). Based on the identified pixels, a marker comparison parameter having a value of an SSIM score is calculated, to be input to an input node 740 of the ANN 700. The SSIM score may e.g. be determined based on (an) area(s) of the pixels 900.

[0112] Figure 10 illustrates an XOR comparison of an iTRACE 2DMI® object marker image and an original marker image, wherein the object marker in fact is an original marker, whereas Figure 11 illustrates an XOR comparison of an iTRACE 2DMI® object marker image and an original marker image, wherein the object marker is a copy marker or clone marker. After processing the object marker image data and / or the original marker image data to align the object marker image to the original marker image, in the XOR comparison, an XOR operation is performed on bits of the object marker image and the bits in the original marker image at the same position in the image. If the bits are different, a value 1 results, and if the bits are the same, a value 0 results. In the Figures 10 and 11, the value 1 is represented in white, whereas the value 0 is represented in black.

[0113] As can be seen between the Figures 10 and 11 , contour lines 910, 920 defined by pixel differences along the line segments of the marker are more pronounced in case of the object marker being a copy marker or clone marker (Figure 11) than in case of the object marker being an original marker (Figure 10). Based on the found pixels, a marker comparison parameter having a value of an XOR score is calculated, to be input to an input node 740 of the ANN 700. The XOR score may e.g. be determined based on (an) area(s) of the pixels along the contour lines 910, 920.

[0114] As described in detail above, in a method, performed in a data processing system, an object marker is authenticated. The method comprises receiving object marker image data of the object marker. From an original marker image database, original marker image data of an original marker are retrieved. Marker comparison parameters based on a comparison of the object marker image data to the retrieved original marker image data are determined. The marker comparison parameters are input to a plurality of input nodes of an artificial neural network, ANN. The ANN has been trained to discriminate between object marker image data of the original marker, and a copy or clone of the original marker. The ANN provides output data, representative of the extent of authenticity of the object marker, at its output node. An authenticity signal based on the output data of the ANN is output.

[0115] As required, detailed embodiments of the present invention are disclosed herein. However, it is to be understood that the disclosed embodiments are merely exemplary of the invention, which can be embodied in various forms. Therefore, specific structural and functional details disclosed herein are not to be interpreted as limiting, but merely as a basis for the claims and as a representative basis for teaching one skilled in the art to variously employ the present invention in virtually any appropriately detailed structure. Further, the terms and phrases used herein are not intended to be limiting, but rather, to provide an understandable description of the invention.

[0116] The terms "a" / "an", as used herein, are defined as one or more than one. The term plurality, as used herein, is defined as two or more than two. The term another, as used herein, is defined as at least a second or more. The terms including and / or having, as used herein, are defined as comprising (i.e., open language, not excluding other elements or steps). Any reference signs in the claims should not be construed as limiting the scope of the claims or the invention.

[0117] The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

[0118] The term coupled, as used herein, is defined as connected, although not necessarily directly.

[0119] A single processor or other unit may fulfil the functions of several items recited in the claims. On the other hand, a function recited in the claims may be fulfilled by more than one processor.

[0120] The terms computer program, software, application and the like as used herein, are defined as a sequence of instructions designed for execution on a computer system or data processing system or data processing terminal. A computer program, software, or application may include a subroutine, a function, a procedure, an object method, an object implementation, an executable application, an applet, a servlet, a source code, an object code, a shared library / dynamic load library and / or other sequence of instructions designed for execution on a computer system or data processing system, or on a data processing terminal interacting with a data processing system, to perform a method of the invention.

[0121] A computer program may be stored and / or distributed on a suitable medium, such as an optical storage medium or a solid-state medium supplied together with or as part of other hardware, but also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems.

Claims

- 27 -CLAIMS1 . A method, performed in a data processing system, of authenticating an object marker associated with an object, the method comprising: receiving object marker image data of the object marker; retrieving, from an original marker image database comprising original marker image data of a plurality of original marker images, original marker image data of an original marker; determining marker comparison parameters based on a comparison of the object marker image data to the retrieved original marker image data; inputting each one of the marker comparison parameters to a respective one of a plurality of input nodes of an artificial neural network, ANN, wherein the ANN further comprises a hidden layer and an output node, wherein the ANN has been trained to discriminate between object marker image data of the original marker, and object marker image data of a copy or clone of the original marker, and wherein the ANN is configured to provide output data, representative of the extent of authenticity of the object marker, at the output node; and outputting an authenticity signal indicating the authenticity of the object marker based on the output data of the ANN.

2. A method, performed in a data processing terminal configured to communicate with the data processing system according to claim 1 , of authenticating an object marker associated with an object, the method comprising: obtaining object marker image data of the object marker; transmitting the object marker image data to the data processing system; and receiving, from the data processing system, an authentication result signal based on the authenticity signal.

3. The method according to claim 1 or 2, wherein each original marker image carries unique original marker identifier data, and the step of retrieving the original marker image data comprises: deriving object marker identifier data from the object marker image data; and retrieving, from the original marker image database, the original marker image data associated with the original marker identifier data being identical to the object marker identifier data.

4. The method according to any one of the preceding claims, wherein the comparison of the object marker image data to the original marker image data comprises the steps of:processing the object marker image data and / or the original marker image data to align the object marker image to the original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the object marker image and the original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the object marker image and the original marker image as a marker comparison parameter.

5. The method according to any one of the preceding claims, further comprising a training of the ANN before the step of receiving object marker image data of the object marker, wherein the training of the ANN comprises the steps of: generating reference original marker image data of a reference original marker image of an original marker; generating additional original marker image data of a plurality of additional original marker images of the original marker, the plurality of additional original marker images being generated under varying circumstances; generating copy marker image data of a plurality of copy marker images of copies of the original marker; generating clone marker image data of a plurality of clone marker images of clones of the original marker;(a) determining at least one first training marker comparison parameter based on a comparison of additional original marker image data of an additional original marker image to the reference original marker image data, inputting each one of the first training marker comparison parameters to a respective one of a plurality of input nodes of the ANN, setting the output data at the output node of the ANN to be representative of the authenticity of the additional original marker image, and back-propagating the set output data to the hidden layer of the ANN;(b) determining at least one second training marker comparison parameter based on a comparison of copy marker image data of a copy marker image to the reference original marker image data, inputting each one of the second training marker comparison parameters to a respective one of a plurality of input nodes of the ANN, setting the output data at the output node of the ANN to be representative of non-authenticity of the copy marker image, and back-propagating the set output data to the hidden layer of the ANN;(c) determining at least one third training marker comparison parameter based on a comparison of clone marker image data of a clone marker image to the reference original marker image data, inputting each one of the third training marker comparison parameters to a respective one of a plurality of input nodes of the ANN, setting the output data at the outputnode of the ANN to be representative of non-authenticity of the clone marker image, and back-propagating the set output data to the hidden layer of the ANN;(d) for completing one training epoch, repeating step (a) for different combinations of additional original marker image data of one of the plurality of additional original marker images and the reference original marker image data, repeating step (b) for different combinations of copy marker image data of one of the plurality of copy marker images and the reference original marker image data, and repeating step (c) for different combinations of clone marker image data of one of the plurality of clone marker images and the reference original marker image data; and(e) repeating steps (a) to (d) in multiple epochs until the mean squared error of the output data of the ANN approaches or reaches a minimum.

6. The method according to claim 5, wherein the comparison of additional original marker image data of an additional original marker image to the reference original marker image data comprises the steps of: processing the additional original marker image data and / or the reference original marker image data to align the additional original marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the additional original marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the additional original marker image and the reference original marker image as a marker comparison parameter.

7. The method according to claim 5, wherein the comparison of copy marker image data of a copy marker image to the reference original marker image data comprises the steps of: processing the copy marker image data and / or the reference original marker image data to align the copy marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the copy marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the copy marker image and the reference original marker image as a marker comparison parameter.

8. The method according to claim 5, wherein the comparison of clone marker image data of a clone marker image to the reference original marker image data comprises the steps of:processing the clone marker image data and / or the reference original marker image data to align the clone marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the clone marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the clone marker image and the reference original marker image as a marker comparison parameter.

9. The method according to any one of the preceding claims, wherein the marker is a two- dimensional marker.

10. The method according to any one of the preceding claims, wherein the marker comprises a plurality of line segments each extending between a first end point and a second end point thereof, wherein the first end points and the second end points of the line segments are located on an edge line comprising at least one edge line segment.

11. A data processing system for authenticating an object marker associated with an object, the data processing system comprising: a receiving component for receiving object marker image data of the object marker; a retrieving component for retrieving, from an original marker image database comprising original marker image data of a plurality of original marker images, original marker image data of an original marker; a comparison component for determining marker comparison parameters based on a comparison of the object marker image data to the original marker image data; an input component for inputting each one of the marker comparison parameters to a respective one of a plurality of input nodes of an artificial neural network, ANN, wherein the ANN further comprises a hidden layer and an output node, wherein the ANN has been trained to discriminate between object marker image data of the original marker, and object marker image data of a copy or clone of the original marker, and wherein the ANN is configured to provide output data representative of the extent of authenticity of the object marker at the output node; and an output component for outputting an authenticity signal indicating the authenticity of the object marker based on the output data of the ANN.

12. The data processing system according to claim 11, wherein the original marker image database is comprised in the data processing system.- 31 -13. The data processing system according to claim 11 or 12, further comprising a camera coupled to the receiving component.

14. A data processing terminal for authenticating an object marker associated with an object, the data processing terminal comprising: an image input component for obtaining object marker image data of the object marker; a transmission component for transmitting the object marker image data to a data processing system according to claim 11 ; and a receiving component for receiving, from the data processing system, an authentication result signal based on the authenticity signal generated in the data processing system.

15. The data processing terminal according to claim 14, wherein the original marker image database is comprised in the data processing terminal.

16. The data processing terminal according to claim 14 or 15, wherein the image input component comprises a camera.

17. An artificial neural network, ANN, for use in the method of any one of claims 1 to 4, the ANN comprising a plurality of input nodes, a hidden layer and an output node, wherein the ANN has been trained to discriminate between object marker image data of the original marker, and object marker image data of a copy or clone of the original marker, wherein the ANN is configured to provide output data representative of the extent of authenticity of the object marker at the output node, and wherein the ANN has been trained by the steps of: generating reference original marker image data of a reference original marker image of an original marker; generating additional original marker image data of a plurality of additional original marker images of the original marker, the plurality of additional original marker images being generated under varying circumstances; generating copy marker image data of a plurality of copy marker images of copies of the original marker; generating clone marker image data of a plurality of clone marker images of clones of the original marker;(a) determining at least one first training marker comparison parameter based on a comparison of additional original marker image data of an additional original marker image to the reference original marker image data, inputting each one of the first training marker- 32 - comparison parameters to a respective one of a plurality of input nodes of the ANN, setting the output data at the output node of the ANN to be representative of the authenticity of the additional original marker image, and back-propagating the set output data to the hidden layer of the ANN;(b) determining at least one second training marker comparison parameter based on a comparison of copy marker image data of a copy marker image to the reference original marker image data, inputting each one of the second training marker comparison parameters to a respective one of a plurality of input nodes of the ANN, setting the output data at the output node of the ANN to be representative of non-authenticity of the copy marker image, and back-propagating the set output data to the hidden layer of the ANN;(c) determining at least one third training marker comparison parameter based on a comparison of clone marker image data of a clone marker image to the reference original marker image data, inputting each one of the third training marker comparison parameters to a respective one of a plurality of input nodes of the ANN, setting the output data at the output node of the ANN to be representative of non-authenticity of the clone marker image, and back-propagating the set output data to the hidden layer of the ANN;(d) for completing one training epoch, repeating step (a) for different combinations of additional original marker image data of one of the plurality of additional original marker images and the reference original marker image data, repeating step (b) for different combinations of copy marker image data of one of the plurality of copy marker images and the reference original marker image data, and repeating step (c) for different combinations of clone marker image data of one of the plurality of clone marker images and the reference original marker image data; and(e) repeating steps (a) to (d) in multiple epochs until the mean squared error of the output data of the ANN approaches or reaches a minimum.

18. The ANN according to claim 17, wherein the comparison of additional original marker image data of an additional original marker image to the reference original marker image data comprises the steps of: processing the additional original marker image data and / or the reference original marker image data to align the additional original marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the additional original marker image and the reference original marker image as a marker comparison parameter; and / or- 33 - determining an XOR Image Analysis score of differences in pixels between the additional original marker image and the reference original marker image as a marker comparison parameter.

19. The ANN according to claim 17, wherein the comparison of copy marker image data of a copy marker image to the reference original marker image data comprises the steps of: processing the copy marker image data and / or the reference original marker image data to align the copy marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the copy marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the copy marker image and the reference original marker image as a marker comparison parameter.

20. The ANN according to claim 17, wherein the comparison of clone marker image data of a clone marker image to the reference original marker image data comprises the steps of: processing the clone marker image data and / or the reference original marker image data to align the clone marker image to the reference original marker image; determining a Structural Similarity Index Measure, SSIM, score of differences in pixels between the clone marker image and the reference original marker image as a marker comparison parameter; and / or determining an XOR Image Analysis score of differences in pixels between the clone marker image and the reference original marker image as a marker comparison parameter.

21. A computer program comprising computer instructions which, when processed in a data processing system, cause the data processing system to carry out the method of any one of claims 1 and 3 to 10.

22. A computer program comprising computer instructions which, when processed in a data processing terminal, cause the data processing terminal to carry out the method of any one of claims 2 to 10.

23. A data carrier comprising computer instructions for the computer program of claim 21 or 22.

Images