Secure authentication

Abstract

A method for measuring blood perfusion of an object, the method comprising: receiving an image generated while the object is illuminated by coherent infrared light, generating a plurality of partial images based on the image, providing the plurality of partial images to a data-driven model based on a location of the plurality of partial images within the image for determining a material property measure, wherein the data-driven model is trained based on historical partial images provided to the data-driven model based on a location of the historical partial images and corresponding material property measures, providing the material property measure.

Description

[0001] Secure Authentication

[0002] Technical Field

[0003] The disclosure relates to secure authentication and provides methods for allowing an object to access a resource, methods for measuring blood perfusion of an object, use of a material property, use of a plurality of partial images, a non-transitory computer-readable storage medium, use of data-driven model, a device and / or system for allowing an object to access a resource.

[0004] Technical Background

[0005] Authentication processes can be spoofed by masks, images or the like representing a user's characteristics. Spoofing items are becoming more realistic and followingly, distinguishing between spoofing object and human becomes more difficult.

[0006] Hence, there is a need to reliably differentiate between humans and spoofing items.

[0007] An object of the present disclosure is to provide a robust, reliable and secure method for authenticating humans.

[0008] Summary

[0009] In an aspect, this disclosure relates to a method for allowing an object to access a resource, the method comprising: receiving a request to access a resource, in response to receiving the request to access the resource triggering to illuminate the object by coherent infrared light, and triggering to generate an image of the object while the object is being illuminated by the coherent infrared light, generating a plurality of partial images based on the image, providing the plurality of partial images based on a location of the plurality of partial images within the image to a data-driven model for determining if the object is a living organism, wherein the data-driven model is trained based on a plurality of historical partial images to determine whether the one or more objects associated with the plurality of historical partial images are living organisms, allowing the object to access a resource based on determining that the object is a living organism.

[0010] In another aspect, it relates to a method for measuring blood perfusion of an object, the method comprising: receiving an image generated while the object is illuminated by coherent infrared light, generate a plurality of partial images based on the image, providing the plurality of partial images to a data-driven model based on a location of the plurality of partial images within the image for determining a blood perfusion measure, wherein the data-driven model is trained based on historical partial images and corresponding blood perfusion measures, optionally receiving the blood perfusion measure from the data-driven model, providing the blood perfusion measure.

[0011] In an aspect, it relates to a method for measuring blood perfusion of an object, the method comprising: receiving an image of the object under illumination by coherent infrared light, generating a plurality of partial images based on the image, determining a blood perfusion measure from the partial images and the location of the partial images within the image by providing the plurality of partial images to a data-driven model, wherein the data- driven model is trained based on historical partial images provided to the data-driven model based on a location of the historical partial images and corresponding blood perfusion measures, providing the blood perfusion measure.

[0012] In another aspect, it relates to a method for allowing an object to access a resource, the method comprising: receiving a request to access a resource, in response to receiving the request to access the resource triggering to illuminate the object by coherent infrared light, and triggering to generate an image of the object while the object is being illuminated by the coherent infrared light, generating a plurality of partial images based on the image, determining if the object is a living organism from the partial images and a location of the partial images within the image by providing the plurality of partial images to a data-driven model, wherein the data-driven model is trained based on a plurality of historical partial images provided based on a location of the plurality of historical partial images to determine whether the one or more objects associated with the plurality of historical partial images are living organisms, allowing the object to access the resource based on determining that the object is a living organism.

[0013] In another aspect, it relates to use of a indication of a material property generated as described herein for allowing an object to access a resource.

[0014] In another aspect, it relates to a method for measuring blood perfusion of an object, the method comprising: receiving an image of the object under illumination by coherent infrared light, generating a plurality of partial images from the image, providing the plurality of partial images to a data-driven model based on a location of the plurality of partial images within the image for determining a blood perfusion measure, wherein the data-driven model is trained with historical partial images provided to the data-driven model based on a location of the historical partial images and corresponding blood perfusion measures, providing the blood perfusion measure.

[0015] In another aspect, it relates to a method for allowing an object to access a resource, the method comprising: receiving a request to access a resource, in response to receiving the request to access the resource triggering to illuminate the object by coherent infrared light, and triggering to generate an image of the object while the object is being illuminated by the coherent infrared light, generating a plurality of partial images from the image, providing the plurality of partial images based on a location of the plurality of partial images within the image to a data-driven model for determining if the object is a living organism, wherein the data-driven model is trained with a plurality of historical partial images provided based on a location of the plurality of historical partial images to determine whether the one or more objects associated with the plurality of historical partial images are living organisms, allowing the object to access the resource based on determining that the object is a living organism.

[0016] In another aspect, it relates to a method for allowing an object to access a resource, the method comprising: receiving an image generated while the object is illuminated by coherent infrared light, generating a plurality of partial images based on the image, providing the plurality of partial images to a data-driven model based on a location of the plurality of partial images within the image for determining if the object is a living organism, wherein the data-driven model is trained based on a plurality of historical partial images to determine whether the one or more object associated with the plurality of historical partial images is a living organism, allowing the object to access a resource based on determining that the object is a living organism.

[0017] In another aspect, it relates to use of a plurality of partial images generated based on an image generated while the object is illuminated by coherent infrared light for allowing an object to access a resource by providing the plurality of partial images to a data-driven model based on a location of the plurality of partial images within the image for determining an indication whether the object is a living organism based on the plurality of partial images, wherein the data-driven model is trained based on historical partial images wherein the data-driven model is trained based on a plurality of historical partial images to determine whether the one or more object associated with the plurality of historical partial images is a living organism.

[0018] In another aspect, it relates to a non-transitory computer-readable storage medium including instructions that, when processed by a computer, configure the computer to perform any one of the methods as described herein.

[0019] In another aspect, it relates to a device and / or system for allowing an object to access a resource, the device and / or system comprising: a processor; and a memory storing instructions that, when executed by the processor, configure the device and / or system to perform any one of the methods as described herein.

[0020] In another aspect, it relates to use of data-driven model suitable for receiving a plurality of partial images based on a location of the plurality of partial images within an image generated, wherein the image is generated while the object is illuminated by coherent infrared light, and, wherein the partial images are generated based on the image to a data-driven model, and determining if the object is a living organism based on the plurality of partial images, wherein the data-driven model is trained based on a plurality of historical partial images to determine whether the one or more objects associated with the plurality of historical partial images are living organisms. In another aspect, it relates to a method for measuring a material property of an object, the method comprising: receiving an image of the object under illumination by coherent infrared light, generating a plurality of partial images based on the image, providing the plurality of partial images to a data-driven model based on a location of the plurality of partial images within the image for determining an material property measure, wherein the data-driven model is trained based on historical partial images provided to the data-driven model based on a location of the historical partial images and corresponding historical material property measures, providing the material property measure.

[0021] Embodiments

[0022] Any disclosure, embodiments and examples described herein relate to the methods, the systems, devices, and computer-readable storage media lined out above and below. Advantageously, the benefits provided by any of the embodiments and examples equally apply to all other embodiments and examples.

[0023] In the following, terminology as used herein and / or the technical field of the present disclosure will be outlined by ways of definitions and / or examples. Where examples are given, it is to be understood that the present disclosure is not limited to said examples.

[0024] The present disclosure provides means for an efficient, robust, stable and reliable method for authenticating an object. Commercial authentication systems can be easily spoofed with hyper realistic spoofing objects such as silicon masks. Blood perfusion can be used as an effective distinguishing feature between a spoofing mask and a real human. Blood perfusion may be detected by providing an image of the object generated while the object is illuminated by coherent infrared light to a data-driven model. The data-driven model may reliably differentiate between living organisms and spoofing objects. In particular, providing a plurality of partial images to a data-driven model based on a location of the plurality of partial images within the image for determining if the object is a living organism and / or an indication of a material property may allow for efficient and secure authentication. Partial images can be processed more easily and hence, enable less computational resource usage for implementing a secure authentication. Providing the partial images based on the location allows for processing of all image data associated with the image while increasing the computational efficiency in terms of hardware requirements and time to compute. For determining an indication of a material property, the context associated with the full image data and hence, the relation between the partial images, improves the accuracy of authentication. Thus, secure authentication based on low-cost and readily available hardware to be performed. This enables, secure authentication for mobile devices such as smartphones with limited space and battery capabilities. Further, providing the partial images based on the location of the partial images within the image allows for an even more reliable authentication as the context of the partial image within the image can be taken into account. Hence, spoofing objects can be reliably detected. Ultimately, the security of authentication, in particular image-based authentication can be improved. This allows for implementing authentication systems based on low-cost hardware for high-security use cases such as boarder control, access control, engine start control or the like.

[0025] These and other objects, which become apparent upon reading the following description, are solved by the subject matters of the independent claims. The dependent claims refer to embodiments of the disclosure.

[0026] In an embodiment, the methods may further comprise receiving a material property measure from the data- driven model. Providing the plurality of partial images based on a location of the plurality of partial images within the image to a data-driven model for determining if the object is a living organism may refer to providing the plurality of partial images based on a location of the plurality of partial images within the image to a data- driven model for determining a material property measure. Allowing the object to access a resource based on determining that the object is a living organism may refer to allowing the object to access the resource based on receiving a blood perfusion measure from the data-driven model indicating that the object may be a living organism. The blood perfusion measure may indicate whether the object may be a living organism. Hence, the blood perfusion measure may indicate that the object may be a living organism or that the object may be a spoofing object. Object may be a human and / or a user, eg a user of a device. The device may be configured for carrying out the steps of any one of the methods as described herein.

[0027] In an embodiment, coherent infrared light may be patterned coherent infrared light.

[0028] Coherent infrared light may comprise of near infrared light, midinfrared light and / or far-infrared light. Near infrared light may be in the range of 780 nm to 3000 nm excluding the value of 3000 nm as being comprised in the range. Mid infrared light may be in the range of 3 m to 15 pm excluding the value of 15 pm as being comprised in the range. Far infrared light may be in the range of 15 pm to 1000 pm. The infrared light may be associated with a wavelength between 700 nm and 800 nm and / or between 1000 nm and 1200 nm.

[0029] In an embodiment, the coherent infrared light may have a wavelength between 700 nm and 800 nm and / or 1100nm and 1200 nm. This is advantageous since the sun light may be less intensive within the respective range. Thus, using light within this range improves the performance of the secure authentication under ambient light conditions.

[0030] Patterned coherent infrared light may comprise a plurality of light beams. Illuminating the object by the patterned coherent infrared light may comprise projecting the plurality of light beams associated with the patterned coherent infrared light onto the object. Projecting the plurality of light beams associated with the patterned coherent infrared light onto the object may comprise projecting a plurality of light spots associated with the plurality of light beams onto the object. In an embodiment, partial image may refer to a part of the image. The part of the image may be smaller than the image. The partial image may comprise a part of the image smaller than the image. The plurality of partial images and / or sections are equal in shape and / or size. The data-driven model may be parametrized and / or trained to receive the plurality of partial images of equal size and / or shape. This allows for an efficient parametrization of the data-driven model and thus, saves resources for implementing a secure authentication even on mobile devices with limited battery and space capabilities.

[0031] In an embodiment, the image may show a plurality of spots related to the illumination by patterned coherent infrared light. The plurality of partial images images may show at least one spot of the plurality of spots per partial image. In particular, one partial image may be associated with one spot. Further, the plurality of partial images may show background related to the plurality of spots. In particular, the plurality of partial images may show background related to the at least one spot associated with at least one of the plurality of partial images. Hence, the plurality of partial images may be generated to associate, in particular show, at least one spot per partial image. Optionally, the plurality of partial images may be generated to associate, in particular show, background related to the at least one spot per partial image. In an embodiment, at least one partial image may be associated with less than 10, preferably less then 5, even more preferably less than two spots. By doing so, the contribution of each spot and its relation to the other, in particular surrounding spots, can be assessed in order to measure a blood perfusion and / or a material property of the object.

[0032] In an embodiment, a region of interest associated with the image may be selected according to a flood image showing the object. This may include identifying one or more parts of the image associated with the object by classifying the one or more part(s) of the image to be associated with the object. An indication of the region of interest may be received. The plurality of partial images may be generated based on the selected region of interest.

[0033] In an embodiment, data-driven model may refer to a model suitable for describing one or more non-linear relations between input data and output data. Input data may refer to data to be provided to the data-driven model and / or to data being received by the data-driven model. Output data may be data to be received from the data-driven model and / or to be provided by the data-driven model. Hence, the data-driven model may determine the output data based on transforming the input data via one or more non-linear relations. In this context, input data may be for example a plurality of partial images. Output data may be for example an indication whether the object associated with the image may be a living organism or a spoofing object.

[0034] The data-driven model may be trained based on a plurality of historical partial images to determine whether the object, preferably associated with the plurality of historical partial images, may be a living organism may refer to the data-driven model may be trained based on a first set comprising a plurality of first partial images and a corresponding indication whether the object associated with the first set may be a living organism and a second set comprising a plurality of second partial images and a corresponding indication whether the object associated with the second set may be a living organism.

[0035] In an embodiment, data-driven model may comprise one or more embedding layers. The embedding layers may be suitable for transforming the provided data such as the partial images into a machine-processable representation. The machine-processable representation may comprise one or more numerical values. The machine-processable representation may comprise one or more first-rank tensors and / or may be a second- rank tensor. In an embodiment, the machine-processable representations of the plurality of partial images may be compressed representions of the plurality of partial images. Any one of the methods may further comprise mapping the plurality of partial images to machine-processable representations of the plurality of partial images. Preferably, one machine-processable representation may be generated per partial image.

[0036] Further, the data-driven model may comprise one or more model blocks such as a transformer block or a graph neural network block. The one or more model blocks may be suitable for transforming the machine- processable representation into a context tensor. The context tensor may be a second-rank tensor. Further, the data-driven model may comprise one or more classification layers. The classification layers may be suitable for transforming the context tensor into an indication on whether the object may be a living organism.

[0037] In an embodiment, object may a living organism and / or a spoofing object. Spoofing object may be presented for authenticating a non-living object. Spoofing object may comprise one or more features associated with an authorized user.

[0038] In an embodiment, historical partial image may be a partial image. Embodiments applying to the partial image may apply to the historical partial image analogously. Historical partial image may be part of training data for training the data-driven model. Historical partial images may be associated with corresponding blood perfusion measures. Historical partial image may be associated with an indication if the object associated with the historical partial image may be a living organism. Hence, historical partial images may have already been classified according to whether the object associated with the historical partial image may be a living organism.

[0039] In an embodiment, providing the plurality of partial images to a data-driven model based on the location of the plurality of partial images within the image for determining if the object may be a living organism, wherein the data-driven model may be trained based on a plurality of historical partial images to determine whether the one or more object associated with the plurality of historical partial images may be a living organism may refer to providing the plurality of partial images to a data-driven model based on the location of the plurality of partial images for determining an indication whether the object may be a living organism based on the plurality of partial images, wherein the data-driven model may be trained based on historical partial images and corresponding historical indications whether a historical object associated with the historical partial images may be a living organism. The methods may further comprise receiving the indication whether the object may be a living organism from the data-driven model. The object may be allowed to access the resource based on receiving the indication that the object may be a living organism. Allowing the object to access a resource based on determining that the object may be a living organism may refer to allowing the object to access a resource based on receiving an indication that the object may be a living organism. During training the indication on whether the object may be a living organism may comprise one or more lables associated with the historical images. Further, the indication whether the object may be a living organism may be indicative of whether the object associated with the image may be a living organism. The indication whether the object may be a living organism may comprise a numerical value and / or boolean value. If the numerical values may be within a predefined range of numerical values the blood perfusion measure may indicate that the object associated with the image may be a living organism. Otherwise, the blood perfusion measure may indicate that the object associated with the image may be a spoofing object. The boolean value may indicate that the object associated with the image may be a spoofing object or a living organism.

[0040] In an embodiment, providing the plurality of partial images based on a location of the partial images to the data-driven model may refer to providing the plurality of partial images to the data-driven model, wherein providing the plurality of partial images may be indicative of a location of the plurality of partial images within the image. In an embodiment, providing the plurality of partial images based on a location of the partial images to the data-driven model comprises providing a data structure associated with the plurality of partial images and the location of the plurality of partial images within the image to the data-driven model, wherein the data-driven model may be parametrized and / or trained to being provided with data structures associated with the plurality of partial images and the location of the plurality of partial images within the image and / or providing a sequence of the plurality of partial images to the data-driven model based on the location of the plurality of partial images within the image, wherein the data-driven model may be parametrized and / or trained to to being provided with a sequences of the plurality of partial images. The sequence of the plurality of partial images may be indicative of a distance between two or more partial images of the plurality of partial images. This may equally apply to the historical partial images. The plurality of partial images may be provided in a sequence, wherein the sequence depends on a location of the plurality of partial images within the image. The sequence may relate to a sequence of locations of the plurality of partial images within the image. The sequence may be determined by the location of the plurality of partial images within the image. In an embodiment, the methods may further comprise mapping the plurality of the partial images to a data structure associated with the plurality of partial images and the location of the plurality of partial images within the image. The data structure associated with the plurality of partial images and the location of the plurality of partial images within the image may be obtained by mapping the plurality of partial images to a data structure associated with the plurality of partial images and the location of the plurality of partial images within the image. In an embodiment, allowing the object to access a resource may include allowing the object to perform at least one operation with a device, computing apparatus and / or system. Resource may be a device, a system, a computing apparatus, a function of a computing apparatus, a function of a device, a function of a system and / or an entity. Additionally and / or alternatively, allowing the object to access a resource may include allowing the object to access an entity. Entity may be physical entity and / or virtual entity. Virtual entity may be a database for example. Physical entity may be an area with restricted access. Area with restricted access may be one of the following: security areas, rooms, apartments, vehicles, parts of the before mentioned examples, or the like.

[0041] In an embodiment, the methods may further comprise receiving a flood image of the object generated while the object may be illuminated by infrared illumination. The flood image may show and / or may be indicative of a contour of the object. Further, the methods may comprise receiving a template image of an authorized user. The template image may show and / or may be indicative of a contour of the authorized user. Further, the methods may comprise providing the flood image and the template image to an authentication data-driven model for determining if the object may correspond to the authorized user. The authentication data-driven model may be trained based on a plurality of historical flood images and a plurality of historical template images to determine if the one or more objects associated with the historical flood images may correspond to one or more authorized users associated with the historical template images. For this purpose, the authentication data-driven model may comprise one or more embedding layers. The one or more embedding layers may be configured for reducing the dimensionality of the flood image. The one or more embedding layers may transform the flood image and the template image into a tensor, in particular a two-dimensional tensor or a one dimensional tensor. Alternatively, receiving the template image may comprise receiving a tensor associated with the template image. The tensor associated with the template image may be obtained by providing the template image to one or more embedding layers, eg of a second data-driven model. Additionally or alternatively, the authentication data-driven model may comprise one or more classification layers. The one or more classification layers may be configured for receiving the tensor associated with the flood image and the tensor associated with the template image and / or classifying the tensor associated with the flood image and the tensor associated with the template image according to whether the flood image and the template image may be associated with the same authorized user. Additionally or alternatively, the authentication data-driven model may comprise one or more mathematical relations for determining a distance between the tensor associated with the flood image and the tensor associated with the template image, in particular an euclidean distance and / or a cosine similarity.

[0042] In an embodiment, the methods may further comprise receiving a flood image of the object generated while the object may be illuminated by infrared illumination. The flood image may show a contour of the object, and, receiving a template image of an authorized user. The template image may show a contour of the authorized user. The methods may further comprise providing the flood image and the template image to a data-driven model for determining if the object may correspond to the authorized user. The data-driven model may be trained based on a plurality of historical flood images and a plurality of historical template images to determine whether the one or more objects associated with the historical flood images may correspond to one or more authorized users associated with the historical template images.

[0043] In an embodiment, providing the plurality of partial images based on a location of the plurality of partial images within the image to a data-driven model may comprise providing the plurality of partial images to the data-driven model in a sequence based on the location of the plurality of partial images within the image. The data-driven model may be further parametrized and / or trained to to being provided by the plurality of partial images in the sequence based on the location of the plurality of partial images within the image. By doing so, an even more reliable authentication is enabled as the context of the partial image within the image can be taken into account. Hence, spoofing objects can be reliably detected.

[0044] In an embodiment, generating the plurality of partial images based on the image comprises dividing the image into a plurality of sections. Further, generating the plurality of partial images based on the image comprises cropping the image according to the plurality of sections to generate the plurality of partial images. The plurality of partial images and / or sections may be equal in shape and / or size.

[0045] In an embodiment, the sequence based on the location of the plurality of partials images within the image may be a linear sequence starting with the partial image associated with the upper most left section, followed by other partial images being associated with the upper most sections from left to right and followed by further partial images associated with further upper most sections from left to right. The data-driven model may comprise a transformer-based architecture. The data-driven model may be parametrized and / or trained to apply self-attention, in particular multi-head self-attention to the plurality of partial images, in particular to the linear sequence and / or a machine processable representation of the linear sequence. The machine- processable of the linear sequence may be obtained by passing the linear sequence through one or more embedding layers. The linear sequence allows for already available components of data-driven models to be used. Hence, resources and time for implementing the secure authentication being highly reliable can be saved.

[0046] In an embodiment, providing the plurality of partial images based on a location of the plurality of partial images within the image to a data-driven model may comprise providing a graph data structure representing the plurality of partial images to the data-driven model. The graph data structure may comprise a plurality of nodes and a plurality of edges. One node of a graph data structure may represent one partial image of the plurality of partial images and wherein an edge of the graph data structure may represent a distance between at least two of the plurality of partial images. The data-driven model may be parametrized and / or trained to being provided with a graph data structure representing the plurality of partial images. The number of nodes may be equal to the number of partial images. The plurality of partial images may be represented by the nodes of the graph. The number of edges may be equal to the number of pairs of partial images. The nodes may comprise node vectors. The edges may comprise edge vectors. The graph data structure may be a linear graph data structure representing a sequence of partial images. The data structure may be a graph data structure associated with the plurality of partial images and the location of the plurality of partial images within the image, wherein the graph data structure may comprise a plurality of nodes and a plurality of edges, wherein one node of the graph data structure may be associated with one partial image of the plurality of partial images and wherein one edge of the graph data structure may be associated with a sequence of and / or a distance between two or more of the plurality of partial images and the data-driven model may be parametrized to being provided with data structures refers to the data-driven model may be parametrized to being provided with the graph data structures. By doing so, a more reliable and robust detection of spoofing object and thus, secure authentication is enabled. The graph data structure may be a directed graph data structure. The directed graph data structure may be associated with a sequence of the plurality of partial images within the images. A direction associated with the directed graph data structure may be indicative of the location of the plurality of partial images within the image

[0047] In an embodiment, one or more of the plurality of partial images and / or sections may comprise at least a part of one light spot generated by illuminating the object by patterned coherent infrared light. The one or more of the plurality of partial images and / or sections may further comprise at least a part of a surrounding of the light spot within a predefined distance from the centre of the light spot. By doing so, background suppression is achieved focusing the attention of the data-driven model onto determining if the object is a living organism or a spoofing object. Thus, secure authentication is enabled.

[0048] In an embodiment, the partial images may relate to a grid of the image and wherein the sequence may be defined by a predefined sequence path connecting partial images of the grid through the predefined path. One or more parts of the grid may be associated with two or more partial images. A first partial image in the sequence may be a predefined partial image and following partial images may be nearest neighbors and / or second nearest neighbors to the previous partial images. Additionally or alternatively, the following partial images may be adjacent partial images to the previous partial images. In particular, the adjacent partial image may be associated with the same part of the image as the previous partial image in the sequence.

[0049] In an embodiment, the methods may further comprise preprocessing of the image by detecting one or more features in the image and augmenting the image according to the detected feature. The detected feature may be a light spot. Augmenting the image according to the detected feature may comrise cropping the image according to the detected one or more features. Preferably, cropping the image according to the detected one or more features may include eliminating at least a part of the surrounding of the image feature within a predefined distance of the centre of the features. In an embodiment, the one or more image features may be spatial features of the object. Preferably, cropping the image according to the detected one or more image features may include eliminating at least a part of the detected one or more image features or elimitating at least a part of a surrounding of the one or more image features. By doing so, background suppression is achieved focusing the attention of the data-driven model onto determining if the object is a living organism or a spoofing object. Thus, secure authentication is enabled.

[0050] In an embodiment, the methods may further comprise preprocessing of the plurality of partial images by applying one or more image augmentation techniques to the plurality of partial images. Applying one or more image augmentation techniques to the plurality of partial images may result in a plurality of partial images with equal size and / or shape. In an embodiment, image augmentation techniques may comprise at least one of scaling, cutting, cropping, rotating, blurring, warping, shearing, resizing, folding, changing the contrast, changing the brightness, adding noise, multiply at least a part of the pixel values, drop out, adjusting colors, applying a convolution, embossing, sharpening, flipping, averaging pixel values or the like. Preferably, applying one or more than one image augmentation techniques may comprise cropping the image according to one or more image features associated with the image and / or shearing the image. Shearing the image may comprise changing a distance between one or more first image features and one or more second image features associated with the image and optionally changing the distance between the one or more first image features and one or more third image features associated with the image.

[0051] By doing so, background suppression is achieved focussing the attention of the data-driven model onto determining if the object is a living organism or a spoofing object. Thus, secure authentication is enabled.

[0052] In an embodiment, material property measure, in particular the blood perfusion measure, may be suitable for determining if the object associated with the image may be a living organism and / or may indicate whether the object associated with the image may be a living organism. The material property measure, in particular the blood perfusion measure, may comprise one or more numerical value and / or a boolean value. The numerical value may indicate a confidence score associated with determining if the object associated with the image may be a living organism. The boolean value may indicate whether the object associated with the image may be a living organism. If the one or more numerical values may be within a predefined range of numerical values the material property measure, in particular the blood perfusion measure, may indicate that the object associated with the image may be a living organism. Otherwise, the one or more numerical values may indicate that the object associated with the image may be a spoofing object.

[0053] In an embodiment, determining if the object is a living organism may be based on the plurality of partial images.

[0054] In an embodiment, the methods may be computer-implemented methods.

[0055] In an embodiment, training the data-driven model based on may refer to training the data-driven model with. In an embodiment, generating a plurality of partial images based on the image may refer to generating the plurality of partial images from the image. In an embodiment, the material property measure may be related to, in particular may include, a type of material and / or a blood perfusion measure. The type of material may include biological or non-biological material, translucent or non-translucent materials, metal or non-metal, skin or non-skin, latex or non- latex, silicon or non- silicon, fabric or non- fabric, reflective or non-reflective, specular reflective or non-specular reflective, foam or non-foam, hair or nonhair, roughness groups or the like. The type of material may include further specifications of the material type such as in class biological the type may be human skin or in class non-biological the type may be plastics, glass, metal. Further sub-types may be assigned to each type. The indication of the material property may be a material property measure. The material property may indicate the type of the material and / or if the object associated with the material property may be a living organism.

[0056] Brief Description of the Several Views of the Drawings

[0057] In the following, the present disclosure is further described with reference to the enclosed figures. The same reference numbers in the drawings and this disclosure are intended to refer to the same or like elements, components, and / or parts.

[0058] FIG. 1A illustrates an embodiment of a device for allowing an object to access a resource 102. FIG. 1 B illustrates an embodiment of a system for allowing an object to access a resource.

[0059] FIG. 2 illustrates an embodiment of a method for allowing an object to access a resource.

[0060] FIG. 3 illustrates an embodiment of generating a plurality of partial images 324 based on the image 322.

[0061] FIG. 4 illustrates an embodiment of providing the plurality of partial images 432 to a data-driven model for determining if the object associated with the image 430.

[0062] FIG. 5 illustrates an embodiment of a method for allowing an object to access a resource.

[0063] FIG. 6 illustrates an embodiment of generating a plurality of partial images 324 based on the image 322.

[0064] FIG. 7 illustrates an embodiment of a providing the plurality of partial images 324 to a data-driven model for determining if the object associated with the image 322.

[0065] FIG. 8 illustrates an embodiment of the plurality of partial images.

[0066] Detailed Description

[0067] The following embodiments are mere examples for implementing the method, the system or application device disclosed herein and shall not be considered limiting.

[0068] FIG. 1A illustrates an embodiment of a device for allowing an object to access a resource 102.

[0069] The device 102 may comprise an illumination source 108 a camera 110 comprising a sensor 112, a processor 104 and / or a memory 114. The illumination source 108 may emit coherent light towards the object 106, preferably coherent infrared light. Infrared light may be unrecognizable to the object 106. The sensor 112 of the camera 110 may be sensitive towards the light emitted by the illumination source 108. Hence, the sensor 112 may be suitable for generating an image of the object 106 while the user may be illuminated by the light emitted from the illumination source 108. The processor 104 may receive the image of the object 106. The processor 104 may process the image of the object 106. By doing so, the processor 104 may determine if the object 106 corresponds to a living organism. This may include determining if the image of the object 106 corresponds to a living organism. For this purpose, the processor may execute instructions stored in a memory 114. Executing the instructions by the processor may result in performing any one of the methods as described in FIG. 2, FIG. 3 and FIG. 6.

[0070] Alternatively, the illumination source 108 and / or the camera 110 may be comprised in a second device and / or a third device. In this example, the processor 104 may be communicatively coupled to the second and / or third device for triggering the illumination source 108 to emit the light and triggering the sensor 112 to generate an image of the object 106 while the object is illuminated by the light.

[0071] The processor 104 may receive a request to access resource such as unlocking the device. For this purpose, the device may comprise a user interface. The object 106 may request to access a resource by the user interface. For example, the device may be a phone. The object 106 may desire to control the phone. For this purpose, the object 106 may need to authenticate. The object 106 may request to access the resource and / or the authentication by using the touch display of the device. This may trigger to provide a signal to the processor 104 of the device. Based on receiving the signal, the processor 104 triggers the illumination source 108 to emit the light.

[0072] FIG. 1 B illustrates an embodiment of a system for allowing an object to access a resource.

[0073] The system may comprise a first device 130 and a second device 128. In an embodiment, a first device 130 may comprise the processor 104 and the memory 114. The first device 130 may be communicatively connected to a second device 128 comprising the illumination source 108 and the camera 122 comprising the sensor 112. For example, the first device 130 may be connected to the second device 128 by means of a cloud service. In particular, the processor 104 and / or the memory 114 may be part of a cloud service. The second device 128 may be configured for providing the image generated by the sensor 112 to the first device 130. Receiving a request to access a resource may refer to receiving a signal triggering the illumination source 108 to emit light at the illumination source 108. In response to receiving the signal, the illumination source 108 may be triggered to illuminate the object 106 by light. Further, the camera 110 may receive a signal triggering to generate an image of the object 106. Receiving a request to access a resource may further comprise receiving the signal at the camera 110.

[0074] The processor 104 may determine if the object 106 associated with the image is a living organism by generating a plurality of partial images based on the image providing the plurality of partial images to a data- driven model. Based on the indication whether the object 106 is a living organism, the processor 104 may provide a signal for providing access to the object 106 Providing the signal for providing access to the object 106 may be referred to as allowing an object 106 to access a resource. This may be as described in the context of FIG. 2, FIG. 3 and / or FIG. 6.

[0075] FIG. 2 illustrates an embodiment of a method for allowing an object to access a resource.

[0076] A request to access a resource may be received 202. For example, a user may desire to unlock a device such as a smartphone and / or may perform an action on the device that may require authentication, eg performing a payment. For this purpose, the user may request to receive access to a resource, in particular of the device. Additionally or alternatively, the user may request to receive access to an area and / or a good. The request may be triggered by a user entering an area and / or interacting with a user interface. The request may be received at the device, at a security point associated with the area and / or a control unit associated with the storage of the good.

[0077] Requesting to receive access to a resource may trigger to illuminate the object by coherent infrared light. The object may be faced with the device, the area and / or the good. The object may be the user. Receiving the request may include receiving a signal indicative of a request to access a resource. Receiving the signal indicative of the request may result in generating and / or providing a signal indicative of a trigger to illuminate the coherent infrared light. The object may be located within the proximity of the illumination source for illuminating the coherent infrared light. Preferably, the illumination source may face the object. Hence, the object may be illuminated by coherent infrared light. The coherent infrared light may interact with the material associated with the object. Depending on the material and the resulting interaction different interference patterns may be obtained by illuminating the object by coherent infrared light. The image may show the interaction of the coherent infrared light with the material of the object. In particular, the coherent infrared light may be infrared coherent infrared light. Infrared light may be invisible to humans. Hence, this enables authenticating humans independent of noticing of the authentication process by humans. Thus, secure authentication is enabled. Further, infrared light may penetrate into deeper layers of the skin such as the dermis. The dermis may comprise blood vessels. The blood vessel may be filled by blood. The amount of blood within a part of the blood vessel may depend on the cardiac cycle. Hence, the movement of blood within an area of skin of a living organism may change over time, preferably periodically. The coherent infrared light may interact with the blood vessel and the blood. Depending on the interaction of the coherent infrared light, at least a part of the coherent infrared light may leave the skin. This leaving light may be collected in an image. The faster the blood may move the more the light may be deflected. Hence, the presence of moving blood, the so-called blood perfusion, may result in blurring of the coherent infrared light in the image as generated.

[0078] Further, the coherent infrared light may be patterned coherent infrared light. This enables to illuminate a smaller area of the object while allowing for secure authentication of authorized users and detecting spoofing objects. Patterned coherent infrared light may comprise a plurality of light beams. Illuminating the patterned coherent infrared light may result in projecting a the plurality of light beams onto the object. Hence, a plurality of light spots may be projected onto the object. In an embodiment, the patterned coherent infrared light may comprise of less than 4000 light beams. In an embodiment, projecting the patterned coherent infrared light onto an object may result in projecting less than 4000 spots. In an embodiment, the patterned coherent infrared light may comprise of less than 3000 light beams, preferably less than 2000 light beams, most preferably less than 1000 light beams. In an embodiment, projecting the patterned coherent infrared light onto an object may result in projecting less than 3000 spots, preferably less than 2000 spots, most preferably less than 1000 spots.

[0079] Hence, the image may be generated while the object is illuminated by coherent infrared light 206. The signal indicative of a trigger to illuminate the coherent infrared light may trigger to generate the image of the object while the object may be illuminated by coherent infrared light.

[0080] The image may be preprocessed by augmenting the image 208. Augmenting the image may comprise applying one or more image augmentation techniques to the image. For example, preprocessing the image may comprise detecting one or more image features in the image 218. Where the coherent infrared light may be patterned coherent infrared light, the image feature may comprise a light spot. Hence detecting the one or more image features in the image may refer to detecting one or more light spots. Further, the image may be augmented according to the detected image features, preferably light spots. Augmenting the image according to the detected image features may refer to cropping the image according to the detected image features. The image may comprise the images features and background. The background may be independent of the blood perfusion of the object. Hence, it may be advantageous to eliminate the background. Augmenting the image according to the detected images features may include eliminating at least a part of the background. Cropping the image according to the image features may include eliminating at least a part of the background based on the detected image features. For example, background within a predefined distance of the centre of one or more image features may be eliminated by cropping the image. An example for the image after preprocessing may be shown in FIG. 3 and FIG. 6.

[0081] A plurality of partial images may be generated based on the image 210 as described within the context of FIG. 3 and FIG. 6. The plurality of partial images may be preprocessed by augmenting the plurality of partial images 212. This may further increase the reliability of authenticating the authorized user.

[0082] The correlation between the image and the material property measure, in particular the blood perfusion measure, of an object may be represented by a data-driven model. Hence, the data-driven model may be trained to obtain the correlation between the image and the material property measure, in particular the blood perfusion measure. The data-driven model may be trained based on a plurality of historical partial images to determine whether the one or more object associated with the plurality of historical partial images is a living organism. For this purpose, a data-driven model may be initialized and one or more sets of historical partial images may be provided to the data-driven model. The historical partial images may be processed by the data-driven model as described within the context of FIG. 3 and FIG. 6. Processing of partial images may be analogous to processing of historical partial images. By processing of the historical images, the data-driven model may result in an indication whether the object associated with the historical image and / or the historical partial images may be a living organism as described within the context of FIG. 3 and FIG. 6. The historical partial images may be generated based on the historical image. The one or more sets of historical partial images may be generated based on one or more historical image. Hence, a set of historical partial images may be generated based on one historical image. The set of historical partial images may be associated with a historical indication whether the object associated with the historical partial images may be a living organism. For example, the sets of historical partial images may be labelled according to whether the object associated with the historical partial images may be a living organism or a spoofing object. The plurality of partial images may be provided to the data-driven model 214.

[0083] Based on determining that the object may be a living organism by providing the plurality of partial images to the data-driven model, the object may be allowed to access the resource 216. This may include unlocking of the device, eg the smartphone, allowing the object to perform the requested action on and / or with the device, providing the object access to the area and / or providing the object access to the good and / or providing a good to the object.

[0084] FIG. 3 illustrates an embodiment of generating a plurality of partial images 324 based on the image 322.

[0085] The image 322 may be generated by the device and / or system as described in the context of FIG. 1A and / or FIG. 1 B. The image 322 may be generated as described in the context of FIG. 2. Further, the image 322 may be a result of the preprocessing as described in the context of FIG. 2. A plurality of partial images 324 may be generated based one the preprocessed image 322. Generating the plurality of partial images may include dividing the image 322 into a plurality of sections and generating a plurality of partial images 324 associated with the plurality of sections. The sections may be associated with a geometrical shape such as a square, rectangle, triangle or any other geometrical shape. The sections may be of equal shape and / or size and / or may differ in shape and / or size. In FIG. 3 the image 322 may be divided into 4 equally shaped and sized sections. Based on these sections, the partial images 324 may be generated. In an embodiment, two of the plurality of the partials images may comprise the same area associate with the image 322. For example, the sections may be define non-overlapping sections associated with the image 322. The partial images 324 may be generated based on the sections by including a part of an another section.

[0086] The image 322 may show a plurality of image features. Image feature may be a light spot. The partial images 324 may show one or more images features. Preferably, the partial images 324 may show one image feature and the surrounding of the image feature. The surrounding of the image feature may be an area around the image feature within a predefined distance to the centre of the image feature. Hence, the number of image features in the image 322 may be equal to the number of generated partial images 324.

[0087] The partial images 324 may be provided to a data-driven model based on a location of the partial images 324 within the image 322. In FIG. 3, the partial images 324 may be provided by flattening the generated partial images. Flattening the partial images may refer to generating a linear sequence of partial images 336 based on the location of the partial images 324 within the image 322. Hence, providing the partial images 324 to the data-driven model model based on the location of the partial images within the image 322 may comprise providing the sequence of partial images 336 to the data-driven model, in particular where the data-driven model may comprise an attention mechanism such as a transformer architecture. For example, the partial images 324 may be provided in a sequence starting with the most upper left partial image 326 followed by other most upper partial images 332 by increasing the distance between the following partial images from the previous partial images 326. These images may be followed by the second most upper left partial image 328 and further followed by the second most upper partial images 330 by increasing the distance between the following partial images from the previous partial images 328.

[0088] FIG. 4 illustrates an embodiment of providing the plurality of partial images 432 to a data-driven model for determining if the object associated with the image 430.

[0089] The sequence of partial images 444 may be provided to the data-driven model, in particular to the partial image embedding 412.

[0090] The data-driven model may comprise a partial image embedding 412, a positional embedding 414, a transformer block 416, a fully connected layer 418, a softmax function or sigmoid function 420 or a combination thereof. The data-driven model maybe a vision transformer. Providing the plurality of partial images 432 may comprise embedding the partial images 432 by partial image embedding 412 and positional embedding 414. Embedding the partial images 432 via partial image embedding 412 may result in a machine- processable representation of the partial images 432 and the relation between the partial images 432. The machine-processable representation of the partial images 432 may be a tensor, in particular a second-rank tensor.

[0091] Applying positional embedding 414 may refer to adding a positional factor to the machine-processable representation obtained via partial image embedding 412. Preferably, the input data may specify a relation between the partial images, in particular a sequence of partial images. The positional factor Ppos may be indicative of the position of the partial image within the image. For example, the positional factor Ppos may be obtained based on the following equation:

[0092] . ( pos \

[0093] Ppos 2z = sin - —

[0094] \ J \ 1000( / where pos may refer to the position of the partial image within the image, / may refer to the dimension associated with the partial image embedding 412 and d may refer to the dimension of the model, eg transformer decoder, transformer encoder or transformer encoder-decoder. This may be referred to as absolute positional embeddings. Alternatively, the positional embedding 414 may be based on rotary positional embeddings (RoPE). Positional embedding 414 is beneficial since it enables the processing of sequential data without requiring further dimensions indicating the position of each partial image. Fol lowingly , the positional embedding 414 reduces the computational resources needed for embedding the input data.

[0095] The machine-processable representation of the partial images 432 may be provided to a transformer block 416 such as an encoder block 552 and / or decoder block 554 as described in the context of FIG. 5. The transformer block 416 may be suitable for applying a normalization, multi-head self-attention, residual connection, concatenation, a multi-layer perceptron or a combination thereof to the machine-processable representation of the partial images 432. This may result in a context tensor. The transformer block may transform the machine-processable representation of the partial images 432 into a context tensor. The context tensor may be a machine-processable representation of the partial images 432 and the relation between the partial images 432. The context tensor may be a tensor, in particular a second-rank tensor. The context tensor may be provided to a fully connected layer 418. The fully connected layer 418 may transform the context tensor into a tensor indicating whether the image 430 and / or the partial images 432 may be associated with a living organism. Then, the tensor may be provided to a softmax function or sigmoid function 420 to determine an indication whether the object associated with the image 430 and / or the partial images 432 may be a living organism. The indication may be provided by the data-driven model. The indication may comprise one or more numerical values. The numerical values may indicate the confidence score associated with determining if the object associated with the image 430 may be a living organism. The indication may comprise a vector. The vector may comprise a plurality of numerical values. For determining if the object may be a living organism, the data-driven model may classify the partial images 432. Classifying the partial images 432 may refer to receiving the plurality of partial images 432 together and providing an indication whether the object may be a living organism based on receiving the plurality of partial images 432. Hence, the fully connected layer 418 and the softmax function or sigmoid function 420 may be suitable for classifying the tensor generated by the transformer block 416, the partial image embedding 412 and the positional embedding 414. For this purpose, the data-driven model may be trained based on historical partial images to embed the historical partial images, transform the historical embedded partial images into historical context tensors. Based on the historical context tensors, the classification part of the data-driven model comprising the softmax function or sigmoid function 420 and the fully connected layer 418 may classify the historical partial images whether the object associated with the historical partial images may be a living organism. In the example, the image 430 may be show a living organism under illumination by coherent infrared light. Hence, providing the plurality of partial images 432 to the data-driven model may result in classifying the plurality of partial images 432 as "skin". Hence, the data-driven model may have determined the object to be a living organism.

[0096] FIG. 5 illustrates an embodiment of a transformer block 416.

[0097] The transformer block may comprise one or more encoder blocks 552 and / or one or more decoder blocks 554. The transformer block may receive the embedded sequence of partial images. Hence, the embedded input data may be the embedded sequence of partial images. The embedded sequence of partial images may be obtained by applying positional embedding 414 and / or partial image embedding 412 as described in the context of FIG. 4.

[0098] The embedded input data may be processed by the encoder block. The embedded input data may be provided to the layer normalization 424 by a residual connection. Multi-head self-attention 422 may be applied to the embedded input data. Multi-head self-attention 422 may comprise the two components multi-head and selfattention. Self-attention may be understood as being a filter applied to the embedded input data. By applying the filter to the embedded input data, the elements associated with the embedded input data contributing to the to be generated output data may be identified for generating the output data. The elements associated with the embedded input data may refer to the plurality of partial images associated with the sequence.

[0099] Hence, the filter may represent the degree of contributing to the to be generated output data by the elements associated with the embedded input data. Applying the filter may be referred to as weighting the elements associated with the embedded input data. This is advantageous specifically regarding long sequences of elements. The filter may be learned and improved during the training by learning to identify the contribution of elements associated with the embedded input data. Self-attention may refer to attention generated based on the input data. Hence, the filter may be determined based on the input data, preferably the embedded input data. The embedded input data may serve as query Q, key K and value V with respect to the self-attention operation. The self-attention may refer to attention based on the received input data. Hence, the filter may be calculated based on the following formula by inserting the respective tensors based on the embedded input data: where dk corresponds to the dimension of the key. For improving the efficiency of the encoder block 552 further, the multiple heads are used to apply the filter resulting in the multi-head self-attention 522. Multi-head self-attention 522 may comprise applying the filter to two or more parts of the embedded input data. Hence, the tensor may be split into two or more parts and the filter may be applied to the two or more parts separately by two or more heads according to the following equation:head i =Attention (Q W{Q, KWiK, VWiV) with parameter matrices where i may refer to the number of heads, dv, dx and dQ may refer to the dimensions of the value, key and query.

[0100] The result of the two or more head may be concatenated according to the following equation: MultiHead(Q, K, V) = Concat(head 1, . . . , headh) W° where jo eg xd and h may refer to the number of heads.

[0101] The embedded input data may be transformed via the multi-head self-attention 522 into a context tensor. The context tensor may represent the sequence of elements and the relation between two or more elements of the input data. The context tensor may be a second rank tensor and / or may comprise one or more first rank tensor(s). After the multi-head self-attention 522 layer normalization 524 may be applied based on the context tensor and / or the embedded input data from the residual connection. Applying layer normalization 524 may refer to normalizing the context tensor. Normalizing the context tensor may lower the values of the entries of the context tensor. This reduces the computational cost associated with processing the context tensor. Further, it improves the training by contributing the loss to converge and preventing instabilities.

[0102] Layer normalization 524 may be followed by passing the context tensor to a feed-forward layer 526 again followed by layer normalization 528 based on the residual connection to the context tensor and / or the output of the feed-forward layer 526. The feed-forward layer 526 may be a feed-forward neural network. The feedforward neural network may comprise of a plurality of fully connected neurons. Passing the context tensor through the feed-forward neural network may result in transforming the context tensor linearly. Additionally or alternatively, the neural network may comprise one or more activation functions such as a rectified linear unit (ReLU). Hence, the neural network may be configured for performing one or more non-linear operations to the context tensor and / or transforming the context tensor non-linearly. After the context tensor has been transformed and / or normalized by the feed-forward layer 526 and the layer normalization 528, the context tensor may be provided to one or more further encoder blocks 552. Having passed the context tensor through the feed-forward layer 526 may adapt the context tensor for the processing by a further attention layer of the one or more further encoder blocks 552 for applying a self-attention filter, preferably multi-head self-attention 522. The context vector after being transformed by the layer normalization 528 and the feed-forward layer 526 may be referred to as hidden state. This hidden state also referred to as output of the encoder block 552 may be provided to one or more fully connected layer 418 as described within the context of FIG. 4.

[0103] Additionally or alternatively, the embedded input data may be provided to a decoder block 554.

[0104] The decoder block 554 may comprise the layer normalizations 532, the masked multi-head self-attention 530, the feed-forward layers 534and / or the layer normalization 536. The embedded input data resulting from passing the input data through the embedding layers as described within the context of FIG. 4. The embedded input data may be provided to the layer normalization 532 via a residual connection. Further, masked multihead self-attention 530 may be applied to the embedded input data. Masked multi-head self-attention 530 corresponds to the multi-head self-attention 522 as described within the context of the encoder block 552 with additionally masking a part of the embedded input data associated with elements later in the sequence than the element to be generated. Additionally or alternatively, the part of the input data associated with elements later in the sequence than the element to be generated may not be received and / or transformed into the embedded input data. Thus, the transformer decoder may be suitable for generating a subsequent element to a sequence, whereas the transformer encoder may be suitable for generating a missing element in within one sequence and / or between two or more sequences. Therefore, the transformer encoder may be configured for classification tasks. The transformer decoder may be configured for element generation.

[0105] Similar to the encoder block 552, a context tensor may be generated by applying the masked multi-head selfattention 530 and the layer normalization 532. The context tensor may be provided to the layer normalization 532 via a residual connection. Further, the feed-forward layer 534 and the layer normalization 536 may be analogous to the feed-forward layer 526 and the layer normalization 528. The context tensor may be provided to one or more further decoder blocks 554.

[0106] FIG. 6 illustrates an embodiment of generating a plurality of partial images 324 based on the image 322.

[0107] The image 640 may be generated by the device and / or system as described in the context of FIG. 1A and / or FIG. 1 B. The image 640 may be generated as described in the context of FIG. 2. Further, the image 322 may be a result of the preprocessing as described in the context of FIG. 2. A plurality of partial images 636 may be generated based one the preprocessed image 640 as described within the context of FIG. 3.

[0108] In an embodiment, the data-driven model may comprise a graph neural network and / or the partial images 636 may be represented as a graph. A graph data structure may be a representation of data comprising nodes and edges. The partial images 636 may be represented as and / or associated with nodes and the relations between the partial images 636 may be represented as and / or associated wit edges. The relation between the partial images 636 may refer to a distance between the two or more partial images 636. For example, distant partial images 636 may be loosely related while close partial images 636 may be closely related.

[0109] FIG. 7 illustrates an embodiment of a providing the plurality of partial images 324 to a data-driven model for determining if the object associated with the image 322.

[0110] In an embodiment, the data-driven model may comprise a graph neural network and / or the partial images 736 may be represented as a graph data structure. Hence, the data-driven model may receive the graph data- driven model structure at one or more embedding layers. The one or more embedding layers may be suitable for embedding the one or more nodes associated with the graph data structure and / or embedding the one or more edges associated with the graph data structure and / or embedding the graph data structure. Hence, the node embedding, edge embedding, graph embedding 614 may be applied by passing the graph representation of partial images 634 through the one or more embedding layers. The embedding layers may be configured for performing node embedding, edge embedding, graph embedding 614.

[0111] The data-driven model may comprise one or more embedding layers for applying node embedding, edge embedding and / or graph embedding 714. Node embedding, edge embedding and / or graph embedding 714 may transform the graph representation of partial images 734 into one or more node vectors, one or more edge vectors and one or more graph vectors. The number of node vectors may be equal to the number of nodes, hence partial images 736. The number of edge vectors may be equal to the number of edges, hence the number of pairs of partial images 736. The number of graph vectors may be equal to the number of graphs, hence the number of images 740. The node vectors, the edge vectors and the graph vectors may be transformed separately. Hence, the data-driven model may comprise one or more functions for transforming the node vectors, one or more functions for transforming the edge vectors and one or more functions for transforming the graph vectors. For this purpose, the data-driven model may comprise one or more graph neural network layers 716. A graph neural network layer may comprise one or more functions for transforming the node vectors, one or more functions for transforming the edge vectors and one or more functions for transforming the graph vectors.

[0112] The graph neural network layer may receive one or more node vectors 744, and transform the one or more node vectors 744 into one or more updated node vectors 750 by applying one or more node functions 756. The one or more node functions 756 may be determined by one or more parameters. Training the graph neural network may comprise updating the one or more parameters associated with the one or more node functions 756.

[0113] Further, the graph neural network layer may receive one or more edge vectors 746 and transform the one or more edge vectors 746 into one or more updated edge vectors 752 by applying one or more edge functions 758. The one or more edge functions 758 may be determined by one or more parameters. Training the graph neural network may comprise updating the one or more parameters associated with the one or more edge functions 758.

[0114] Further, the graph neural network layer may receive one or more graph vectors 748 and transform the one or more graph vectors 748 into one or more updated graph vectors 754 by applying one or more graph functions 760. The one or more graph functions 760 may be determined by one or more parameters. Training the graph neural network may comprise updating the one or more parameters associated with the one or more graph functions 760.

[0115] By transforming the graph representation of partial images 734 the parameters of the graph specifying the nodes, edges and the overall graph may be updated. Updating the parameters of the graph may result in a second graph. The second graph may be different to the graph representation of partial images 734 provided to the data-driven model.

[0116] The second graph may be classified according to whether the images 740 may show a living organism. Being a living organism or a spoofing object may be a property of the image 740, hence a property of the graph. Thus, the second graph may be provided to one or more classification layers for classifying the graph. The images 740 may be associated with an object being a living organism. Hence, the data-driven model may provide an indication that the object may be a living organism. This may be seen in FIG. 7 as "skin”.

[0117] FIG. 8 illustrates an embodiment of the plurality of partial images.

[0118] The present disclosure has been described in conjunction with preferred embodiments and examples as well. However, other variations can be understood and effected by those persons skilled in the art and practicing the claimed subject-matter, from the studies of the drawings, this disclosure and the claims. Notably, in particular, the any steps presented can be performed in any order, i.e. the present disclosure is not limited to a specific order of these steps. Moreover, it is also not required that the different steps are performed at a certain place or at one node of a distributed system, i.e. each of the steps may be performed at different nodes using different equipment / data processing.

[0119] As used herein ..determining" also includes ..initiating or causing to determine", "generating" also includes ..initiating and / or causing to generate" and "providing” also includes "initiating or causing to determine, generate, select, send and / or receive”. "Initiating or causing to perform an action” includes any processing signal that triggers a computing node or device to perform the respective action.

[0120] In the claims as well as in the description the word "comprising” or "including” or similar wording does not exclude other elements or steps and shall not be construed limiting to the elements or steps lined out. The indefinite article "a” or "an” does not exclude a plurality. A single element or other unit may fulfill the functions of several entities or items recited in the claims. The mere fact that certain measures are recited in the mutual different dependent claims does not indicate that a combination of these measures cannot be used in an advantageous implementation or further elements may be included.

[0121] Providing in the scope of this disclosure may include any interface configured to provide data. This may include an application programming interface, a human-machine interface such as a display and / or a software module interface. Providing may include communication of data or submission of data to the interface, in particular display to a user or use of the data by the receiving entity. Any disclosure and embodiments described herein relate to the methods, the systems, devices, the computer program element lined out above and vice versa. Advantageously, the benefits provided by any of the embodiments and examples equally apply to all other embodiments and examples and vice versa.

Claims

ClaimsWhat is claimed is:1 . A method for measuring a material property of an object, the method comprising: receiving an image of the object under illumination by coherent infrared light, generating a plurality of partial images based on the image, providing the plurality of partial images to a data-driven model based on a location of the plurality of partial images within the image for determining an material property measure, wherein the data- driven model is trained based on historical partial images provided to the data-driven model based on a location of the historical partial images and corresponding historical material property measures, providing the material property measure.

2. A method for allowing an object to access a resource, the method comprising: receiving a request to access a resource, in response to receiving the request to access the resource triggering to illuminate the object by coherent infrared light, and triggering to generate an image of the object while the object is being illuminated by the coherent infrared light, generating a plurality of partial images based on the image, providing the plurality of partial images based on a location of the plurality of partial images within the image to a data-driven model for determining if the object is a living organism, wherein the data- driven model is trained based on a plurality of historical partial images provided based on a location of the plurality of historical partial images to determine whether the one or more objects associated with the plurality of historical partial images are living organisms, allowing the object to access the resource based on determining that the object is a living organism.

3. The method of claim 2, further comprising receiving a flood image generated while the object is illuminated by infrared illumination and a template image generated while an authorized user is illuminated by infrared illumination, and providing the flood image and the template image to an authentication data-driven model for determining if the object corresponds to the authorized user, wherein the authentication data-driven model is trained based on a plurality of historical flood images and a plurality of historical template images to determine whether the one or more objects associated with the historical flood images correspond to one or more authorized users associated with the historical template images.

4. The method of any one of claims 1 to 3, wherein providing the plurality of partial images based on a location of the partial images to the data-driven model comprises providing a sequence of the plurality of partial images to the data-driven model based on the location of the plurality of partial images within the image, wherein the data-driven model is parametrized to receive a sequence of the plurality of partial images as input.

5. The method of claim 4, wherein the sequence of the plurality of partial images comprises the plurality of partial images in an order according to the location of the plurality of partial images within the image.

6. The method of any one of claims 1 to 5, wherein providing the plurality of partial images based on a location of the partial images to the data-driven model comprises providing a data structure associated with the plurality of partial images and the location of the plurality of partial images within the image to the data-driven model, wherein the data-driven model is parametrized to being provided with data structures associated with the plurality of partial images and the location of the plurality of partial images within the image.

7. The method of claim 6, wherein the data structure is a graph data structure associated with the plurality of partial images and the location of the plurality of partial images within the image, wherein the graph data structure comprises a plurality of nodes and a plurality of edges, wherein one node of the graph data structure is associated with one partial image of the plurality of partial images and wherein one edge of the graph data structure is associated with a sequence of and / or a distance between two or more of the plurality of partial images and the data-driven model is parametrized to being provided with data structures refers to the data-driven model is parametrized to being provided with the graph data structures.

8. The method of claim 7, wherein the graph data structure is a directed graph data structure, wherein a direction associated with the directed graph data structure is indicative of the location of the plurality of partial images within the image.

9. The method of any one of claims 4-5 or 7-8, wherein the partial images relate to a grid of the image and wherein the sequence is defined by a predefined sequence path connecting partial images of the grid through the predefined path.

10. The method of any one of claims 1 to 8, wherein generating the plurality of partial images based on the image comprises dividing the image into a plurality of sections and / or cropping the image according to the plurality of sections to generate the plurality of partial images.

11. The method of any one of claims 1 to 10, wherein the coherent infrared light is patterned coherent infrared light and wherein the patterned coherent infrared light comprise of less than 4000 light beams and / or wherein projecting the patterned coherent infrared light onto an object results in projecting less than 4000 spots onto the object.

12. The method of claim 11 , wherein one or more of the plurality of partial images and / or sections comprise at least a part of one light spot generated by illuminating the object by the patterned coherent infrared light.

13. Use of a plurality of partial images generated based on an image generated while the object is illuminated by coherent infrared light in a method according to any one of claims 1 to 12.

14. Use of a data-driven model configured for receiving a plurality of partial images based on a location of the plurality of partial images within an image in a method according to any one of claims 1 to 12.

15. A device and / or system for allowing an object to access a resource, the device and / or system comprising: a processor; and a memory storing instructions that, when executed by the processor, configure the device and / or system to perform the method according to any one of claims 1 to 12.

Images