Method and system for automatic object recognition and authentication

Abstract

The present invention relates to the field of tracking and anti-counterfeiting protection of objects, such as products, and in particular to automatic identification and authentication of such objects. In particular, the present invention relates to a system and method for receiving object data representing one or more distinguishing characteristics of a physical object or a group of physical objects. The method comprises: processing the object data by means of an object recognition process based on machine learning to obtain distinguishing data representing one or more collision-resistant virtual representations of the physical object or the group of physical objects; comparing the distinguishing data and at least one of the original hash values ​​derived therefrom by applying a predetermined cryptographic hash function thereto with corresponding reference data stored in one or more data repositories with restricted access; and, if said comparison with the reference data results in a match, outputting identification data comprising a digital signature of said hash value. Potential applications of the present invention include in particular object authenticity checks for tracking and tracing and anti-counterfeiting purposes, such as, but not limited to, the fields of automotive / aerospace spare parts, 3D printed objects, pharmaceutical products or other health-related products such as healthcare products, laboratory disposables and clinical test disposables.

Description

Technical Field

[0001] The present invention relates to the field of tracking and anti-counterfeiting protection of objects, such as products, like for example pharmaceutical products or other health-related products, and in particular to the automatic identification and authentication of such objects. In particular, the present invention is directed to a method and a system for automatic object identification, a method for training such a system, a method for authenticating an object using an object authentication device communicating with said system, and such an object authentication device itself, as well as to an associated computer program corresponding to said method. Moreover, the present invention relates to a system for automatic object authentication, comprising said system for automatic object identification and one or more said object authentication devices, said system for automatic object authentication being particularly useful as an anti-counterfeiting protection and product tracking system. Background Art

[0002] In many industries, product counterfeiting is a major problem that not only seriously affects the revenue of the original product manufacturer, but can even seriously threaten the health or even life of the consumer or operator of the counterfeit product (i.e., fake product). These safety-related product categories include, among others, parts for automobiles and aircraft, components for building buildings or other infrastructure, food, and even medical devices and pharmaceuticals.

[0003] In addition, in a wide range of different industries, the traceability of goods and physical objects is a key requirement. This is particularly applicable to logistics and supply chain infrastructures and highly regulated / structured workflow environments. Examples are industrial workplaces that are controlled by official regulators (such as the FDA (U.S. Food and Drug Administration)), and / or certified (for example, according to GMP (Good Manufacturing Practice), GLP (Good Laboratory Practice), GCP (Good Clinical Practice) or DIN ISO or other similar standards and rules). Each of these regulated environments specifically requires audit trails and auditable technologies. Another example is the traceability of high-value products (such as industrial spare parts) in order to prove the authenticity and intended use of these parts in the secondary market.

[0004] In order to limit counterfeiting and provide integrity of the supply chain and workflow, including identification and authentication of products within the workflow and supply chain, various industries have developed many different protection measures and identification solutions. Widely used protection measures include adding so-called security features to products, which are difficult to counterfeit. For example, holograms, optically variable inks, security threads and embedded magnetic particles are known security features that are difficult to reproduce by counterfeiters. Although some of these security features are "public", that is, they can be easily seen or otherwise identified by users of the product, other security features are "covert", that is, they are hidden and can only be detected by using specific equipment (such as UV light sources, spectrometers, microscopes or magnetic field detectors, or even more sophisticated criminal investigation equipment). Examples of hidden security features are especially printing using cold light inks or inks that are only visible in the infrared part of the electromagnetic spectrum and invisible in the visible part of the spectrum, specific material compositions and magnetic pigments.

[0005] A specific set of security features, used in particular in cryptography, is known as a "physically unclonable function" (PUF). PUFs are sometimes also referred to as "physically unclonable functions" or "physical random functions". A PUF is a physical entity implemented in a physical structure and is easy to assess but difficult to predict, even for an attacker with physical access to the PUF. A PUF depends on the uniqueness of its physical microstructure, which typically includes a random component that is already inherently present in the physical entity, or is explicitly introduced into or generated in the physical entity during its manufacturing process, and is essentially uncontrollable and unpredictable. Thus, even PUFs manufactured by exactly the same manufacturing process are different at least in their random components and can therefore be distinguished. Although in most cases PUFs are covert features, this is not restrictive and public PUFs are also possible. In addition, PUFs are ideal for enabling passive (i.e., without active broadcast) identification of physical objects.

[0006] PUFs are particularly known in connection with their implementation in integrated circuits, and in particular for deriving cryptographic keys therefrom, for example in chips for smart cards or other security-related chips, by minimal unavoidable variations of microstructures produced on the chip within given process-related tolerances. Examples of such chip-related explanations and applications of PUFs are disclosed in the article “Background on Physical Unclonable Functions (PUFs)” (Virginia Tech, Department of Electrical and Computer Engineering, 2011), which can be found on the Internet at the hyperlink: http: / / rijndael.ece.vt.edu / puf / background.html.

[0007] However, other types of PUFs are also known, such as a random distribution of fibers in paper used as a substrate for making banknotes, where the distribution and orientation of the fibers can be detected by a specific detector and used as a security feature of the banknote. To evaluate a PUF, a so-called challenge-response authentication scheme is used. A "challenge" is a physical stimulus applied to the PUF, and a "response" is its reaction to the stimulus. The response depends on the uncontrollable and unpredictable nature of the physical microstructure and can therefore be used to authenticate the PUF and therefore also the physical object of which the PUF forms a part. A specific challenge and its corresponding response together form a so-called "challenge-response pair" (CRP).

[0008] An anti-counterfeiting protection system based on the use of PUFs for authenticating products is described in each of the two European patent applications EP 16205928.1 and EP 16205920.8, each of which is incorporated herein by reference in its entirety.

[0009] Asymmetric cryptography, sometimes also referred to as "public key cryptography" or "public / private key cryptography", is a known technique based on cryptographic systems using key pairs, where each pair includes a public key and a private key. The public key may be widely disseminated and often even publicly available, while the private key is kept secret and is usually known only to its owner or holder. Asymmetric cryptography enables (i) authentication, which is when a public key is used to verify the holder of the paired private key who initiated a particular information (e.g., a message or stored data containing the information) by digitally signing it with its private key, and (ii) protection of information (e.g., a message or stored data) by encryption, whereby only the owner / holder of the paired private key can decrypt a message encrypted by someone else with the public key.

[0010] Recently, blockchain technology has been developed, wherein a blockchain is a public ledger in the form of a distributed database containing multiple blocks of data, and it maintains a growing list of data records and is reinforced by cryptographic means to prevent tampering and revision. A prominent application of blockchain technology is virtual Bitcoin for currency transactions in the Internet. Another known blockchain platform is provided by, for example, the Ethereum project. In essence, a blockchain can be described as a decentralized protocol for recording transactions between parties, which transparently captures and stores any modifications to its distributed database and saves them "permanently" (i.e., as long as the blockchain exists). Storing information in a blockchain involves digitally signing the information to be stored in the blocks of the blockchain. In addition, maintaining a blockchain involves a process called "blockchain mining", in which so-called "miners" are part of the blockchain infrastructure, verifying and sealing each block so that the information contained therein is saved "forever" and the block can no longer be modified.

[0011] Another new ledger technology is known by the name of “Tangle”, which is a blockless and permissionless distributed ledger architecture that is scalable, lightweight, and provides consensus in a decentralized peer-to-peer system. A prominent related technology that uses Tangle as a technical foundation is called “IOTA”, which is a transaction settlement and data integrity layer for the Internet of Things. Summary of the invention

[0012] It is an object of the present invention to provide an improved way of efficiently identifying and authenticating physical objects such as products.

[0013] A solution to this problem is provided by the teaching of the attached independent claim. Various preferred embodiments of the invention are provided by the teaching of the dependent claims.

[0014] Furthermore, a complete object authentication scheme is presented herein, including a method and system for object recognition, a method for training the system and an apparatus for object authentication as well as related methods and corresponding computer programs as different aspects that can form part of a comprehensive multi-component object authentication solution for effectively protecting physical objects from counterfeiting and tampering.

[0015] A first aspect of an object authentication scheme provided herein is directed to a method for automatic object recognition. The method comprises: (i) receiving object data representing one or more distinguishing features of a physical object or a group of physical objects; (ii) processing the object data by means of a machine learning-based object recognition process to obtain distinguishing data representing one or more collision-resistant virtual representations of the physical object or the group of physical objects; (iii) comparing at least one of the distinguishing data and a raw hash value derived therefrom by applying a predetermined cryptographic hash function thereto with corresponding reference data stored in one or more data repositories with restricted access; and (iv) if the comparison with the reference data results in a match, i.e., consistency with respect to one or more predefined matching criteria, then outputting identification data including a digital signature of the hash value.

[0016] Optionally, the identification data may include further information such as, for example, time or location data or other metadata relating to the identified object or group of objects and / or the circumstances of the identification process.

[0017] The matching criteria may be specifically defined such that a "match" occurs if either the distinguishing data or the original hash value or both or a predefined combination thereof is contained in or otherwise represented by the reference data.

[0018] As used herein, the term "physical object" refers to any kind of physical object, in particular any kind of man-made product (such as, for example, but not limited to, a pharmaceutical product or other health-related product), or a natural object (such as, for example, but not limited to, a vegetable or a piece of natural raw material); or packaging of any one or more of the above. The physical object itself may include multiple parts, such as a consumer product and its packaging.

[0019] As used herein, the term "authentication" refers to confirming the authenticity of the properties of a physical object that an entity claims to be authentic, particularly its kind and its originality.

[0020] As used herein, the term "object data" refers to data describing or otherwise representing an object or group of objects of interest, including being described by one or more distinguishing characteristics thereof. In this document, the term "distinguishing characteristic" refers to at least one characteristic property of an object or group of objects that is suitable for distinguishing the object or group of objects from other objects in an at least substantially conflict-resistant or even bijective manner.

[0021] As used herein, the term "machine learning-based object recognition process" refers to a process in which one or more computers are used to recognize a physical object or a group of physical objects based on input data (such as sensory data) (e.g., image or video data captured by one or more cameras used as sensors), characterize the object or group of objects, and use one or more machine learning-based algorithms. In the context of the present invention, the "object data" is used as input data. Machine learning is an application of artificial intelligence (AI) that provides systems with the ability to automatically learn and improve from experience without being explicitly programmed. Machine learning can include, but is not limited to: (i) supervised learning, in which a computer is presented with example inputs and their expected outputs given by a "teacher", and the goal is to learn general rules that map inputs to outputs; (ii) semi-supervised learning, in which the computer is given only an incomplete training signal: a training set that lacks some (often many) target outputs; (iii) active learning, in which the computer is only given training labels for a limited set of examples (based on a budget) and must also optimize its object selection to obtain its labels; (iv) reinforcement learning, in which training data (in the form of rewards and penalties) is given only as feedback to the program's actions in a dynamic environment (such as driving a vehicle or playing a game with an opponent); or (v) unsupervised learning, in which the learning algorithm is given no labels and is left to find structure in its input.

[0022] As used herein, the term "collision-resistant virtual representation" of a physical object or group of physical objects refers to its data representation, which is defined to make it difficult (and in particular, almost impossible in practice) to find two at least substantially different objects or groups of objects such that their respective data representations are equal (i.e., indistinguishable).

[0023] The term "cryptographic hash function" as used herein refers to a special type of hash function, i.e. a mathematical function or algorithm that maps data of arbitrary size to a bit string of fixed size (hash value), which is designed to be also a one-way function, i.e. a function that is easy to calculate for each input but difficult to invert given an image of a random input. Preferably, the cryptographic hash function is a so-called "collision-resistant" hash function, i.e. a hash function that is designed to make it difficult (in particular, almost impossible in practice) to find two different data sets d1 and d2 such that hash(d1)=hash(d2). Prominent examples of such hash functions are hash functions of the SHA family (e.g. SHA-3 function) or hash functions of the BLAKE family (e.g. BLAKE2 function). In particular, so-called "provably secure cryptographic hash functions" can be used. These are hash functions whose specific security level can be mathematically proven. In the object authentication solution of the present invention, the security of the cryptographic hash function can be further improved by the fact that the analysis of the product or product group to be checked occurs at a specific location and time, where the physical object or object group actually exists at such a location and time. This can be used to increase the absolute security level that can be achieved, or to allow the use of cryptographic hash functions that work with smaller data sets (e.g., shorter data strings as input and / or output) while still providing a given required security level.

[0024] As used herein, the term "digital signature" or "digital signing" etc. refers to (using) a set of one or more digital values ​​that confirms the identity of the sender or initiator of digital data and the integrity of the latter. In order to create a digital signature, a hash value is generated from the data to be protected by applying an appropriate cryptographic hash function. This hash value is then encrypted with the private key (sometimes also referred to as a "security key") of an asymmetric cryptographic system, such as based on the RSA cryptographic system, where the private key is usually only known to the sender / initiator. Typically, a digital signature includes the digital data itself and a hash value derived therefrom by the sender / initiator. The recipient can then apply the same cryptographic hash function to the received digital data, decrypt the hash value included in the digital signature using the public key corresponding to the private key, and compare the decrypted hash value from the digital signature with the hash value generated by applying the cryptographic hash function to the received digital data. If the two hash values ​​match, this indicates that the digital information has not been modified, so its integrity is not compromised. Furthermore, the authenticity of the sender / originator of the digital data is confirmed by an asymmetric cryptographic system, which ensures that encryption with a public key only works if the encrypted information was encrypted with a private key mathematically paired with the public key. The representation of the digital signature can be realized in particular using an RFID transmitter or a single or multidimensional barcode (such as a QR code or a DATAMATRIX code) or simply as a multi-digit number.

[0025] The term "restricted access data repository" refers to a data storage device, such as a database, wherein the data stored therein can only be accessed with prior authorization, in particular authentication of the entity or person attempting access. Without limitation, such restricted access may be achieved with the aid of password protection or even two-factor or multi-factor authentication, for example by means of a plurality of independently provided passwords or other identification means, such as digital signatures.

[0026] This method of authenticating an object defines one of several aspects of the overall object authentication solution. Within the overall solution, it is used to identify the physical object to be authenticated based on the corresponding object data and with the help of a machine learning-based object recognition process applied thereto, and to return a hash code that corresponds to the identified object according to previously generated and stored reference data and can therefore be used for its identification. The object data can be received, for example, from an object authentication device as described below, which can form another component of the object authentication solution and is particularly used for sensor-based acquisition of feature information describing the object. The returned hash code can then be used by the object authentication device to authenticate the object by means of a comparison involving, on the one hand, the hash code or its derivative and, on the other hand, independently acquired information related to the object (such as, identity information provided on or in association with the object or group of objects itself).

[0027] In some embodiments, similarly applicable to other aspects of the present solution, the physical object includes one or more of the following items for consumption (consumer goods) or use: pharmaceutical or cosmetic compounds or compositions; medical devices; laboratory equipment; spare parts or components of equipment or systems; pesticides or herbicides; seed materials; coatings, inks, paints, dyes, pigments, varnishes, impregnating substances, functional additives; raw materials for additive manufacturing of products, products of additive manufacturing processes, i.e. 3D printed products. In particular, all of these items have in common the need to prevent counterfeiting in order to avoid malfunctions, health threats or other risks.

[0028] A second aspect of the object authentication solution is directed to a method of training a system for automatic object recognition. The method comprises: (i) receiving object data representing one or more distinguishing features of a physical object or a group of physical objects; (ii) processing the object data by means of a machine learning-based object recognition process to obtain raw distinguishing data representing one or more collision-resistant virtual representations of the physical object or the group of physical objects; and (iii) storing reference data including the raw distinguishing data and a raw hash value derived therefrom by applying a predetermined cryptographic hash function thereto in one or more restricted-access data repositories.

[0029] This method of training a system for automatic object authentication defines another aspect of the overall object authentication solution. Within the overall solution, it is used to generate and store reference data to prepare and thereby enable the system to perform the method of the first aspect during authentication of a physical object or group of physical objects.

[0030] In the following, selected embodiments of such a method of training a system are described, which embodiments can be arbitrarily combined with each other or with other aspects of the solution described herein, unless such a combination is explicitly excluded, inconsistent or technically impossible.

[0031] In some embodiments, storing the reference data includes storing the original distinguishing data in a first data repository in the data repository and storing the identification data in a second data repository separated from the data repository, such that neither the first data repository nor the second data repository stores both the original distinguishing data and the corresponding original hash value. This separation allows the original distinguishing data and the original hash value to be stored in different spatial locations, such as a data center, thereby enabling, for example, distributed computing and distributed data storage. In addition, access and control of the data repositories can thus be separated, which allows, for example, outsourcing the operation of one of the data repositories, and an increased level of data security is achieved due to the separation of the original distinguishing data from the corresponding original hash value.

[0032] In some embodiments, processing the object data to obtain the original distinguishing data comprises generating a plurality of anti-collision virtual representations of the physical object or group of physical objects by means of performing one or more predefined modification operations, such that at least two of these virtual representations represent the same physical object or group of physical objects, but each virtual representation is in a different condition. In particular, in some embodiments thereof, the different conditions relate to one or more of the following: (i) different age points in the life cycle of the physical object or group of physical objects; (ii) different environmental conditions to which the physical object or group of physical objects may be exposed during the detection of the corresponding one or more distinguishing features of the physical object or group of physical objects during a subsequent object authentication process. Such environmental conditions (ii) of each item may, for example, specifically relate to one or more of the lighting, temperature, air pressure or other parameters of the environment in which the object or group of objects may be located during the authentication process. The above different age points (i) of each item can be particularly used to create corresponding virtual representations of the object or group of objects associated with its different ages, and thus can be used to improve the recognition of objects that are generally subject to aging effects. These embodiments can be particularly used to improve the reliability of object recognition, in particular the success level of object recognition, because the effects of changes related to aging or different environmental conditions are thus incorporated into the recognition process. In some related embodiments, at least one of the predefined modification operations is at least partially defined by one or more tensors to be applied to vector representations of the object data or data derived from the object data. This allows a particularly efficient and well-structured implementation, in which only linear operations need to be performed (thus performing simple and fast operations). The different virtual representations can then each be used as distinguishing data for comparison purposes. For example, in the case of various virtual representations of the same object but for different ages, a comparison can be performed on two or more virtual representations to check whether at least one of them results in a match and therefore in a successful identification of the object based on the associated object data.

[0033] In some embodiments, storing the reference data comprises storing combined data in one or more restricted access data repositories, the combined data being formed as a combination of the reference data and additional adjunct values ​​defined independently of the reference data. In particular, this approach may be used in situations where two or more data repositories are used and the respective data to be stored therein must be distributed to one or more of them via one or more respective communication links. Adding the adjunct values ​​serves to increase the entropy of the data to be transmitted, thereby increasing the achievable level of protection against interception, such as, for example, against brute force attacks against the data stream on the communication link, or, if the adjunct values ​​are also stored in the data repository, against brute force attacks against the data repository itself.

[0034] In some related embodiments, the combined data is formed by: (i) combining the original distinguishing data or the original hash value derived therefrom (on the one hand) with the auxiliary value (on the other hand) in a predetermined reversible (i.e., injective) manner to obtain a combined value representing the two; and (ii) including the combined value in the reference data. For example, but not limitation, in some of these embodiments, the combined data is formed by applying a cryptographic hash function to input data including both the original hash value and the auxiliary value (optionally, other information) to obtain the combined value.

[0035] In some embodiments, processing the object data comprises processing data representing one or more of the following properties of the physical object or group of physical objects, respectively: (i) characteristic biometric or physiological properties of a living being or group of living beings that respectively forms at least a part of the physical object or group of physical objects; (ii) characteristic electrical, magnetic or electromagnetic properties, such as color or wavelength; (iii) characteristic acoustic properties; (iv) characteristic geometric properties; (v) characteristic chemical properties; (vi) characteristic mechanical or thermodynamic properties. These types of properties have the advantage that they can usually be detected automatically with the aid of corresponding existing types of sensors, without the need for human intervention.

[0036] In some embodiments, the method further comprises iteratively retraining the system for automatic object recognition based on additional object data representing distinguishing features of a physical object or group of physical objects, for which corresponding raw reference data have previously been generated and stored in one or more data repositories during a previous iterative step of training the system, wherein the additional object data is used as input data for the machine learning-based object recognition process. In this way, the system can be effectively trained to improve its ability in recognizing a given physical object or group of objects, such as the correct recognition rate (true positive rate, true negative rate). In particular, to the extent that the corresponding object data of various iterations correspond to different situations or conditions, this can also involve improving its ability to recognize objects of the group of objects under a certain (wider) range of different conditions, under which sensor-based measurements of the corresponding object / group of objects occurred for the purpose of generating the corresponding object data.

[0037] In some embodiments, storing the reference data comprises one or more of: (i) storing the reference data in at least one of the data repositories in a digitally signed form; (ii) storing or causing one or more nodes of a blockchain environment to store the identification data in a digitally signed form in a block of a blockchain associated with the blockchain environment; (iii) storing or causing one or more nodes of a blockless distributed ledger environment (such as, for example, a Tangle ledger environment) to store the identification data in a digitally signed form in at least one node of the blockless distributed ledger environment; (iv) storing the identification data in a digitally signed form in a storage of a public / private key environment. The digital signature serves as a protection measure based on which the originality of the stored data can be verified. This also applies in particular when retrieving and transmitting the stored data over a communication link (e.g., to an object authentication device), as described herein. In addition, storage in a public / private key environment adds encryption, so that the stored and / or transmitted data is further protected from interception.

[0038] Specifically, storing identification data in a blockchain or in a blockless distributed ledger (e.g., a Tangle-based ledger) enables secure and reliable storage of read results with very high data integrity, making it essentially impossible to manipulate or erase or otherwise tamper with or lose such data, for example due to accidental or intentional deletion or due to data corruption. Therefore, the complete storage history remains available. In addition, the stored information can be accessed as long as the blockchain or blockless distributed ledger is accessible, respectively. This allows secure and distributed storage and access to stored identification data, for example for integrity verification or authentication purposes, such as checking whether the claimed product supplier is actually the originator of the product. Based on one or more of these embodiments, the physical world to which the object to be authenticated belongs can be connected to the power of blockchain technology. Therefore, a high degree of traceability of the origin and supply chain of physical objects such as products can be achieved.

[0039] In some embodiments of the method of the first aspect or the second aspect, the corresponding method is performed by a system including two or more separate devices, which jointly perform the method and are connected to each other by one or more data links protected to prevent unauthorized interception. The security of the data link can particularly include data encryption. Of course, other protective measures are also possible, for example, physical protection (especially to prevent interception or destruction) in the case of physical data lines. Therefore, in particular in view of man-in-the-middle attacks, the security of the system and method can be improved.

[0040] A third aspect of the present solution is directed to a system for automatic object recognition, the system being configured to perform the method of one or more of the first and second aspects of the present solution, as described herein.

[0041] A fourth aspect of the present solution is directed to a computer program comprising instructions which, when executed on one or more processors of a system for automatic object recognition according to the third aspect, cause the system to perform a method according to either or both of the first and second aspects of the present solution.

[0042] Thus, the explanations provided above with respect to the methods apply with appropriate modifications to the system according to the third aspect of the present solution and to the computer program according to the fourth aspect of the present solution.

[0043] The fifth aspect of the present solution is directed to a method for authenticating an object using an object authentication device, comprising: (i) sensor-based detection of one or more distinguishing characteristics of a physical object or a group of physical objects; (ii) generating object data representing the one or more distinguishing characteristics of the physical object or the group of physical objects; (iii) transmitting the object data to a system according to the third aspect of the present solution; (iv) receiving digitally signed identification data from the system in response to the transmission of the object data; (v) reading a mark provided on the physical object or the group of physical objects, respectively, or in combination with the physical object or the group of physical objects, to obtain digitally signed, optionally encrypted, identity information therefrom; (vi) verifying the correctness of each of the digital signature of the identification data and the digital signature of the identity information; (vii) comparing the received identification data with the obtained identity information to verify the authenticity of the object or the group of objects based on the result of such comparison; and (viii) outputting authentication information indicating whether the physical object or the group of physical objects is determined to be authentic based on the result of the verification.

[0044] In particular, the determination of authenticity may require that both verifications are successful, i.e., the digital signature correctly identifies the expected / claimed originator, e.g., the correct object identification system and object originator (e.g., the manufacturer in the case of a product), and that the comparison of the received identification data with the obtained identity information results in a match according to one or more predetermined matching criteria. The term "verifying the correctness of a digital signature" as used herein refers to a general method of verifying the originality of a digital signature, in particular including reading it by applying the associated public key of the assumed originator in order to check whether it is original (i.e., signed with the associated secret private key of said originator).

[0045] In particular, the object authentication device may be a mobile device, such as a device specifically designed for authentication purposes or a general purpose device, such as a smartphone or a portable computer specifically programmed to perform the authentication method and use one or more sensors thereof for detecting object data, such as a camera, a microphone or an RFID reader. Thus, the object authentication device may be a device that can be used variably at different locations, such as along a supply chain of a product. On the one hand, it serves as a device for generating object data related to an object or a group of objects to be authenticated by means of sensor-based detection of one or more distinguishing features thereof. The object data is then transmitted to a system for object recognition, which system may in particular be a central entity serving a plurality of object authentication devices, and the object data is processed by means of a machine learning-based object recognition process to obtain raw distinguishing data representing one or more conflict-resistant virtual representations of a physical object or a group of physical objects.

[0046] The system for object identification also compares at least one of the distinguishing data and a raw hash value derived therefrom by applying a predetermined cryptographic hash function thereto with reference data stored in one or more access-restricted data repositories. If the comparison with the reference data results in a match, i.e., if the object or group of objects is identified, the system outputs identification data (i.e., identification data associated with the identified object or group of objects) including a digital signature of the hash value, and specifically transmits it to a corresponding object authentication device.

[0047] On the other hand, the object authentication device then serves as a device for actually authenticating the object or group of objects (i.e. for verifying its originality) by comparing the received identification data with identity information obtained from the object or group of objects itself (e.g. from a barcode, RFID chip or any other suitable marking provided on the object / group of objects or on the packaging or materials provided therewith). In particular, the marking may comprise an optionally encrypted representation of a digital signature of the identity information or a representation of a pointer indicating the location where the digital signature can be accessed.

[0048] If applied, the encryption of the representation of the digital signature and / or of the representation of the pointer indicating the location where the digital signature can be accessed adds a further level of security, since before the digital signature can be read the corresponding representation first needs to be decrypted, which requires knowledge of the encryption scheme and the correct cryptographic key. The encryption can in particular be based on known symmetric or asymmetric encryption schemes, for example according to the well-known AES (symmetric) or RSA (asymmetric) cryptographic schemes.

[0049] The term "pointer indicating the location where the digital signature can be accessed" as used herein may in particular be a pointer to a local or remote database or to a server address or an internet address, such as a hyperlink, etc., at which the digital signature can be accessed (e.g. downloaded). The pointer may in particular be implemented using an RFID transmitter or a one-dimensional or multi-dimensional barcode (such as a QR code or a DATAMATRIX code) as its representation.

[0050] Specifically, in some embodiments, the comparison of the received identification data with the identity information is achieved by means of a comparison of two corresponding hash values ​​(i.e., the original hash value represented by the identification data and the corresponding hash value obtained from / via the tag). If the two hash values ​​match, then this indicates that the physical object is authentic and the tag has not been tampered with. Otherwise, i.e., if they do not match, then this indicates that some fraud may have occurred since the originator applied the tag to the physical object.

[0051] In some embodiments, in the event of an authentication failure or a determined lack of authenticity, the object authentication device may itself output, or cause another entity to output, an alarm signal to indicate the failure or lack of authenticity to a human user or to another device, respectively.

[0052] Thus, within the framework of the present solution, an object authentication device can interact with the object recognition system to securely authenticate an object or a group of objects, wherein, unlike the above-mentioned PUF-based solutions, the object to be authenticated does not need to be specifically marked with a PUF and the distinguishing features of the object do not need to be predefined and fixed, but can instead be determined and continuously refined by a machine learning-based recognition process used by the object recognition system.

[0053] In the following, further selected embodiments of the object authentication method are described, which embodiments can be arbitrarily combined with each other or with other aspects of the solution described herein, unless such a combination is explicitly excluded, inconsistent or technically impossible.

[0054] In some embodiments, the sensor-based detection of a distinguishing feature of a physical object or group of physical objects includes detecting at least one distinguishing feature that is at least substantially invariant under changes in environmental conditions in which the at least one distinguishing feature is detected. This may be the case, for example, if the distinguishing feature is provided by an object-specific characteristic signal emitted by an RFID tag attached to the object. Moreover, in some cases, specific properties of the material of the object may show such invariance. In this way, the reliability of the entire object authentication process may be ensured.

[0055] In some embodiments, the method further comprises: including the original hash value or another object-specific information contained in the identification data in the output authentication information. This can have multiple purposes. In particular, if the authentication information is output at a human-machine interface (such as a display), this can be used to inform the user of the object authentication device of the original hash value or another object-specific information, respectively. Moreover, the output can be used to similarly inform another automated process, for example via a corresponding machine-to-machine interface.

[0056] Furthermore, in some related embodiments, the method further comprises a storage process, wherein the storage process comprises digitally signing and storing or causing another device (such as a separate and optionally even remotely located computer configured to perform (i) blockchain mining or (ii) write to a node of a blockless distributed ledger) to store the original hash value or the other object-specific information or both in a digitally signed form into (i) a block of the first blockchain or (ii) a node of the first blockless distributed ledger, respectively. This enables secure, reliable storage with very high data integrity, making it essentially impossible to manipulate or erase or otherwise gradually reduce or lose such data, for example due to accidental or intentional deletion or due to data corruption. Thus, the complete storage history remains available. Furthermore, the stored information can be accessed as long as the blockchain can be accessed. This allows secure and distributed storage and access to the stored data, for example for integrity verification purposes, such as checking whether the supplier of a product (object) is actually the originator of the product. Based on this embodiment, the physical world to which the object belongs can be connected to the power of blockchain technology. Thus, a high degree of traceability of the origin and supply chain of physical objects such as products can be achieved.

[0057] In some related embodiments, the storing process further comprises: (i) reading a tag provided on or in association with the physical object or group of objects to be authenticated, or detecting one or more selected characteristics of the object or group of objects to obtain supplementary information related thereto; and (ii) digitally signing and storing or causing another device to store the supplementary information in a digitally signed form in a block of a second blockchain separate from the first blockchain or in a node of a second blockless distributed ledger separate from the first blockless distributed ledger. For example, the supplementary information may be information related to a supply chain, and may specifically include one or more of the following: (i) location information related to the location at which the supplementary information was acquired by the object authentication device; (ii) authentication information of a user of the object authentication device; (iii) time and / or date information indicating a point in time at which the supplementary information was acquired by the object authentication device; (iv) product identification, serial number and / or batch number of the object marked by the tag; (v) manufacturing date or expiration date of the object marked by the tag; (vi) information identifying the manufacturing location of the product, etc.

[0058] These embodiments allow the supplementary information to be additionally stored and thus saved to the respective second blockchain, thereby providing the advantages discussed in connection with the immediately preceding embodiments also in relation to the supplementary information. Using different blockchains or blockless distributed ledgers for the original hash value or said other object-specific information on the one hand and for the supplementary information on the other hand further provides the advantage of easily supporting a combination of an existing (second) blockchain or blockless distributed ledger, respectively, for the supplementary information and an additional first blockchain or blockless distributed ledger, respectively, for the original hash value or said other object-specific information.

[0059] Thus, different access rights can be easily implemented and the management of the blockchain can be in the hands of different institutions. In particular, these embodiments can be used to verify whether the supplier of a product is actually its originator and whether the supply chain is as expected. In addition, this can be used to (i) check the tag / object in view of whether it has been forged or tampered with, and (ii) read and output additional information from the tag, such as supply chain or other logistics information. However, in addition, a combination of both uses (i) and (ii) can be used to further improve the security aspects of the present object authentication solution, because such additional information (such as supply chain information) can be used to retroactively identify the locations or people involved in the supply chain where potential fraud may have occurred, as well as potential related dates or time frames. Thus, an object authentication device suitable for performing the methods of these embodiments is a dual-purpose or even multi-purpose device, which increases ease of use and reduces the number of different devices required to read a complete composite security tag.

[0060] In some related embodiments, the storage process in the case of a blockchain further includes: (i) when storing the original hash value and / or the other object-specific information in a block of a first blockchain, including a cross-blockchain pointer pointing to the block of the first blockchain, which logically maps the block of the first blockchain to the corresponding block of the second blockchain; and (ii) when storing the supplemental information in a block of a second blockchain, including a cross-blockchain pointer pointing to the block of the second blockchain, which logically maps the block of the second blockchain to the corresponding block of the first blockchain.

[0061] Similarly, in the case of a blockless distributed ledger, the storage process further includes: (i) when storing at least one of the hash values ​​in a node of a first blockless distributed ledger, including a cross-ledger pointer pointing to the node of the first blockless distributed ledger, which logically maps the node of the first blockless distributed ledger to the corresponding node of the second blockless distributed ledger; and (ii) when storing the supplemental information in a node of a second blockless distributed ledger, including a cross-ledger pointer pointing to the node of the second blockless distributed ledger, which logically maps the node of the second blockless distributed ledger to the corresponding node of the first blockless distributed ledger.

[0062] In this way, the two blockchains or two blockless distributed ledgers can be interconnected via a cross-blockchain pointer or a cross-ledger pointer, respectively, which can be used to further increase the achievable security level of the present object authentication solution. In particular, this can be used to track attempts to tamper with or counterfeit a tagged object at different points along the supply chain. For example, this embodiment allows tracking the location and / or time point of such an attempt, or, in the case of mandatory authentication at the object authentication device, tracking the identity of the user or entity involved in such an attempt.

[0063] In some embodiments, the authentication information is at least partially output in the form of a one-dimensional or multi-dimensional barcode (such as a DATAMATRIX code or a QR code). This allows the output authentication information to be further processed using a readily available barcode scanner, which is particularly advantageous in the case where the object authentication device is integrated into or interacts with an automated production line or other processing line, where its output needs to be further processed by an algorithm that is processed by the production line rather than a human user.

[0064] In some embodiments, the method further comprises: if the physical object or group of physical objects is not determined to be authentic based on the authentication information, then repeating the method of object authentication until the earlier of (i) successful authentication or (ii) satisfying a predetermined termination criterion. For example, the termination criterion may be defined in terms of the number of failed authentication attempts.

[0065] In some embodiments, the method further comprises performing user authentication and allowing or denying the user to perform object authentication with the object authentication device based on the result of the user authentication. This can be advantageously used to further improve the security of the solution by preventing unauthorized users from successfully interacting with the object authentication device and thereby participating in the security chain provided by the present solution. In addition, this can be used to obtain user identities or other user-related information that can be used to increase transparency of the flow of physical objects (particularly products) to be authenticated along the supply chain. In the case of security concerns, such information can be used to track potential threats to security provided by the overall solution and identify locations or persons that may be associated with such threats.

[0066] In some embodiments, the method further comprises transmitting the authentication information and / or identification data (in whole or in part) and / or other information derived from the authentication data and / or identification data to the opposite side via a communication link. In particular, the communication may be suitable for sending and receiving data via a wired, wireless or optical communication link (by way of example but not limitation, such as a communication link based on a wireless LAN, Bluetooth, a cellular network or a classic telephone line). Such a communication link can be used for a variety of different purposes, including for sending the acquired information (e.g., the output authentication information) to the opposite side, which can, for example, be a central security instance, such as a trust center comprising a central security server, which can form an integral part of the present authentication solution.

[0067] In some related embodiments, the method also includes capturing security-related information and sending it to the opposite side via a communication link. The opposite side may be, for example, the trust center mentioned above. In particular, the sending of such security-related information may occur randomly, or may be specifically triggered according to a predetermined triggering scheme or remotely (e.g., by the opposite side). This allows remote monitoring of the security status of the object authentication device itself and / or security-related events to which the object authentication device is involved. Such security-related events may be, for example, the detection of a tag / object that has been forged or tampered with based on the authentication information or other security-related information provided by the object authentication device.

[0068] Specifically, according to relevant embodiments, security-related information includes one or more of the following: (i) location information characterizing the current or past location of the object authentication device; (ii) user data characterizing or identifying the user of the object authentication device; (iii) network data characterizing the communication link; (iv) information characterizing attempts or actual actions detected by at least one sensor of the object authentication device or the corresponding reaction of the object authentication device (for example, as described above); (v) authentication information generated by the object authentication device.

[0069] In some related embodiments, the method also includes detecting security events in information contained in a signal received from the opposite side via the communication link. In particular, where an authorized opposite side (e.g., a central security center) sends information containing such security events to the object authentication device, this can be used to trigger a transition of the object authentication device to a secure mode or even its deactivation, to avoid the negative impact that the object authentication device might otherwise have on the entire authentication system. For example, such a negative impact may result if any compromising behavior occurs (such as unauthorized intrusion or firmware / software modification at the object authentication device or use by unauthorized personnel or at an unauthorized location occurs and has been communicated to the opposite side or otherwise detected by the opposite side).

[0070] In some related embodiments, the method also includes sensor-based detection of one or more of the following as security events: (i) an attempt or actual physical intrusion into the object authentication device; (ii) an attempt or actual local or remote access to internal control functions of the object authentication device, wherein such access is not available to a user of the object authentication device during normal operation of the object authentication device. In particular, such attempted access may be directed to taking over control of or modifying the functions of the object authentication device. Therefore, this embodiment can be advantageously used to further increase the security level of the present object authentication solution, and in particular to protect the object authentication device itself and the entire solution presented herein from unauthorized intrusion and tampering.

[0071] In some related embodiments, the method further includes executing one or more of the following security measures in response to detecting the security event: (i) locking the object authentication device, such as to limit or prevent its further use; (ii) self-destructing at least one functional part of the object authentication device or destroying data stored therein to prevent the user from further using or accessing it; (iii) outputting an error message. In particular, as described above, the security measures can be considered as specific measures for putting the object authentication device into a safe mode or for disabling it.

[0072] A sixth aspect of the present solution is directed to an object authentication device adapted to perform the method of the fifth aspect.

[0073] In some embodiments, the object authentication device is integrated or otherwise forms an integral part of one or more of the following: a handheld device, such as a product or barcode scanning device; production, quality control or commissioning equipment; a production or quality control or commissioning line; a flying object, such as a drone; a robot, such as an agricultural robot; agricultural machinery. This allows the functionality of the object authentication device to be integrated into a system with additional or broader functionality, particularly in an automated or semi-automated manner. For example, in the case of a production quality control or commissioning line, the object authentication device can be integrated into the production line in a manner that enables it to automatically read the markings on the products traveling along the production line so as to perform an initial capture of the relevant data. The captured data can then be stored in a relevant database or compared with the stored data to verify that the production or commissioning line has produced and commissioned the expected set of products, respectively. Similarly, at one of the multiple nodes of the supply chain (such as a logistics center), such an object authentication device can be integrated inline into the identification and transportation system (e.g., a conveyor belt) so as to automatically or semi-automatically (e.g., in the case of a handheld device) check and verify the authenticity of the product before delivering the product to the next node in the supply chain. The same applies to the final node (i.e., the recipient and / or end user of the product).

[0074] According to another preferred embodiment, the object authentication device is a portable electronic communication terminal. Without limitation, the object authentication device may be, for example, a smartphone or a portable computer (e.g., a tablet computer). The communication capabilities (e.g., for cellular communication) present in any case in the electronic communication terminal may then be used to establish a communication link to the system for object identification.

[0075] A seventh aspect of the present solution is directed to a computer program comprising instructions which, when executed on one or more processors of an object authentication device according to the sixth aspect, cause the object authentication device to perform the method of the fifth aspect.

[0076] The computer program of the fourth and / or seventh aspect can be implemented in particular in the form of a data carrier on which one or more programs for executing the corresponding method are stored. This can be advantageous if the computer program product is intended to be traded as a standalone product in a standalone product independent of the processor platform on which the one or more programs are to be executed. In another embodiment, the computer program product is provided as a file on a data processing unit, in particular on a server, and can be downloaded via a data connection (e.g., the Internet or a dedicated data connection, such as a dedicated or local area network).

[0077] Thus, the explanations provided above with respect to the method of the fifth aspect apply with appropriate modifications to the object authentication device according to the sixth aspect and the computer program according to the seventh aspect of the present solution.

[0078] An eighth aspect of the present solution is directed to a system for automatic object authentication, comprising the system for automatic object recognition of the third aspect and the object authentication device of the sixth aspect, the system and the object authentication device being configured to jointly recognize and authenticate a physical object or a group of physical objects. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Further advantages, features and applications of the present object authentication solution are provided in the following detailed description and accompanying drawings, in which:

[0080] Further advantages, features and applications of the present object authentication solution are provided in the following detailed description and accompanying drawings, in which:

[0081] Figure 1 Schematically illustrates an overview of a preferred embodiment of the present object authentication solution, comprising a system for automatic object authentication;

[0082] Figure 2 Schematically illustrates the training of a preferred embodiment of the object authentication solution Figure 1 The method of the system shown in and its application to a physical object as a consumer good comprising a group of pharmaceutical tablets arranged in a blister pack and an associated outer packaging for the blister pack;

[0083] Figure 3A Schematically illustrates the use of a preferred embodiment of the object authentication solution in case of successful identification of the inspected object Figure 1 A method for automatic object recognition and authentication of the system shown in;

[0084] Figure 3B Schematically illustrates the Figure 3A The same method of automatic object authentication as in, but in the case where the inspected object is not successfully identified;

[0085] Figure 4A and 4B A flow chart illustrating a method for automatically authenticating an object with an object authentication device according to a preferred embodiment of the present object authentication solution is shown;

[0086] Figure 5 is a schematic diagram of a preferred embodiment of the present object authentication solution involving a PKI environment; and

[0087] Figure 6 Schematically, is the evolution of a set of two cross-connected blockchains along the supply chain of a product that is marked with a security mark according to a preferred embodiment of the present object authentication solution.

[0088] In the figures, the same reference signs are used for identical or mutually corresponding elements of the solutions described herein. DETAILED DESCRIPTION

[0089] Figure 1 System aspects of an overall object authentication solution 1 according to a preferred embodiment of the present invention are illustrated. Method aspects of solution 1 will be described below with reference to further figures. Solution 1 comprises a system 2 for automatic object authentication, which in turn comprises one or more object authentication devices 4 (only one of which is shown), each of which may in particular be a mobile device, such as a mobile object authentication device or even a mobile computer, for example a tablet or a smartphone having at least one sensor for capturing or measuring characteristics of a physical object A or a group of physical objects to be authenticated. Each object authentication device 4 comprises a sensor unit 4a, a processing unit 4b and a memory 4c. The memory 4c stores a computer program configured to be executed on the processing unit 4b in order to control the respective object authentication device 4, including its sensor unit 4a, and in particular to cause the respective object authentication device 4 to perform an object authentication method, for example as described below with reference to Figure 4A / 4B and 5A / 5B. The object authentication device 4 may also include a security device 4d, which includes one or more sensors for detecting security events (such as attempts or actual behavior of physically intruding into the object authentication device 4, or attempts or actual behavior of accessing the internal control functions of the object authentication device 4 locally or remotely without authorization). Preferably, the security device 4d interacts with or also includes a security defense arrangement 4e to protect the object authentication device 4 in the event of a security event being detected. In particular, the security defense arrangement 4e may be adapted to perform a similar Figure 4A The steps of step S5 are described in more detail below. For example, the security defense arrangement 4e can be configured to lock the user interface of the object authentication device 4 in the event of a security event being detected, or to activate the self-destruction of a security chip contained in the object authentication device 4, so as to protect data stored therein, including, for example, private cryptographic keys or other security-related data (such as authentication data). In addition to or in place of the security device 4d, the object authentication device 4 may also include a monitoring device 4f, which is configured to detect security events indicated in information contained in a signal received from the opposite side via the communication link. For example, in the event that such an opposite side (e.g., a trust center) becomes aware of a wider attempt to attack the security and integrity of, for example, object authentication devices 4 distributed on site along a given supply chain, such a signal can be used to actively trigger (at least temporarily) blocking any further use of the object authentication device 4 on site to prevent tampering with the object authentication device 4 through such an attack.

[0090] The system 2 for object authentication also comprises a system 3 for object identification, which can be implemented as a central backend system working together with the one or more object authentication devices 4. The system 3 can also serve as the opposite side or trust center mentioned above with respect to the security aspects of the object authentication devices 4.

[0091] The system 3 for object recognition comprises a recognition subsystem 3a, which in turn comprises a data analysis module 5, which is configured to analyze object data provided by one or more object authentication devices 4 using one or more machine learning-based algorithms (e.g., one or more neural networks), which algorithms may particularly include algorithms for pattern recognition, etc. To this end, the data analysis module 5 comprises a processing unit 5a having one or more CPUs and a memory 5b. The memory 5b stores one or more computer programs, which are configured to execute and implement the one or more algorithms on the processing unit 5a (or alternatively on a distributed processing platform (not shown)). The recognition subsystem 3a also comprises a data processing module 6, which is configured to also process the output data of the data analysis module 5, as will be described in more detail below. To this end, the data processing module 6 comprises a processing unit 6a having one or more CPUs and a memory 6b. The memory 6b stores one or more computer programs, which are configured to execute and implement the data processing on the data processing module 6. The data analysis module 5 and the data processing module 6 jointly form a data processing platform 3b of the recognition subsystem 3a. In fact, in some embodiments, the data analysis module 5 and the data processing module 6 may be jointly implemented as a single module forming a data processing platform 3b, which may specifically but not limitedly include a single processing unit and a single memory.

[0092] The recognition subsystem 3a further comprises a first data repository 7, which can be implemented in particular as a database. As will be explained in more detail below, the first data repository 7 is configured to store and provide reference data generated by the data processing performed by the data processing module 6 based on the analysis results output by the data analysis module 5 during the training system 3 for object recognition.

[0093] The system 3 for object identification further comprises a second data repository 8, which is implemented separately from the identification subsystem 3a. The second data repository 8 is configured to store and provide a subset of reference data resulting from the data processing performed by the data processing module 6, as will be explained in detail below. External to the system 2 for automatic object authentication, the object authentication solution 1 may comprise an external data ledger 9, which may in particular be implemented by a blockchain environment, a blockless distributed ledger environment or a PKI environment or a combination of one or more of the above mentioned. Figure 1As shown in , each of the above-mentioned components 4 to 9 is connected to each other via a set of protected data links L1 to L6. Each of the data links L1 to L5 can be individually selected as wired, wireless or optical type or any other suitable data transmission type. The protection of the data links can in particular include data encryption. Moreover, other protection measures are of course also possible, such as physical protection in the case of physical data lines (in particular against interception or destruction).

[0094] Figure 2 Schematically illustrates a training according to a preferred embodiment of the object authentication solution 1 Figure 1 3. In preparation for training, an object authentication device 4 (in this case typically a central object authentication device 4-1 provided at the location of the system 3) uses its sensor unit 4a to capture (in particular measure) characteristics of a known physical object A and outputs corresponding results in the form of object data OD. The object A may in particular be a product that is subsequently to be distributed and is therefore used in the training, so that it can be authenticated by the system 3 once the training is completed. The training is typically performed as part of the production or post-production process of the respective product A before it is distributed to the supply chain.

[0095] The training may be targeted to a specific single instantiation of object A, or alternatively, to just its product type in such a way that different instantiations of the same product type will result in the same training results, so these training results may be used to later authenticate any instantiation of a product of this product type, not just the specific individual product A. Figure 2 In the non-limiting example given, product A is a multi-part product comprising a plurality of consumer goods A1 in the form of a collection of pharmaceutical tablets arranged in a plurality of blister packs and corresponding product outer packaging A2 for the blister packs of A1.

[0096] During the training process, the data analysis module 5 receives the object data OD as input and uses its one or more algorithms based on machine learning to analyze the object data in order to derive, based on the received object data OD, distinguishing data representing one or more collision-resistant virtual representations of the physical object A. This derived distinguishing data is referred to hereinafter as "raw" distinguishing data ODD, in order to distinguish it from similar distinguishing data DD generated in the subsequent authentication process. The raw distinguishing data ODD is then forwarded to the data processing module 6, where it is processed to obtain reference data, which can be used as a reference in the subsequent object recognition / authentication process. The processing may in particular include: generating a plurality of collision-resistant virtual representations of the physical object A by means of performing one or more predefined modification operations, so that at least two of these virtual representations represent the same physical object A, but each virtual representation is in a different condition. The different conditions may in particular refer to different age points in the life cycle of the object A, or different environmental conditions to which the object may be exposed in the future, such as different lighting, temperature or pressure conditions. This allows the creation of reference data that provides a good match with the distinguishing data derived from the captured object data in the subsequent object recognition process, even if the age or environmental conditions are substantially different from the age or environmental conditions at the time of training.

[0097] The generated reference data are then grouped into a first subset RD1 of reference data and a different second subset RD2, each of which is digitally signed and stored in two different data repositories. In particular, the first subset RD1 is defined as an output CD(ID, AV1) such as comprising a data combination function CD(), the function CD() being configured to take as input the identification data ID and a predefined first auxiliary value AV1 and to perform a data combination operation (such as concatenation, mixing and scrambling, or otherwise mathematically combining the inputs in a predetermined reversible manner) so as to generate said output. The identification data ID comprises at least a hash value OH derived from the original distinguishing data ODD or a part thereof by applying a predetermined cryptographic hash function, and the identification data ID optionally comprises metadata MD comprising additional information related to the product or training. In particular, the metadata MD may comprise information indicating the time or place of the training, the user identity of a user operating the object authentication device 4 or the device identity of the device 4, a serial number and / or a date or place of production, or one or more preferred invariant characteristics of the object A. On the other hand, the auxiliary value AV1 can be any known value, even a random value, which is added as a further input to the combination function CD() in order to increase the entropy of the resulting first subset RD1 and thus enhance the security that can be achieved by encrypting or digitally signing RD1. Furthermore, the first subset RD1 can be stored in an external data ledger 9, in particular for the purpose of creating redundancy and reducing the risk of data loss and for the purpose of enhancing the accessibility of the RD1 data. In a similar manner, a second subset RD2 of reference data is generated by the data processing module 6, wherein RD2 in particular comprises information obtained by applying the same or different data combination function CD() to the original distinguishing data ODD or a part thereof and a second auxiliary value AV2, wherein the second auxiliary value AV2 plays a similar role as AV1 with respect to RD1.

[0098] In particular, the training process may be defined as iterative, in which case the data analysis module 5 receives a previous second subset PRD2 of reference data, generated in a previous training iteration regarding the same object A, as further input when processing the object data OD. After or even as part of the training process, the first subset RD1 of reference data including the identification data ID is forwarded to the object authentication device 4-1 and / or is used to create a mark M (in particular, for product A, respectively M A ), such as a one-dimensional or multi-dimensional barcode (e.g., a QR code or a data matrix code) or an RFID tag fixed or otherwise attached to or combined with the object A. The marker M (M A) includes an encoding of the RD1 data. The product A thus marked is then ready to be distributed to the supply chain, and the system 3 is trained to recognize the product A again on a later occasion based on the object data OD captured from the product A at a later point in time and possibly at a different location.

[0099] Figure 3A and 3B Schematically illustrates the use of a preferred embodiment according to the present authentication solution Figure 1 A method for automatic object recognition and authentication of the system shown in. Figure 3A It refers to the situation where the inspected object A is successfully identified, while Figure 3B It refers to the situation where the inspected object B is not successfully identified.

[0100] refer to Figure 3A , again using the object authentication device 4 to capture / measure characteristic information associated with the physical object A (such as a product) to be investigated and generate corresponding object data OD. However, now, the object authentication device 4 will typically be an object authentication device 4-2 specifically designed for use in the field rather than at the factory level, such as at a node in the supply chain or during customs inspection or anywhere else it is desired to track and / or authenticate products to detect counterfeiting. Thus, the object authentication device 4-2 will typically be the same as the one described above with reference to Figure 2 The object authentication device 4-1 used during the described training process (which is usually performed at factory level) is different.

[0101] Then, the object data is passed through Figure 1 The data link L1 shown in is sent to the system 3 for automatic object recognition. In a similar manner, as described above for the training process, the data analysis module 5 generates the distinguishing data DD and forwards it to the data processing module 6. The data processing module 6 accesses the first data repository 7 to retrieve the reference data RD2 stored therein during the training process and compares it with the distinguishing data DD received from the data analysis module 5. In particular, this comparison can be performed based on (i) the original hash code OH included in the RD2 reference data, usually in different object-related data sets, one for each of usually a plurality of different objects for which training has been previously performed, and (ii) the corresponding hash code derived from the distinguishing data DD. If the comparison results in a match, according to some predefined matching criteria for one of the object-related data sets included in the RD2 reference data (the matching criteria may in a simple case require that the two compared hash codes are mathematically equal), the data processing module 6 forwards the index IN value associated with the object A and the data set for which a match has been found.

[0102] The second data repository 8 then selects the reference data subset RD1 corresponding to the object A according to the index IN and forwards it to the object authentication device 4-2. As mentioned above, the digitally signed RD1 data includes identification data related to the object A. As described so far, this process can be called an object identification process, because it causes the identification data of the identified object A to be returned as a result.

[0103] However, this process can be extended to become an object authentication process, as shown below. Figure 4A and 4B As described in detail. As a result of the object authentication process, authentication data AD will be output by the object authentication device 4-2, and in a storage step, the received digitally signed RD1 data or at least the selected identification information ID included in the RD1 received from the second data repository 8 will be stored in an external data ledger 9 (e.g., a first blockchain). Optionally, the supplementary information SI can also be stored in the same or different external data ledger 9 (e.g., a second blockchain), as described in more detail below. The supplementary information SI may in particular include information related to one or more of the following: the time and / or place of the object authentication performed by the object authentication device 4-2, the user identity of the user operating the object authentication device 4-2, the device identity of the device 4-2, the serial number and / or the date or place of production, or one or more preferred invariant characteristics of the object A, and the identity of the successfully authenticated object A, which is taken out or derived from the identification data ID included in the received RD1 data.

[0104] Figure 3B Involving a Figure 3A Different from the situation where the object B under inspection is not successfully identified. This may be the case, for example, if no previous training has been performed on object B, or if problems such as detection errors or data loss or errors occur during the authentication process. Although the first part of the process can be similar to Figure 3A , but if, for example, a failed authentication is identified by the data processing module 6, the error signal ERR is forwarded to the second data repository 8 instead of the index IN, and a corresponding error message ERM is sent to the object authentication device 4-2 to inform it of the failed authentication. The object authentication device 4-2 can then repeat the process or output authentication data AD indicating that the authentication attempt failed. In addition, the error message ERM or another equivalent information can be stored in the external data ledger, optionally again together with the corresponding supplementary information SI, as described above with reference to Figure 3A as described.

[0105] Figure 4A and 4BA flow chart (divided into two parts connected via a connector "C") illustrating a preferred embodiment of a method for automatic object authentication with an object authentication device 4 (such as, for example, the device 4-2 discussed above, to which reference is made in the following method description) according to a preferred embodiment of the present invention is shown together. The method optionally includes a first stage, which includes steps S1 to S7, which is used to enhance the security of the object authentication device itself that performs the method.

[0106] Step S1 is an access monitoring step, in which the sensor output of the sensor within the object authentication device is evaluated in order to detect an attempt or actual behavior of physically intruding into the object authentication device or an attempt or actual behavior of locally or remotely accessing the internal control functions of the object authentication device (such as a processing device or a communication device) as a security event. If it is determined in a further step S2 that a security event is detected in step S1 (S2; yes), then the method performs a security defense step S5 as a final step, in which an error message indicating the security event is output at the user interface and / or sent via a communication link to an opposite side (such as a predetermined trust center), which opposite side can particularly communicate with Figure 1 3 or forms part of the system 3 for automatic object recognition shown in . In particular, it can be implemented as part of the data analysis module 5 or the data processing module 6, or as a separate trust center module (not shown) within the system 3, such as a security server. In addition, the object authentication device can be locked and / or the object authentication device or at least the data stored therein can be self-destructed to avoid unauthorized access to the data or any function of the object authentication device. Otherwise (S2; No), the method proceeds to the information monitoring step S3.

[0107] In the information monitoring step S3, the Figure 1 The method receives a signal from a central authority of an object authentication solution (such as a trust center providing a security server (e.g., within system 3) via a link L1 or L6 in the example above) and evaluates the signal in order to detect whether the information contained in the signal indicates a security event. If it is determined in a further step S4 that a security event is indicated in the information in step S3 (S4; Yes), the method proceeds to and performs a security defense step S5 as a final step. Otherwise (S4; No), the method proceeds to an authentication step S5.

[0108] In an authentication step S5, the user of the object authentication device is authenticated, for example via a suitable user interface (such as a keyboard for entering a password or a fingerprint sensor, etc.). If it is determined in a further step S7 that the authentication of step S6 has failed (S7; No), the method returns to step S7, or alternatively (not shown) returns to the authentication step S6. Otherwise (S7; Yes), the method proceeds to a second phase, in which the object A (or B) to be checked is detected to generate object data OD and output it to the system 3.

[0109] Thus, this second phase comprises an object detection step S8, wherein one or more distinguishing features of the inspected object (or group of objects, as the case may be) are detected by means of the sensor unit 4c of the object authentication device 4-2, and corresponding object data OD representing these features are generated.

[0110] In a subsequent communication step S9, the generated object data OD is transmitted to the object recognition system 3 via the data link L1, so as to enable it to perform the above-referenced Figure 3A / 3B. Step S9 also comprises receiving, in response to said transmission of OD and through the communication link L6 from the system 3, the digitally signed identification data comprised in the reference data subset RD1 resulting from the execution of the object identification method.

[0111] The method further comprises a reading and signature verification step S10, in which a mark M provided on the object to be authenticated (specifically, mark M for object A) is read. A ). The tag M may itself contain digitally signed identity information associated with the object A, or may in an alternative contain a pointer to a data source (such as a server in the Internet) where such information can be accessed. Step S10 also includes verifying the correctness of each of the received identification data and the digital signature of the identity information read from the tag M in order to confirm the originality of the corresponding digitally signed data.

[0112] If the signature verification in step S10 fails (S10-Yes), the method proceeds to step S12b, where authentication data to be output later are set to indicate that the authenticity of the object being checked cannot be confirmed. Otherwise (S10-No), the method proceeds to an authenticity verification step S11, in which the received identification data (in RD1) are compared with the identity information obtained from the mark M. If, as a result of this comparison, it is determined according to the corresponding predetermined matching criteria that the identification data matches the identity information (S11-Yes), then in step S12a, authentication data to be output later are set, such as to indicate that the authenticity of the object being checked is confirmed. Otherwise (S11-No), the method proceeds to step S12b, which has already been described above.

[0113] The method further comprises a storing step S13, wherein data representing the original hash value OH is received as part of the identification data (in RD1) in a block of a first blockchain of an external data ledger 9, and data representing the identity information is at least partially received in a block of a separate second blockchain, which may in particular belong to the same external data ledger 9. As part of the storing, relevant cross-blockchain pointers are added to each of the two blockchains. The cross-blockchain pointers correspond to each other in the sense that they contain data created and stored in the same authentication event. In particular, the second blockchain may be related to supply chain information such as the time, location and user identification of the current authentication event. On the other hand, the first blockchain is used to track authentication information, in particular whether the physical object carrying the mark M has been successfully authenticated as authentic (i.e., not forged or tampered with) at the current authentication event. In alternative embodiments, the blockchain may be replaced by a blockless distributed ledger or by a public-private key (PKI) environment.

[0114] The method further comprises an output step S14, which may be performed before, simultaneously with or (as shown) after the storage step S13, wherein the authentication data AD defined in step S12a or S12b, respectively, is output in a data stream or file provided, for example, on a user interface of the object authentication device 4 or at an electronic or optical interface of the object authentication device 4. The output data AD output in step S14 may also include all or part of the identification information in RD1 and / or the identity information read from the tag M. The output data AD may be used for authentication purposes in the field (e.g., at various nodes along the supply chain of the marked product), or even initially at a manufacturing or commissioning site when the physical object is initially marked, in order to verify the tag and capture its content for further use, for example for storing it in a database for subsequent authentication purposes.

[0115] Furthermore, the method may optionally comprise a security tracking step S15, in which the authentication data output in the output step S14 and optionally also the timestamp and / or the current position of each reading event of the object authentication device (each of which may be considered as security-related information SI) are sent via a communication link to a predetermined central server, which, as already described above, may, for example, form part of a trust center which in turn may again form part of the system 3 for object identification.

[0116] Figure 5 and Figure 6 The diagram illustrates additional preferred security-related aspects of the overall object authentication solution discussed above. In particular, Figure 5An alternative schematic overview of a basic embodiment of an overall object authentication solution 1 is shown, which allows verification of the identity of a marked object M at a receiver R participating in a supply chain (e.g. according to Figure 2 ) is genuine and actually provided by the assumed original manufacturer OM located upstream in the supply chain.

[0117] To this end, the original manufacturer OM is equipped with a device for applying a marking M to the product A that is subsequently shipped along the supply chain. A The device, as referenced above Figure 2 As described, it includes an object authentication device 4 (4-1) (such as Figure 1 ) or in combination with the equipment shown in . Mark M A Including equal to the system 3 in the implementation Figure 2 The training method is performed by providing a product-related reference data subset RD1 or identity information derived therefrom. The private key pair of the manufacturer OM is used by the tag M A The information represented (including in particular identity information) is digitally signed.

[0118] Thus, the object authentication device 4-2 is configured to detect the distinguishing features of the product A and access the information included in the mark M. A identity information in. In addition, the object authentication device 4-2 is configured to generate a public / private key pair of an asymmetric cryptographic system, store the private key (security key, SK) in a protected storage space of the object authentication device 4-2, and provide the public key (PUK) together with the identity information and optionally additional security-related information (such as the current time and / or location) to a central security server 10 located in a trust center, which is owned by a trusted third party. Thus, the trust center plays the role of a registration authority, in which the public keys of one or more object authentication devices 4 are registered and stored. Preferably, any communication to and from the trust center is protected by encryption, in particular to prevent "man-in-the-middle attacks". In some embodiments, the trust center 10 can form a reference Figure 1 3 and may be implemented specifically as part of its data processing platform 3 b. Alternatively, the trust center may form part of or a function of the external data ledger 9.

[0119] In order to increase the available security level, the public key can be provided to a certificate authority of a public key infrastructure (PKI), in particular to a relevant certificate authority server 12, wherein the public key is authenticated and included in a cryptographic certificate, which is available to the manufacturer OM and the verification authority (server) 11. Now, any further node in the supply chain, such as a receiver R, which is equipped with an object authentication device 4 (4-2) as described herein, can request a certificate from the verification authority 11 to use it to check the authenticity of the marked product allegedly originating from the manufacturer OM. To this end, the object authentication device 4-2 at the receiver R runs Figure 3A / 3B and / or Figure 4A / B authentication method to generate authentication data AD and based thereon confirm that OM is in fact the originator of product A or otherwise the product A or its mark M being checked A has been forged or otherwise tampered with.

[0120] The result of this authentication process (i.e., authentication data, which may optionally also include security-related information, such as the time and location of the authentication process and / or the identity of the user of the object authentication device 4-2 performing the authentication) is forwarded to and stored on the central security server 10 of the trust center. This allows centralized monitoring of the supply chain and early identification of any counterfeiting or tampering issues occurring along the supply chain. The central security server 10 may also be configured to generate or integrate tracking and tracing data reflecting the processing of product A along the supply chain based on the matching results and security-related information provided by any object authentication device 4 involved in the supply chain and make it available via a data interface API.

[0121] Figure 6 Another preferred embodiment of the object authentication solution is to use blockchain technology to securely store authentication data AD generated along the supply chain and make it available. Specifically, Figure 6 Schematically illustrates the evolution of a set of two cross-connected blockchains in parallel with the supply chain of a product A, which is marked with a corresponding mark M according to a preferred embodiment of the present object authentication solution A In particular, Figure 5 and Figure 6 The embodiments may be combined in a single solution 1.

[0122] Figure 6 The solution comprises a first blockchain BC-PUF configured to securely store authentication information, in particular by Figure 2The training process of the present invention is to derive raw hash values ​​OH from various products and make them available as described herein. In addition, a second blockchain BC-SCM is provided, which is configured to securely store and make available supply chain information (such as the serial number of the product, the date and location of the product certification and its marking, etc.). In particular, such supply chain data can be stored in the second blockchain BC-SCM in the form of an associated hash value generated from such data by applying a suitable hash function, or in addition thereto. The two blockchains BC-RD1 and BC-SCM, both configured to track the movement of products along the supply chain, link their related blocks (i.e., blocks containing data related to the same checkpoint along the supply chain) through cross-blockchain pointers, thereby providing references from and to corresponding blocks.

[0123] At the first node of the supply chain owned by OM, the original manufacturer of product A, this product A is marked with M A are marked, as described herein, e.g. Figure 2 Again, as in the above reference Figure 1 The object authentication device 4 (4-1) can be used for this purpose. In this marking process, the distinguishing characteristics of product A are identified by the object authentication device 4-2 and using Figure 3A / 3B and / or Figure 4A and 4B Detection of the authentication method of the invention is carried out by receiving from the system 3 reference data RD1 (or in an alternative an error message ERM) comprising in particular the original hash value OH of the successfully identified product A. Optionally, by comparing this original hash value OH with the marker M A The original hash value OH is then compared with the corresponding hash value provided to confirm this original hash value OH. The hash value OH is then stored as the initial original hash value in the first block of the blockchain BC-RD1 as part of the first storage transaction #1 initiated by the manufacturer OM.

[0124] Product A's mark M A The second digital signature also includes a second hash value derived from supply chain related data associated with the manufacturer OM. A This second hash value is read and stored as part of the first transaction #1 initiated by the manufacturer OM to the first block of the second supply chain BC-SCM, optionally together with further supply chain related data. Both first blocks contain data corresponding to the initial steps of the supply chain owned by the manufacturer OM, so a cross-block pointer is added in each of the two blocks to the corresponding block in the other blockchain to allow cross-references.

[0125] In the next step along the supply chain, product A arrives at a second intermediate node C, which node C may be owned, for example, by a logistics company responsible for the further transportation of the product along the supply chain. Node C is equipped with a further object authentication device 4-2-C and thus authenticates the product A by identifying the mark M on the object authentication device 4-2-C. A run Figure 3A / 3B and / or Figure 4A / 4B method to perform a check of product A. If this check confirms that manufacturer OM is the originator of product A, then a corresponding transaction #2 confirming the positive check is stored in the second block of the first blockchain BC-RD1. Otherwise, the stored transaction #2 indicates a negative result of the check, thereby indicating a negative relationship with product A or its marker M A Furthermore, an alarm or error message may be output, for example, on a user interface of the object authentication device 4-2-C, or an alarm / error message may be sent to the central trusted center 10 via a communication link to indicate said negative result.

[0126] The second block is cross-linked to the previous (i.e., first) block of the blockchain by adding the block hash of the previous block. Entering the first blockchain BC-RD1 confirms that product A was checked at node C with the corresponding result. The original raw hash value OH may remain available via the cross-link with the first block. Similarly, as in the previous node, the supply chain information is obtained from the marker M A A second digital signature of the node and additional data related to the node is generated and stored in the second blockchain BC-SCM as transaction #2. Moreover, in this second supply chain BC-SCM, the second block is cross-linked to the previous first block by storing the block hash of the previous block in the second block. Once again, a cross-blockchain pointer is added in each second block to allow cross-references between them.

[0127] In the next step along the supply chain, product A arrives at a third intermediate node d, which may be, for example, a remote logistics station not equipped with an object authentication device 4 but only with a conventional scanner 13 which can only read the mark M of product A. A Unlike in the previous nodes, at node d, only data related to the supply chain are written as transaction #3 to the third block of the second supply chain BC-SCM, similar to that in node C. However, since the scanner 13 cannot detect the distinguishing features of product A and cannot generate the relevant object data OD, no data is stored in the first supply chain BC-RD1.

[0128] Finally, in a fourth step along the supply chain, product A reaches node E, which may be, for example, the final destination or a local retailer of product A. At this node E, a similar process is performed using another object authentication device 4-2-E as at the previous node C, whereupon similar entries are added to the respective further blocks of the blockchain PC-RD1 and BC-SCM.

[0129] The two blockchains serve as a secure public ledger of all said transactions that have occurred and have been stored since said blockchains were launched. Furthermore, blockchains offer an extremely high level of integrity, since they cannot be manipulated (in practice) and therefore their use further enhances the security of the overall object authentication solution presented herein. In particular, the data stored in the two blockchains can be used to check whether the manufacturer OM is actually the originator of the product A and whether the supply chain is as expected. This check can be performed at each node OM, C, E along the supply chain that is equipped with an object authentication device 4, so it can be checked that the product A and its mark M A and access data stored in both blockchains.

[0130] While at least one exemplary embodiment of the present object authentication solution has been described above, it must be noted that a large number of variations exist thereto. Furthermore, it should be appreciated that the described exemplary embodiments merely illustrate non-limiting examples of how the present object authentication solution may be implemented and are not intended to limit the scope, application, or configuration of the apparatus and methods described herein. Rather, the foregoing description will provide those skilled in the art with a construction of at least one exemplary embodiment for implementing the solution, wherein it must be understood that various changes may be made to the functionality and apparatus of the exemplary embodiments without departing from the subject matter defined by the appended claims and their legal equivalents.

[0131] Reference numerals list

[0132] 1 Object Authentication Solution

[0133] 2 Systems for automatic object authentication

[0134] 3 Systems for Automatic Object Recognition

[0135] 3a Identification subsystem

[0136] 3b Data processing platform

[0137] 4 Object Authentication Device

[0138] 4-1 Central object authentication device of system 2

[0139] 4-2 On-site object authentication equipment

[0140] 4a Sensor unit of object authentication device

[0141] 4b Processing unit of the object authentication device

[0142] 4c Object Authentication Device Memory

[0143] 4d security equipment

[0144] 4e Security Defense Arrangement

[0145] 4f Monitoring equipment

[0146] 5 Data Analysis Module Based on Machine Learning

[0147] 5a Processing unit of the data analysis module

[0148] 5b Data analysis module memory

[0149] 6 Data Processing Module

[0150] 6a Processing unit of the data processing module

[0151] 6b Data processing module memory

[0152] 7 First Data Repository

[0153] 8 Secondary Data Repository

[0154] 9 External data ledgers, such as blockchain or blockless distributed ledgers or PKI environments

[0155] 10 Central Security Server

[0156] 11 Verification Agency (Server)

[0157] 12 Certificate Authority Server

[0158] 13 Scanner

[0159] A, B (physical) object to be authenticated

[0160] A1 Consumer goods, such as a group of pharmaceutical tablets arranged in a blister pack

[0161] A2 Product outer packaging, especially the outer packaging of blister packaging used for A1

[0162] API Trust Center Data Interface

[0163] AV1 First Auxiliary Value

[0164] AV2 Second auxiliary value

[0165] BC-RD1 The first blockchain for reference data

[0166] BC-SCN Second blockchain for supply chain data

[0167] C, D, E nodes in the supply chain

[0168] CD() data combination function

[0169] DD Distinguishing Data

[0170] ERR Error signal

[0171] ERM error messages

[0172] ID Identification data

[0173] IN Index

[0174] L1-L6 protected data links

[0175] M, M A , M B mark

[0176] MD Metadata

[0177] OD object data

[0178] ODD Original Differentiated Data

[0179] OH original hash value

[0180] OM Original Manufacturer

[0181] PUK Public Key

[0182] PRD2 Second subset of previous reference data

[0183] R Receivers, such as customers along the supply chain

[0184] RD1 The first subset of reference data, including identification data

[0185] RD2 Second subset of reference data

[0186] SI Supplementary Information

[0187] SK Security (private) key

Claims

1. A method of training a system for automatic object recognition, the method comprising: receiving object data representing one or more distinguishing characteristics of a physical object or group of physical objects; Processing the object data by means of a machine learning based object recognition process to obtain raw distinguishing data by: generating a plurality of collision-resistant virtual representations of the physical object or group of physical objects by means of performing one or more predefined modification operations, such that at least two of the virtual representations represent the same physical object or group of physical objects, but each virtual representation is in a different condition; Reference data comprising the original distinguishing data and an original hash value derived therefrom by applying a predetermined cryptographic hash function thereto is stored in one or more data repositories to which access is restricted.

2. The method of claim 1 , wherein storing the reference data comprises storing the original distinguishing data in a first data repository in the data repository and storing the identification data including the original hash value in a separate second data repository in the data repository, such that neither the first data repository nor the second data repository stores both the original distinguishing data and the corresponding original hash value.

3. A method as claimed in claim 1 or 2, wherein storing the reference data comprises storing combined data into the one or more data repositories with restricted access, the combined data being formed as a combination of the reference data and additional auxiliary values ​​defined independently of the reference data.

4. The method of claim 1 or 2, further comprising iteratively retraining the system for automatic object authentication based on additional object data representing distinguishing features of a physical object or group of physical objects, for which corresponding raw reference data had previously been generated during a previous iterative step of training the system and stored in the one or more data repositories, wherein the additional object data is used as input data for the machine learning based object recognition process.

5. The method of claim 2, wherein storing the reference data comprises one or more of the following steps: storing the reference data in at least one of the data repositories in digitally signed form; storing the identification data in a digitally signed form or causing one or more nodes of the blockchain environment to store the identification data in a digitally signed form in a block of a blockchain associated with the blockchain environment; storing or causing one or more nodes of a blockless distributed ledger environment to store the identification data in digitally signed form in at least one node of the blockless distributed ledger environment; The identification data is stored in a storage device of a public / private key environment in a digitally signed form.

6. A method as claimed in claim 1 or 2, wherein the method is performed by a system comprising two or more separate devices, which jointly perform the method and are communicatively connected to each other through one or more data links that are protected to prevent unauthorized interception.

7. A method for automatic object recognition, comprising: receiving object data representing one or more distinguishing characteristics of a physical object or group of physical objects; processing the object data by means of a machine learning based object recognition process trained by the method according to claim 1 to obtain distinguishing data representing one or more collision-resistant virtual representations of the physical object or group of physical objects; comparing at least one of the distinguishing data and a raw hash value derived therefrom by applying a predetermined cryptographic hash function thereto with corresponding reference data stored in one or more data repositories to which access is restricted; If the comparison with the reference data results in a match, digitally signed identification data comprising the hash value are output.

8. A system for automatic object recognition, the system being configured to perform the method of one or more of the preceding claims.

9. A computer program comprising instructions which, when executed on one or more processors of a system for automatic object recognition according to claim 8, cause the system to perform the method according to any one of claims 1 to 7.

10. A method for authenticating an object using an object authentication device, comprising: performing sensor-based detection of one or more distinguishing features of a physical object or group of physical objects; generating object data representing the one or more distinguishing characteristics of the physical object or group of physical objects; transmitting the object data to the system according to claim 8; receiving digitally signed identification data from said system in response to said transmitting of said object data; reading a marking provided on or in conjunction with said physical object or group of physical objects, respectively, to obtain digitally signed identity information therefrom; verifying the correctness of each of the digital signature of the identification data and the digital signature of the identity information; comparing the received identification data with the obtained identity information to verify the authenticity of the object or group of objects based on the result of the comparison; as well as Authentication information is output, the authentication information indicating whether the physical object or the physical object group is determined to be authentic according to the result of the verification.

11. The method of claim 10, wherein performing the sensor-based detection of one or more distinguishing characteristics of the physical object or group of physical objects comprises: At least one distinguishing feature is detected which is at least substantially constant in the event of changes in environmental conditions in which the at least one distinguishing feature is detected.

12. The method according to claim 10 or 11, further comprising: The original hash value or another object-specific information contained in the identification data is included in the output authentication information.

13. The method of claim 12, further comprising a storage process, wherein the storage process comprises digitally signing and storing or causing another device to store the original hash value or the other object-specific information in a digitally signed form in a block of the first blockchain or a node of the first blockless distributed ledger.

14. The method according to claim 13, wherein the storing process further comprises: reading indicia provided on or in association with the physical object or group of objects to be authenticated, or detecting one or more selected characteristics of the object or group of objects to obtain supplementary information related thereto; as well as Digitally sign and store or cause another device to store the supplemental information in digitally signed form in a block of a second blockchain separate from the first blockchain or in a node of a second blockless distributed ledger separate from the first blockless distributed ledger.

15. The method according to claim 10 or 11, further comprising one or more of the following steps: performing user authentication, and permitting or denying the user to perform object authentication with the object authentication device based on a result of the user authentication; transmitting said authentication information and / or said identification data in whole or in part and / or further information derived therefrom to the opposite side via a communication link; Capturing security-related information and sending it to the opposite side via a communication link; A security event is detected in information contained in a signal received from the opposite side via the communication link.

16. The method of claim 10 or 11, further comprising detecting one or more of the following as security events based on sensors: - Attempted or actual physical intrusion into the object authentication device; - an attempt or actual act to gain local or remote access to internal control functions of the object authentication device, wherein such access is not available to a user of the object authentication device during its normal operation.

17. The method of claim 10 or 11, further comprising performing one or more of the following security measures in response to detecting a security event: - locking the object authentication device, such as to limit or prevent its further use; - self-destruction of at least one functional part of the object authentication device or destruction of data stored therein to prevent further use or access thereof by the user; - Outputs error messages.

18. An object authentication device adapted to perform the method according to any one of claims 10 to 17.

19. A computer program comprising instructions which, when executed on one or more processors of an object authentication device according to claim 18, cause the object authentication device to perform a method according to any one of claims 10 to 17.

20. A system for automatic object authentication, comprising the system for automatic object recognition as claimed in claim 8 and the object authentication device as claimed in claim 18, wherein the system as claimed in claim 8 and the device as claimed in claim 18 are configured to jointly recognize and authenticate a physical object or a group of physical objects.

Images