Method for authenticating an object - Patent application

Abstract

The present invention relates to a method for authenticating an object 114. A pattern image is received showing the object while illuminated with a light pattern 113 having one or more pattern features. From the pattern image, a pattern feature located on the object is selected based on information indicating the location and extent of the object in the pattern image, and a plurality of cropped pattern images are generated by cropping the pattern image based on the selected pattern features, the cropped pattern images having a predetermined size and including at least a portion of one of the selected pattern features. The object is authenticated by providing the cropped images to a machine learning based identification model trained to be able to authenticate the object based on the cropped pattern image as input. The authentication of the object is then output.

Description

[Technical field]

[0001] The present invention relates to a method, an apparatus and a computer-readable data medium for authenticating an object.Furthermore, the present invention relates to a training method, a training apparatus and a training computer-readable data medium for training a machine learning based identification model suitable for authenticating an object, such that the identification model is available for the method, the apparatus and the computer-readable data medium for authenticating an object.Furthermore, the present invention relates to the use of the authentication of an object obtained by the method for authenticating an object for access control purposes. [Background technology]

[0002] It is generally known that neural networks can be trained to detect whether an image contains a desired object, such as a real face or a spoof mask, particularly for the purpose of identification in an unlocking process. However, it has been found that training such neural networks requires a large number of images, and even then, recognition is often unreliable. Moreover, the huge number of input images for training a neural network in this approach easily leads to overfitting of the neural network, which results in a further decrease in accuracy.

[0003] Therefore, it would be advantageous if a neural network could be trained with less training data, thus avoiding overfitting, while at the same time increasing the discrimination accuracy of the neural network for authenticating objects. Summary of the Invention [Problem to be solved by the invention]

[0004] It is an object of the present invention to provide a method, an apparatus and a computer-readable data medium for accurately authenticating an object, which allows for the use of less training data to train a machine learning-based identification model for authenticating the object. It is also an object of the present invention to provide a training method, a training apparatus and a computer-readable data medium that allows for the use of less training data and less computational resources to train the identification model, thereby providing an identification model that can be used in a method, an apparatus and a computer-readable data medium for authenticating an object. [Means for solving the problem]

[0005] In a first aspect of the present invention, a computer-implemented method for authenticating an object is provided, the method comprising: i) receiving a pattern image showing an object while the object is illuminated with a light pattern including one or more pattern features; ii) selecting a pattern feature located on the object from the pattern image based on information indicating a position and extent of the object in the pattern image; iii) generating a plurality of cropped pattern images by cropping the pattern image based on the selected pattern features, the cropped pattern image having a predetermined size and including at least a portion of one of the selected pattern features; iv) authenticating the object by providing the cropped pattern image to a machine learning based identification model trained to be capable of authenticating the object based on the cropped pattern image as an input; and v) outputting an authentication of the object.

[0006] Because the multiple cropped pattern images are generated by cropping the pattern image so that each cropped pattern image having a predetermined size includes at least a portion of one selected pattern feature selected to be located on the object, and because the cropped pattern images are used as inputs to the machine learning-based identification model, the complete pattern image, especially the complete pattern image that also includes a potentially huge amount of background, is not used for authentication. In particular, the cropped pattern image already focuses on the object to be authenticated without including a significant amount of background. Thus, the part of the training process of the identification model that is necessary to train the identification model to distinguish between the object and the potentially very diverse background can be avoided, and the required training data can be reduced. Furthermore, cropping the pattern image into smaller sections, i.e., splitting the object to be authenticated into multiple images each showing only a part of the object, has the further advantage that during the training of the machine learning-based identification model, authentication can be avoided from being strongly based on the correlation of features of completely different areas of the object. In this context, it has been found that cropping the pattern image forces the discrimination model to only base its recognition on the correlation of features that are close to each other on the object, i.e. features found in the same area of ​​the object, which leads to a higher recognition accuracy. Furthermore, it has been found that cropping the pattern image allows the use of machine learning models with fewer parameters, e.g. fewer neurons in the case of neural networks. This has the advantage that the risk of overfitting the machine learning model is reduced, which can lead to a higher reliability of the output of the machine learning model. Thus, the method allows the recognition of objects with higher accuracy and reliability, utilizing a machine learning discrimination model that can be trained with less computational cost and in particular with less training data.

[0007] The method relates to a computer-implemented method and can therefore be performed, for example, by a general-purpose or special-purpose computing device adapted to perform the method by executing a respective computer program. Furthermore, the method can also be performed by a plurality of computing devices, for example a computer network or any other kind of distributed computing, in which case the steps of the method can be performed by one or more computing units.

[0008] The method allows for authentication of objects. Authentication of objects refers in particular to identifying a particular object for the purpose of determining whether the particular object can access a predetermined resource. In general, identification of objects refers to determining the identity of the object. Identity can also refer to a general identity, e.g. a class identity indicating that the object is part of a predetermined object class, or a specific identity, which refers to determining whether the object refers to a predetermined unique object. An example of determining the general identity of an object refers, for example, to determining whether an object on an image is a human, i.e. belonging to the class of humans, or a chair, i.e. belonging to the class of chairs. Examples of specific identification can include identifying a predetermined individual, e.g. the owner of a smartphone, identifying a specific individual chair, e.g. a specific chair belonging to a specific owner, etc. Preferably, the method is adapted to authenticate humans in order to grant access to a locked resource with limited access. In a preferred embodiment, authentication of humans is utilized to grant or deny access to a computing device, such as a smartphone, laptop, tablet, etc., or to resources provided by the computing device, such as predetermined programs, digital payment options, etc.

[0009] In a first step, a pattern image showing the object is received. For example, the pattern image may be received from a camera unit that captures an image while the object is illuminated with a light pattern. However, the pattern image may also be received from a storage unit in which the pattern image is already saved. Furthermore, the pattern image may also be received by a user input, for example when the user indicates which image of a plurality of images stored in the storage should be used as the pattern image. A pattern image refers to an image captured while the object is illuminated with a light pattern including one or more pattern features. For example, the capture of the pattern image may be initiated by a user by providing a respective input to the respective device, in which case the light pattern generating unit may be adapted to generate the light pattern and the camera may be adapted to capture the pattern image while the object is illuminated with the light pattern. However, the generation of the light pattern and the capture of the image may also be performed automatically based on one or more predetermined events or may be continuous, for example for quality control of products.

[0010] In general, the light pattern on the object can be generated by any kind of light pattern generating unit. Preferably, the light pattern is generated using laser light, in particular infrared laser light. Using infrared light has the advantage that this light is less irritating to a human user when it is irradiated on the user's face. In particular, it is preferred to use one or more vertical cavity surface emitting lasers (VCSELs) to generate the light pattern comprising a plurality of laser light spots. However, other light sources can also be used to generate the light pattern, for example LED light sources of one or more colors. Preferably, the light pattern illuminating the object refers to a regular light pattern comprising regularly arranged pattern features. However, in other embodiments, the light pattern can refer to an irregular pattern or even to any pattern. In general, a pattern feature of a light pattern refers to a part of the light pattern that can be distinguished from other pattern features, for example due to the non-illuminated distance between the pattern features or due to a different arrangement of light in different pattern features. Preferably, the pattern feature refers to one or more light spots arranged in a predetermined pattern, which light pattern is preferably repeated in a predetermined pattern. In particular, the light pattern preferably refers to a cloud of points, which points refer to light spots, in which case the pattern feature may refer to one light spot. In this case, the light pattern may refer to, for example, a hexagonal or triclinic lattice of light spots that are substantially similar and include circular shapes. Utilizing a hexagonal or triclinic pattern for the light spots has the advantage that the arrangement of the light spots provides different distance relationships for the light spots, which prevents the risk of the machine learning-based identification model being misled by a pattern that is too regular during training. However, the pattern feature may also refer to one or more light spots, for example, a hexagon that includes six light spots, in which case, for example, the feature pattern, i.e., a hexagon, may be repeated to form a regular light pattern.

[0011] The method then further includes selecting a pattern feature located on the object from the pattern image based on information indicative of the location and extent of the object in the pattern image. The information indicative of the location and extent of the object in the pattern image can be received in a number of ways. For example, the pattern image can be presented to a user, who can optionally indicate the location and extent of the object in the pattern image based on a visible light image of the object. Furthermore, information from the pattern image, and in particular information from the pattern features in the pattern image, can itself be utilized to determine the location and extent of the object. For example, in a preferred embodiment, the selection of the pattern feature can include first deriving information indicative of the location and extent of the object from the pattern image. In this embodiment, known methods for deriving information from the pattern image can be utilized. In a preferred embodiment, known methods for determining the distance at which a pattern feature is reflected from the camera can be utilized to receive information regarding the extent and location of the object. For example, pattern features that are within a predetermined distance range relative to each other can be considered as belonging to the same object and can therefore be selected. Furthermore, a contour of the object can be determined, for example by comparing the distance of adjacent pattern features to each other. If the distance of adjacent pattern features exceeds a predetermined threshold, the contour can therefore be determined. Furthermore, in additional or alternative embodiments, information indicative of the location and extent of the object within the pattern image can also be derived from the pattern image by deriving the material from the reflectance properties of each pattern feature. In this case, already known methods for deriving material properties from the properties of reflected light can still be utilized, and it can be determined that the pattern feature indicative of the material associated with the object is selected as the pattern feature located on the object. For example, a pattern feature indicative of being reflected by human skin can in this case be determined as belonging to a human face and therefore selected as being located on the object.

[0012] In a further additional or alternative embodiment, a further flood light image is received and the selection of the pattern features is preferably based on selecting pattern features located on the object by determining a contour of the object indicative of the object's location and extent based on the flood light image and selecting pattern features lying within said contour, where the flood light image shows the object while it is illuminated with flood light. In this embodiment, instead of the flood light image, a natural light image showing the object illuminated by natural or artificial indoor light can also be used. The determination of the object contour indicative of the object's location and extent based on the flood light image can be performed according to any known feature extraction method for visible light images. In particular, the image is presented to a user and the user can indicate the contour of the object in the flood light image. However, more sophisticated automatic algorithms such as machine learning algorithms or simple feature detection algorithms can also be used. In general, in all embodiments utilizing multiple images, e.g. multiple pattern images for distance determination of feature patterns, or (one) flood light image for contour detection, it is preferred that the respective images are taken at the same time, or at least in a predetermined time range before and after the time the pattern image is taken, and further taken by the same camera or a camera having a predetermined distance to the camera taking the pattern image. This allows the location of features in one image, e.g. the flood light image, to be directly derived from the location of features in the pattern image. However, each image can also be pre-processed. This can include determining the location and range of objects in the images, and centering, scaling, size normalizing, and rotating each object such that a normalized orientation is provided for all images that allows the location of features in one image to be derived and this derived location to be transferred to the other image. For example, in the flood light image, a feature detection algorithm can detect the object, center the object in the image, and scale the image to a normalized scale.In the pattern image, the distance measurement of the feature pattern is also used to determine the contour of the object, and the pattern image can also be preprocessed to center the object and scale it to a normalized scale. Both images can therefore be used to transfer the position from one image to the other, even if the position is slightly shifted in the original image, for example because the user moves a little in front of the camera. In general, for the method to work accurately, the position and extent of the object only need to be approximately determined, i.e., the contour does not need to be determined exactly. Therefore, methods that only approximate the position and extent of the object can also be used, or methods such as those described above can be used with less precision, e.g., with less computational resources.

[0013] The pattern features located on the object are then selected from the pattern image, for example by determining whether the position of the pattern feature is located within the contour of the object. In this context, the selection of the pattern features can also include determining the position of each pattern feature in the pattern image. For example, a respective feature detection algorithm can be utilized. Since the pattern features have a predetermined shape and are also clearly distinguishable from other parts of the image that are not illuminated by the pattern feature, such a feature recognition method can be based on easy rules. For example, it can be determined that the pixels of the pattern image that contain a light intensity above a predetermined threshold are part of the pattern feature. Furthermore, the light intensity of neighboring pixels can also be taken into account to determine the position of the pattern feature, depending on the geometric shape of the pattern feature. Furthermore, a 2D shape recognition algorithm can also be utilized to recognize the predetermined 2D shape of the pattern feature. However, more sophisticated feature extraction methods can also be utilized or the user can perform the position determination by respective input. The pattern features on the object can then be selected by comparing the position of the pattern feature with the indicated position and range of the object and by selecting the pattern features that are within the boundary of the object.

[0014] However, in some embodiments, the determination of the position of the pattern feature can be omitted. For example, if information about the range and position of the object is derived for the pattern image itself, for example by determining the distance of the feature patterns relative to each other, the selection can also be based directly on the distance determination. In particular, in this example, pattern features that are adjacent to each other and within a predetermined distance range from each other can be directly selected as being on the object. Therefore, in such cases, there is no need to determine the position of the pattern feature.

[0015] In a further step, a plurality of cropped pattern images are generated by cropping the pattern image based on the selected pattern feature. Generally, cropping an image refers to removing all areas of the pattern image outside the cropped pattern image. Preferably, several cropped pattern images refer to at least two cropped pattern images, more preferably more than two cropped pattern images. Cropping the pattern image to generate the plurality of cropped pattern images is performed such that the cropped pattern images, e.g., each cropped pattern image, have a predetermined size and include at least a portion of the selected pattern feature, preferably one of the selected pattern features. A portion of the selected pattern feature can refer, for example, to one-half or one-quarter of the selected pattern feature. In particular, the complete selected pattern feature does not need to be part of the cropped image, since the algorithm can also be trained to recognize an object based on a cropped image including a portion of the selected pattern feature. In particular, when the selected pattern feature refers to a laser spot light, the selected pattern feature follows a Gaussian intensity function with a maximum intensity at the center of the selected pattern feature. In such a case, the identification model can perform authentication based on, for example, the characteristics of the intensity function, where, even if only a part of the intensity function, i.e. only a part of the selected pattern features, is visible in the cropped image, accurate authentication can be performed, since these parts already allow the respective characteristics of the intensity curve to be determined. In a preferred embodiment, the pattern image is cropped such that the cropped image includes parts of the selected pattern features. For example, the pattern image is cropped by utilizing the selected pattern features as boundary points of the cropping boundary of the cropped image. However, in order to increase the accuracy of authentication, it is preferred that the pattern image is cropped such that the cropped image includes at least one complete selected pattern image.The predetermined size can be determined, for example, in the form of a predetermined area to be covered by the cropped image, or in the form of any other characteristic that can determine the size of the area, for example a radius in the case of a circular area. Preferably, the cropped image refers to a rectangular image determined by a predetermined height and width. More preferably, the cropped image refers to a quadratic image. This has the advantage that the information around the pattern feature is weighted equally in all directions. However, the cropped images can generally have any shape and can therefore also be characterized by different size characteristics. For example, the cropped images can be circular and characterized by a respective radius. Preferably, each cropped image has the same predetermined size. However, for different cropped images, different predetermined sizes can also be utilized, for example, for an image cropped at the center of the object, a larger size can be predetermined than for an image cropped within a predetermined distance from the object's contour. Preferably, the cropped pattern image is positioned around the selected pattern feature. Further preferably, for each selected pattern feature, a cropped image is generated including at least the respective selected pattern feature, preferably centered within the cropped image. Further, the cropped pattern image preferably includes multiple selected pattern features, e.g., in addition to the central pattern feature, all or part of adjacent selected pattern features.

[0016] In a further step, the method includes authenticating the object by providing the cropped pattern image to a machine learning based identification model. The machine learning based identification model is trained so that it can authenticate the object based on the cropped pattern image as input. In general, the machine learning based identification model is trained using the cropped pattern image generated according to the same rules as the cropped pattern image that is subsequently used to authenticate the object. The machine learning based identification model can utilize any known machine learning algorithm. However, neural networks are particularly suitable for use in the context of image feature and object recognition techniques. Thus, the machine learning based identification model is preferably based on a neural network algorithm. More preferably, the neural network refers to a convolutional neural network. In this case, using the cropped pattern image is particularly suitable as an input for the machine learning based identification model, since the cropping of the pattern image prevents the convolutional neural network algorithm from basing the authentication decision too much on the correlation of features in completely different areas of the image. This leads to a significantly higher authentication accuracy and a reduction in false recognition. In a preferred embodiment, the discrimination model is adapted to learn to distinguish between different materials in the pattern image based on characteristics of the pattern features reflecting the respective materials, and to perform authentication based on the material distinction. Further details of such discrimination model are described, for example, in application WO 2020 / 187719 A1.

[0017] By its training, the trained identification model can thus authenticate the object when provided with the cropped pattern image as input. In particular, the output of the identification model can refer to simple verification or non-verification, or whether the object refers to a predetermined specific object such as an individual or to a general object class such as a human. However, the output can also refer to determining which of a predetermined number of object classes the authenticated object belongs to, or which of a predetermined number of specific objects it belongs to. For example, it can be determined that of three specific users A, B, and C of a device, the current user is identified as user A.

[0018] The result, i.e. the output of the identification model, may then be output. In particular, the method includes a step of outputting the authentication of the object. For example, the result of the identification model, i.e. the authentication of the object, may be provided to an output unit of the user device, which provides a visual or audio output to the user to inform the user of the result. Furthermore, the output may be further processed, e.g. to unlock a door or a device, if it is determined that the potential user identity allows the respective access. Furthermore, the output of the object authentication result may be utilized to control a device, not only to provide access to restricted resources, but also, for example, to control the movement or position of an automated device or robot based on the authentication of the object.

[0019] In a further aspect of the present invention, an apparatus for authenticating an object is presented, comprising: i) an input interface for receiving a pattern image indicative of an object while the object is illuminated with a light pattern including one or more pattern features; ii) a processor configured to: a) select pattern features located on the object from the pattern image based on information indicative of a position and extent of the object in the pattern image; b) generate a plurality of cropped pattern images by cropping the pattern image based on the selected pattern features, the cropped pattern images having a predetermined size and including at least a portion of one of the selected pattern features; and c) authenticate the object by providing the cropped pattern images to a machine learning based identification model trained to be capable of authenticating the object based on the cropped pattern image as an input; and iii) an output interface for outputting authentication of the object.

[0020] In a further aspect of the present invention, a method for training a machine learning based identification model suitable for authenticating an object is presented, the method comprising: i) receiving a training dataset based on a set of historical data including a) a cropped pattern image of the object; and b) an authenticity of the object, wherein a cropped image is generated from the pattern image, the pattern image showing the object while the object is illuminated with a light pattern including one or more pattern features, and generating the cropped pattern image includes: a) selecting a pattern feature located on the object from the pattern image based on information indicating a position and extent of the object in the pattern image; and b) generating a plurality of cropped pattern images by cropping the pattern image based on the selected pattern features, the cropped pattern images having a predetermined size and including at least a portion of the selected pattern features; ii) training a trainable machine learning based identification model by adjusting a parameterization of the identification model based on the training dataset, such that the trained identification model is adapted to authenticate the object when provided with the cropped pattern image of the object as an input; and iii) outputting the trained identification model.

[0021] In a further aspect of the present invention, an apparatus for training a machine learning based identification model suitable for authenticating an object is presented, the apparatus comprising: i) an input interface for receiving a training dataset based on a set of historical data including a) a cropped pattern image of the object; and b) an authenticity of the object, wherein a cropped image is generated from the pattern image, the pattern image showing the object while the object is illuminated with a light pattern including one or more pattern features, and generating the cropped pattern image includes a) selecting a pattern feature from the pattern image located on the object based on information indicating a position and extent of the object in the pattern image; and b) generating a plurality of cropped pattern images by cropping the pattern image based on the selected pattern features, wherein the cropped pattern image has a predetermined size and includes at least a portion of one of the selected pattern features; ii) a processor configured to train a trainable machine learning based identification model by adjusting a parameterization of the identification model based on the training dataset, such that the trained identification model is adapted to authenticate the object when provided with the cropped pattern image of the object as an input; and iii) an output interface for outputting the trained identification model.

[0022] In a further aspect of the invention, a use of the authenticity of an object obtained by the method described above is presented, comprising access control.

[0023] In a further aspect of the invention, a non-transitory computer readable data medium is presented, said data medium storing a computer program comprising instructions for causing a computer to perform the steps of the above-mentioned method.

[0024] In a further aspect of the invention, a non-transitory computer readable data medium is presented, said data medium storing a computer program comprising instructions for causing a computer to carry out the steps of the training method described above.

[0025] In a further aspect of the present invention, a method for determining authenticity of an object is presented, the method comprising the steps of: a) receiving a pattern image and a flood light image of the object, the pattern image showing the object while illuminated with a light pattern, and the flood light image showing the object while it is illuminated with flood light; b) determining a contour of the object from the flood light image; c) selecting a pattern feature from the pattern image that is located on the object based on the contour obtained from the flood light image; c) generating a plurality of cropped images by cropping the pattern image, the cropped image having a predetermined width and height and including one of the selected pattern features at its center; d) determining authenticity of the object by providing the cropped images to a neural network trained with a historical dataset including cropped images of the object; and e) outputting the authenticity of the object.

[0026] In a further aspect of the present invention, a system for determining authenticity of an object is presented comprising: a) an input for receiving a pattern image and a flood light image of the object, wherein the pattern image shows the object while illuminated with a light pattern and the flood light image shows the object while illuminated with flood light; b) a processor configured to: i) determine a contour of the object from the flood light image; ii) select a pattern feature from the pattern image that is located on the object based on the contour obtained from the flood light image; iii) generate a plurality of cropped images by cropping the pattern image, the cropped image having a predetermined width and height and including one of the selected pattern features at its center; and iv) determine authenticity of the object by providing the cropped images to a neural network trained with a historical dataset comprising cropped images of the object; and c) an output for outputting the authenticity of the object.

[0027] In a further aspect, a method for training a neural network suitable for determining authenticity of an object is presented, the method comprising: a) receiving a training dataset based on a set of historical data including cropped images of an object and authenticity of the object, the cropped images being generated from a pattern image and a flood light image of the object, the pattern image showing the object while illuminated with a light pattern, and the flood light image showing the object while illuminated with flood light, generating the cropped images comprising: b) determining a contour of the object from the flood light image, selecting a pattern feature from the pattern image that is located on the object based on the contour obtained from the flood light image, and generating a plurality of cropped images by cropping the pattern image, the cropped image having a predetermined width and height and including one of the selected pattern features at its center; c) training the neural network by adjusting a parameterization according to the training dataset; and d) outputting the trained neural network.

[0028] It is to be understood that the above-mentioned method, the above-mentioned device and the above-mentioned computer-readable data medium for authenticating an object have similar and / or identical preferred embodiments, in particular as defined in the dependent claims. Furthermore, the above-mentioned training method, the above-mentioned training device and the above-mentioned training computer-readable data medium have similar and / or identical preferred embodiments, in particular as defined in the dependent claims.

[0029] It is to be understood that a preferred embodiment of the invention can also be any combination of the dependent claims or the above embodiments with the respective independent claim.

[0030] These and other aspects of the invention will be apparent from and elucidated with reference to the embodiment(s) described hereinafter. [Brief description of the drawings]

[0031] [Figure 1] FIG. 1 illustrates a schematic and exemplary embodiment of a system including an apparatus for authenticating an object. [Diagram 2] FIG. 2 shows, diagrammatically and exemplarily, a flow chart of a method for authenticating an object. [Diagram 3] FIG. 1 shows, diagrammatically and exemplarily, a flow chart of a method for training a discriminative model for authenticating an object. [Figure 4] FIG. 1 illustrates a schematic and exemplary image recording device; [Diagram 5] FIG. 2 shows a schematic and exemplary image processing device that can be used by a device for authenticating an object; [Figure 6] FIG. 2 shows a schematic and exemplary diagram of a more detailed embodiment of a method for authenticating an object. [Figure 7] 3A-3C show schematic and exemplary diagrams of cropping an image according to a method for authenticating an object; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0032] Detailed Description of the Embodiments FIG. 1 shows an embodiment of a system 100 comprising a locking device 110, an apparatus 120 for authenticating an object, and a training apparatus 130 for training an identification model utilized in the apparatus 120. The locking device 110 generally represents a device or part of a device adapted to manage the access of a user or an object to further resources. For example, the locking device 110 may be part of a user device such as a smartphone and may manage the user's access to the smartphone and / or to the smartphone's resources. However, the locking device 110 may also represent a door lock mechanism adapted to manage the access of a person to a restricted area, such as, for example, an office building, a laboratory, or other area. In a further example, the locking device 110 may also refer to an access control system in a sorting facility in which a number of products are sorted according to predefined classes, in this example the locking device 110 manages which additional procedures an authenticated object 114 has access to. According to the examples of the locking device 110, the object 114 to be authenticated may also refer to any object. Preferably, the object 114 may refer to a human being or a part of a human being such as a face, but the object may also refer to an inanimate object such as any kind of industrial product.

[0033] The locking device 110 comprises a light pattern generating unit 111 adapted to generate a light pattern 113 on at least one surface of the object 114. The light pattern may refer to any predetermined light pattern. However, preferably, the light pattern is generated by the light pattern generating unit 111 by utilizing infrared laser light. In general, the light pattern 113 comprises a plurality of pattern features which refer to portions of the light pattern that together form a light pattern. Furthermore, the light pattern preferably refers to a regular light pattern of light spots, such as pattern features which may be arranged in a triangular, cubic or hexagonal pattern. Furthermore, the locking device 110 comprises a camera 112 adapted to receive light reflected by the object 114 illuminated with the light pattern 113 and to generate a pattern image of the object 114 from the reflected light. In particular, the pattern image generated by the camera 112 shows the light pattern 113 reflected by the object 114.

[0034] Furthermore, the system 100 comprises a device 120 for authenticating an object. In this preferred example, authenticating an object refers not only to authenticating the object, i.e. to determining whether the object 114 refers to a specific object, e.g. a specific user, but also to determining whether the object 114 refers to a specific object, e.g. a specific user. The device 120 comprises an interface unit 121 for receiving input data, a processor 120 for generally processing the input data, and an output interface 123 for generally outputting data. For example, the device 120 can be part of the same user equipment as the lock device 110, e.g. part of a computing unit of a smartphone or tablet. However, the device can also be a standalone device or part of a general network or server system, in which case it is preferably communicatively coupled to the lock device 110.

[0035] The input interface is particularly adapted to receive the pattern image generated by the camera 112 of the locking device 110. Optionally, the input interface can be further adapted to receive further data from the locking device 110 or from another device, e.g. a display. For example, if the camera 112 of the locking device 110 is further adapted to generate a visible light image of the object 114 not illuminated by the light pattern or a flood light image of the object 114 generated by the object 114 while illuminated by the flood light, the input interface can be further adapted to receive this additional image. Additionally or alternatively, the input interface can also receive information indicating the position and range of the object 114 within the pattern image from another device, e.g. a user input device where a user inputs this information. The input interface 121 is then adapted to provide the received data to the processor 122 for further processing.

[0036] The processor 122 is then adapted to select pattern features located on the object from the pattern image, for example by processing the respective computer control signals. In particular, the selection of the pattern features is based on information indicative of the position and extent of the object in the pattern image. Furthermore, the selection may also include determining the position of the pattern features in the pattern image, for example via feature recognition methods. For example, since the general light patterns utilized to illuminate the object 114 are known, a respective known feature algorithm adapted to recognize the light patterns in the pattern image may be utilized. In particular, a trained machine learning algorithm is preferably utilized to determine the position of the pattern features in the pattern image. Furthermore, the information indicative of the position and extent of the object in the pattern image may be provided or determined in a number of different ways. For example, the information may be provided by a user based on the pattern image or based on a visible light image, such as a flood light image. For example, the user may utilize the input unit to indicate the position and extent of the object on the flood light image, for example by tracing the contour of the object in the flood light image. In this case, the flood light image is preferably taken such that there is no substantial difference in the position and extent of the object 114 between the pattern image and the flood light image. The contours thus shown provide information about the location and extent of the object in the pattern image if the functional relationship between the location in the pattern image and the location in the flood light image is known, for example because both were taken by the same camera, or by a camera that provides a pre-determined functional relationship. However, the selection of the pattern features can also include automatically determining information indicative of the location and extent of the object. For example, the pattern features themselves allow for deriving information about the reflective properties of the object 114. Thus, for example, based on the pattern image, it can be determined whether a certain pattern feature is reflected by a material that refers to the expected material of the object 114, or by something different. An example of this is described, for example, in application WO 2020 / 187719 A1.In this way, information about the location and range of the object can also be determined based on the reflection properties of the pattern features. Furthermore, the pattern features using laser light also allow for distance determination between the reflection of the pattern features visible in the pattern image and the camera. Since for most objects 114, it is expected that all pattern features reflected by the object will be found within a certain predeterminable distance range from each other, while pattern features reflected from the background of the object may be expected to show quite different distance patterns, the distance information provided by the pattern features can also be utilized to provide information indicative of the location and range of the object. Furthermore, additional inputs can also be utilized to derive information indicative of the location and range of the object, such as when the distance of the pattern features is determined with reference to further pattern images, for example, using two images taken from slightly different angles to utilize the parallax effect for distance determination. Furthermore, in the preferred embodiment described in detail with respect to Figures 5 and 6, a visible light image, for example a flood light image, is provided and utilized to determine information about the location and range of the object, in particular to determine the contour of the object. Preferably, known feature recognition algorithms can be utilized to determine the location and range of the object, for example by determining the contour of the object. A preferred embodiment utilizes machine learning algorithms such as neural networks that are trained to determine the contours of objects in an image.

[0037] Based on the selected pattern features, a number of cropped pattern images are generated by cropping the pattern image. In particular, the pattern image is cropped such that the cropped pattern image has a predetermined size and includes at least a part of one selected pattern feature. In general, the cropping can be performed according to any predetermined rule that satisfies the above-mentioned conditions. The respective rule is generally the same rule applied for the cropped image for which the respective discrimination model for which the cropped pattern image is used as input was trained. However, it has been found to be particularly advantageous to generate for each selected pattern feature a cropped pattern image centered on the selected pattern feature in order to obtain accurate authentication results with less training data. Furthermore, it is preferred that the size of the cropped pattern image is predetermined such that each cropped pattern image includes at least two selected pattern features, more preferably such that the cropped pattern image includes all adjacent pattern features of the pattern feature on which it is centered, but is small enough not to cover all selected pattern features. In general, the optimal predetermined size of the cropped pattern image can be based on the respective application, for example based on the object to be authenticated. For example, such an optimal size can be found for each application by, for example, training a discriminative model at each different size and comparing the accuracy and reliability of each trained discriminative model, however, the discriminative model can operate with adequate accuracy for any size of cropped image that falls within the above range.

[0038] The cropped pattern image thus determined is then provided as input to an identification model trained to verify the authentication of the object based on the cropped pattern image provided as input, in particular whether the object refers to a particular object or not. Preferably, the identification model is trained by utilizing a training device 130 adapted to train a machine learning based identification model. The training device 130 can be realized as part of the device 120, for example utilizing the same processor and / or the same output and input interfaces. However, the training device 130 can also be realized as part of an entirely different computing device, for example provided on a server or a computing network, and then preferably communicatively coupled to the device 120 in order to provide the trained identification model to the device 120.

[0039] In general, the training device 130 may comprise an input interface 131 for receiving a training data set for training the identification model. The training data set preferably refers to historical data including a) cropped pattern images of objects and b) the authenticity of the objects. In particular, the training data set is preferably provided for a number of different objects accordingly. The selection of the objects provided in the training data set may be based, for example, on the intended application of the identification model. For example, if the identification model is to be utilized for authenticating individual users, the training data set may include a number of human faces and also a number of images of the individual users to be authenticated, each authentication being provided with a cropped pattern image of the different users' faces, for example as an annotation. In general, the same rules for cropping the pattern images are applied for training the identification model as are applied later, for example, by the device 120 when the object 141 is to be authenticated. Thus, the cropped pattern images in the training data set may also be generated according to the principles and rules described above. The processor 132 of the training device 130 is then configured to train a trainable machine learning based discriminatory model based on the provided training data set, in particular by adjusting the parameterization of the discriminatory model. Preferably, the discriminatory model is based on a neural network algorithm, in particular a convolutional neural network algorithm, where parameterization refers to determining the respective parameters of the neural network. In particular, any known training method for training a respective machine learning based algorithm, such as a neural network or a convolutional neural network, may be utilized by the processor 132. After the training process, the discriminatory model is adapted to recognize that it is the respective trained object when a cropped pattern image of the object is provided as input.Such a trained discrimination model may then be provided by the processor 132 to an output interface 133 of the training device 130 in order to provide the discrimination model to the device 120 , and in particular to the processor 122 for authenticating the object 114 .

[0040] The processor 122 then utilizes an identification model, for example provided by the training device 130 and then stored in the local storage of the device 120, and provides the cropped pattern image as input to the identification model. Based on the input cropped pattern image of the object 114, the identification model is then adapted to provide as output an authentication of the object 114. A result of this authentication can then be output by the output interface 123 of the device 120. In particular, the output may comprise providing the authentication information of the object to the user, for example by a respective visual or audio output. However, it is preferred that the output is provided in particular to the locking device 120, which is then adapted to manage the locking or unlocking of the respective resource based on the determined authentication of the object. For example, if it is verified that the object 114, in particular the user, has access to the respective resource, the locking device 110 may be adapted to provide this access to the respective user. However, if it is determined that the object 114 does not have access to the respective requested resource, the locking device 110 may be adapted to deny the respective access.

[0041] FIG. 2 shows a schematic and exemplary computer-implemented method 200 for authenticating an object. In particular, the computer-implemented method 200 comprises the functionality as described above with respect to the device 120 shown in FIG. 1. In a first step 210, the method 200 comprises receiving a pattern image showing the object while illuminated with a light pattern including one or more pattern features. In step 220, a pattern feature located on the object is then selected from the pattern image based on information indicating the location and extent of the object in the pattern image. Furthermore, in step 230, a plurality of cropped pattern images are generated by cropping the pattern image based on the selected pattern features. In particular, the cropped pattern images are generated according to the principles and rules as described above with respect to FIG. 1. Furthermore, in step 240, the object is authenticated by providing the cropped pattern image as an input to a machine learning based identification model trained to be able to authenticate the object based on the cropped pattern image as an input. In step 250, the result provided by the identification model, i.e. the authentication of the object, is provided as an output, for example to use the authentication to unlock a resource.

[0042] FIG. 3 shows, in a schematic and exemplary manner, a computer-implemented method 300 for training a machine learning-based identification model suitable for object authentication and used in the method described with respect to FIG. 2. In particular, the method 300 comprises, in a first step 310, receiving a training data set based on historical data including a) cropped pattern images of the object and b) authentication of the object. In particular, the training data set may be provided according to the rules and principles described above with respect to the training device 130 of FIG. 1. In step 320, a trainable machine learning-based identification model is then trained by adjusting a parameterization of the identification model based on the training data set such that the trained identification model is then adapted to authenticate the object when provided with a cropped image of the object as input. In step 330, the trained identification model may then be output, for example provided to the device 120 as described with respect to FIG. 1.

[0043] In the following, further preferred and more detailed embodiments of the present invention will be described. In general, a neural network can be trained to detect whether an image contains a desired object, such as a real face or a spoof mask, as authentication for the unlocking process. However, if the entire image is taken as input, a large number of images are required and the recognition reliability is often low. Moreover, this approach requires a large number of input parameters for the neural network, which is prone to overfitting.

[0044] The inventors have found that the above problem is solved by splitting the pattern image into partial pattern images and using only those partial pattern images that contain the object to be authenticated. To this end, in a preferred embodiment, an image recording device, for example a mobile phone, can be provided with two projectors (one for illuminating a flood light, for example an LED, and another for illuminating a light pattern, for example a VCSEL array, as shown in FIG. 4). A camera of the image recording device can then capture the object illuminated by the flood light at one time and by the light pattern at another time. These images, i.e. the pattern image and the flood light image, are then passed to an image processor, which may be configured to execute a neural network. The image processor may refer to a realization of the processor 122 of the device 120 described with reference to FIG. 1. In a preferred embodiment, the processor is adapted to perform at least the image processing in a secure environment to avoid external access to the operation. A schematic example of such a system is shown in FIG. 5.

[0045] FIG. 6 shows a preferred example of the respective image processing for determining the cropped pattern image. In this example, in a first step, preferably two different images are received from the object, namely a pattern image and a flood light image. To generate the pattern image, the object is illuminated with pattern light and recorded by a corresponding camera, for example an IR camera. The pattern is typically regular, i.e. it contains repeating features. Particularly preferred patterns are point clouds, for example a hexagonal or triangular lattice of spots, somewhat similar to a circle. To generate the flood light image, the object is illuminated with flood light or, optionally, only with ambient light, and recorded by a corresponding camera.

[0046] In the laser spot image, i.e. the pattern image, the position of each laser spot can be determined, for example, by determining a local intensity maximum in the pattern image. In the flood light image, objects can be identified by their shape. There are various known methods for this, such as convolutional neural networks trained on certain objects (e.g. faces). However, methods that are not based on machine learning methods, such as rule-based methods, can also be used. Once objects are identified in the flood light image, the image can be further pre-processed. This pre-processing can include centering the objects, scaling them to a normalized size, and / or rotating them to a normalized orientation.

[0047] Information about the contours of the target object from the flood light image can then be used to indicate the location and extent of the object in the pattern image and can be used to find their pattern features, such as light spots in the pattern image located on the object. Preferably, for each such pattern feature, the pattern image is cropped such that the cropped pattern image shows the pattern feature at the center of the pattern feature and a preset amount of the neighborhood of the pattern feature. Preferably, the cropped image includes not only the central pattern feature, but also some of the neighborhood features. The selected cropping size is a compromise between overfitting the neural network if too large an image is cropped, and too low accuracy of object recognition if too small an image is cropped. Thus, the size of the cropped image can be selected based on the requirements of a particular use case.

[0048] FIG. 7 shows an example of cropping a face image. For simplicity and to better illustrate the principle, the flood light image and the pattern image are superimposed in the left image of FIG. 7. The face contour is identified from the flood light image such that only the spots of the laser spot image that fall on the face area are selected. This selection is shown in the center image of FIG. 7. Now, for each light spot as a pattern feature, a cropped image is generated that has the light spot at the center and also includes some neighborhood around the light spot. For three spots, the respective bounding boxes are shown in the right image of FIG. 7.

[0049] The set of partial images obtained by cropping around the pattern features is used as input for a neural network trained on a historical dataset generated in the same way as described above. This approach has the advantage that the neural network requires significantly fewer input parameters and therefore fewer nodes. This reduces the demand for training images, i.e. images that are known to contain the object of interest or not, as well as spoofed images. It also reduces the risk of overfitting the network, improving recognition accuracy. In particular, when convolutional neural networks (CNNs) are used, this cropping approach suppresses the relationships of distant object features, which are usually not very useful for object authentication, rather close relationships are important.

[0050] The trained neural network can then be used to authenticate objects. An example is face recognition, where an image of a person is taken and evaluated whether it is really a person authorized to enter a particular application, or a different person or a mask of an authorized person. In this case, the neural network can use inputs such as those described above and have as output a decision whether this is the correct person or not.

[0051] Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims.

[0052] For the processes and methods disclosed herein, the operations performed in the processes and methods may be performed in different orders. Furthermore, the operations described are provided only as examples, and some of the operations may be optional, combined into fewer steps and operations, supplemented with additional operations, or expanded into additional operations, without detracting from the essence of the disclosed embodiments.

[0053] In the claims, the word "comprising" does not exclude other elements or steps and the indefinite article "a" or "an" does not exclude a plurality.

[0054] A single unit or device may fulfill the functions of several items recited in the claims. 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.

[0055] The procedures performed by one or more units or devices, such as receiving the pattern image, selecting the characteristic pattern, cropping the pattern image, determining the authenticity of the object, etc., can be performed by any number of other units or devices. These procedures can be implemented as program code means of a computer program and / or as dedicated hardware.

[0056] The computer program product may be stored / distributed on a suitable medium, such as an optical storage medium or a solid-state medium, may be supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunications systems.

[0057] The units described herein may be processing units that are part of a classical computing system. The processing unit may include a general-purpose processor, or may include a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), or any other specialized circuit. The memory may be a physical system memory, which may be volatile, non-volatile, or a combination of the two. The term "memory" may include a computer readable storage medium, such as a non-volatile mass storage. If the computing system is distributed, the processing and / or memory capabilities may also be distributed. The computing system may include multiple structures as "executable components." The term "executable components" is a structure well understood in the computing arts as a structure that may be software, hardware, or a combination thereof. For example, when implemented in software, one skilled in the art will understand that the structure of the executable components may include software objects, routines, methods, etc. that may be executed on the computing system. This may include both executable components in the heap of the computing system or executable components on a computer readable storage medium. The structure of the executable components may reside on a computer-readable medium and, when interpreted by one or more processors, e.g., processor threads, of a computing system, cause the computing system to perform functions. Such structure may be directly computer readable by the processor, e.g., as where the executable components were binary, or may be structured to be interpretable and / or compilable to generate such binary directly interpretable by the processor, e.g., in a single stage or multiple stages. In other examples, the structure may be hard-coded or hard-wired logic gates that are implemented exclusively or nearly exclusively in hardware, such as in a field programmable gate array (FPGA), application specific integrated circuit (ASIC), or other specialized circuitry.Thus, the term "executable component" is a term that describes a structure well understood by one of ordinary skill in the art of computing, whether implemented in software, hardware, or a combination thereof. Any embodiment herein is described with reference to acts performed by one or more processing units of a computing system. When such acts are implemented in software, one or more processors direct the operation of the computing system in response to executing the computer-executable instructions that make up the executable component. A computing system may also include communication channels that enable the computing system to communicate with other computing systems, for example, over a network. A "network" is defined as one or more data links that enable the transmission of electronic data between computing systems and / or modules and / or other electronic devices. When information is transferred or provided to a computing system over a network or other communications connection (e.g., either wired, wireless, or a combination of wired or wireless), the computing system properly regards the connection as a transmission medium. A transmission medium may include a network and / or data links that may be used to transmit desired program code means in the form of computer-executable instructions or data structures and that may be accessed by a general-purpose or special-purpose computing system or combination thereof. Although not all computing systems require a user interface, in some embodiments a computing system includes a user interface system for use in interfacing with a user. The user interface serves as an input or output mechanism to a user, for example via a display.

[0058] Those skilled in the art will appreciate that at least a portion of the present invention may be implemented in a networked computing environment having many types of computing system configurations, including personal computers, desktop computers, laptop computers, message processors, handheld devices, multiprocessor systems, microprocessor-based or programmable consumer electronics, network PCs, minicomputers, mainframe computers, cell phones, PDAs, pagers, routers, switches, data centers, wearables such as glasses, etc. The present invention may also be implemented in a distributed system environment where tasks are performed by both local and remote computing systems that are linked, for example, via a network, by either wired data links, wireless data links, or a combination of wired and wireless data links. In a distributed system environment, program modules may be located in both local and remote memory storage devices.

[0059] Those skilled in the art will also appreciate that at least a portion of the present invention may be implemented in a cloud computing environment. A cloud computing environment may be distributed, but this is not required. If distributed, a cloud computing environment may be distributed internationally within an organization and / or may have components owned across multiple organizations. For purposes of this specification and the claims that follow, "cloud computing" is defined as a model that enables on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services). The definition of "cloud computing" is not limited to any of the many other advantages that such a model may derive when deployed. The computing system of the figure includes various components or functional blocks that may implement various embodiments disclosed herein, as described. The various components or functional blocks may be implemented on a local computing system or may be implemented on a distributed computing system that includes elements that reside in the cloud or that implement aspects of cloud computing. The various components or functional blocks may be implemented as software, hardware, or a combination of software and hardware. The computing system shown in the figure may have more or fewer components than those shown in the figure and may combine some of the components as appropriate.

[0060] Any reference signs in the claims shall not be construed as limiting the scope.

[0061] The present invention relates to a method for authenticating an object, comprising: receiving a pattern image showing an object while illuminated with a light pattern including one or more pattern features; selecting a pattern feature located on the object from the pattern image based on information indicating the location and extent of the object in the pattern image; generating a plurality of cropped pattern images by cropping the pattern image based on the selected pattern features, the cropped pattern images having a predetermined size and including at least a portion of one of the selected pattern features; authenticating the object by providing the cropped pattern image to a machine learning based identification model trained to be capable of authenticating the object based on the cropped pattern image as an input; and outputting the authentication of the object.

Claims

1. 1. A computer-implemented method for authenticating an object (114), the method comprising: receiving (210) a pattern image showing an object (114) while illuminated with a light pattern (113) including one or more pattern features; selecting (220) pattern features located on the object (114) from the pattern image based on information indicating the location and extent of the object (114) within the pattern image; generating (230) a plurality of cropped pattern images by cropping the pattern image based on the selected pattern features, the cropped pattern images having a predetermined size and including at least a portion of one of the selected pattern features; authenticating the object (114) by providing the cropped pattern image to a machine learning-based discrimination model trained to be able to authenticate the object (114) based on the cropped pattern image as input; outputting (250) an authentication of the object (114); A method comprising:

2. 2. The method of claim 1, further comprising receiving a flood light image, wherein the selection of the pattern features is based on determining a contour of the object (114) indicating a location and extent of the object (114) based on the flood light image, and selecting pattern features located on the object (114) by selecting pattern features located within the contour, wherein the flood light image shows the object (114) while illuminated with flood light.

3. The method of claim 1 or 2, wherein the pattern image is cropped such that the cropped image is centered around a selected pattern feature.

4. The method of claim 1 or 2, wherein for each selected pattern feature, a cropped pattern image is generated that includes at least the respective selected pattern feature.

5. The method of claim 1 or 2, wherein the size of the cropped image is determined by a predetermined height and width.

6. The method according to claim 1 or 2, wherein the machine learning based discriminative model is based on a neural network algorithm, preferably a convolutional neural network.

7. The method of claim 1 or 2, wherein the light pattern (113) refers to a regular pattern comprising regularly arranged pattern features.

8. The method of claim 1 or 2, wherein a pattern feature refers to one or more light spots arranged in a predetermined pattern, and the light pattern (113) is based on a repetition of the predetermined pattern.

9. The method according to claim 1 or 2, wherein the light pattern (113) is generated by utilizing laser light, preferably infrared laser light.

10. An apparatus (120) for authenticating an object (114), said apparatus (120) comprising: an input interface (121) for receiving a pattern image showing an object (114) while illuminated with a light pattern (113) including one or more pattern features; A processor (122) comprising: selecting, from the pattern image, pattern features located on the object (114) based on information indicating the location and extent of the object (114) within the pattern image; generating a plurality of cropped pattern images by cropping the pattern image based on the selected pattern features, the cropped pattern images having a predetermined size and including at least a portion of one of the selected pattern features; authenticating the object (114) by providing the cropped pattern image as an input to a machine learning-based discrimination model trained to be capable of authenticating the object (114) based on the cropped pattern image; a processor (122) configured to: an output interface (123) for outputting the authentication of said object (114); An apparatus (120) comprising:

11. 1. A method for training a machine learning-based discrimination model suitable for authenticating an object (114), the method (300) comprising: a step (310) of receiving a training dataset based on a set of historical data including a) cropped pattern images of an object (114) and b) authenticity of the object (114), wherein the cropped images are generated from a pattern image, the pattern image showing the object (114) while illuminated with a light pattern (113) including one or more pattern features, and generating the cropped pattern images includes: a) selecting, from the pattern image, pattern features located on the object (114) based on information indicating a position and extent of the object (114) within the pattern image; and b) generating a plurality of cropped pattern images by cropping the pattern image based on the selected pattern features, wherein the cropped pattern images have a predetermined size and include at least a portion of one of the selected pattern features; training (320) a trainable machine learning-based discriminative model by adjusting parameterization of the discriminative model based on a training dataset, such that the trained discriminative model is adapted to authenticate the object (114) when provided with a cropped pattern image of the object (114) as input; and outputting the trained discriminative model (330).

12. 1. An apparatus for training a machine learning-based discrimination model suitable for authenticating an object (114), the apparatus (130) comprising: an input interface (131) for receiving a training dataset based on a set of historical data including: a) a cropped pattern image of an object (114); and b) an authenticity of the object (114), wherein the cropped image is generated from a pattern image, the pattern image showing the object (114) while illuminated with a light pattern (113) including one or more pattern features, and generating the cropped pattern image includes: a) selecting, from the pattern image, a pattern feature located on the object (114) based on information indicating a position and extent of the object (114) within the pattern image; and b) generating a plurality of cropped pattern images by cropping the pattern image based on the selected pattern features, wherein the cropped pattern image has a predetermined size and includes at least a portion of one of the selected pattern features; a processor (132) configured to train a trainable machine learning based discrimination model by adjusting parameterization of the discrimination model based on the training dataset such that the trained discrimination model is adapted to authenticate the object (114) when provided with a cropped pattern image of the object (114) as input; an output interface (13) for outputting the trained discriminative model; An apparatus (130) comprising:

13. Use of an object (114) obtained by the method according to claim 1 or 2 for access control of the authenticity.

14. A non-transitory computer-readable data medium storing a computer program comprising instructions for causing a computer to carry out the steps of the method according to claim 1 or 2.

15. A non-transitory computer-readable data medium storing a computer program comprising instructions for causing a computer to perform the steps of the method of claim 11.

Images