Determining probability of injection attack

Abstract

Methods are provided of determining probability of injection attack during video-capture of a person-related entity using a video capturing device including a camera. These methods comprise providing to the camera, during the video-capture of the person-related entity, one or more alteration commands each instructing the camera to perform alteration of a respective setting parameter of the camera and defining a magnitude, a direction and a timing of the alteration. Such methods further comprise determining, for at least one of the performed alterations, indicia of expected impact on the capture by the performed alteration and determining the probability of injection attack depending on the determined indicia. Systems, computer systems and computer programs that are suitable for performing said methods are also provided.

Description

[0001] DETERMINING PROBABILITY OF INJECTION ATTACK

[0002] This application claims the benefit of European Patent Application EP24382826.6 filed 26 July 2024.

[0003] The present disclosure relates to methods of determining probability of injection attack, and to systems and computer programs suitable for performing said methods.

[0004] BACKGROUND

[0005] Biometric systems are being introduced as a common authentication factor in a large number of applications, e.g., physical access to buildings, access to banking applications, unlocking of smartphones and laptops, etc. As a consequence, biometrics are becoming the target of fraudsters willing to impersonate others by stealing their identities. These attempts of fraudulently impersonating others are also denominated spoofing attacks.

[0006] Injection attacks are a type of spoofing where the attacker attempts to inject a biometric evidence that replaces the original data sample captured by the capturing device. In facial biometric systems, this may involve using a virtual camera that by-passes the camera on the capturing device (e.g. the camera on a smartphone). The virtual camera may play a previously recorded video (or pre-recorded video) or may play a real-time composition or sequence of images that the fraudster can use to trick the system. In remote identity verification systems, a virtual camera may be used to play a static image or a pre-recorded video of a fraudulent identity document, such as a digitally manipulated or created (deepfake) identity document.

[0007] An object of this disclosure is to provide new methods, systems and computer programs aimed at improving prior art manners of determining probability of injection attack.

[0008] SUMMARY

[0009] In an aspect, methods are provided of determining probability of injection attack during video-capture of a person-related entity using a video capturing device including a camera. These methods (also denominated determining methods herein) comprise providing to the camera, during the video-capture of the person-related entity, one or more alteration commands each instructing the camera to perform alteration of a respective setting parameter of the camera and defining a magnitude, a direction and a timing of the alteration. Such determining methods further comprise determining, for at least one of the performed alterations, indicia of expected impact on the capture by the performed alteration (or alterations) and determining the probability of injection attack depending on the determined indicia.

[0010] Such determining methods are based on deliberately changing capturing conditions during video capture, optionally in imperceptible or almost imperceptible manner to the user of the video capturing device, or at least in a way that no suspicions nor alerts are raised from the user. These changes must have an impact on the (ongoing or completed) capture which may permit concluding whether an injection attack has been suffered. If no impact is detected by forced changes on the capture, it may mean that the camera has been by-passed and a false video content has been injected and, therefore, an injection attack has been suffered. Otherwise, if impact expected is detected on the capture, it may be concluded that no injection attack has been suffered.

[0011] The (respective) setting parameter to be altered by the camera according to any of the provided alteration commands may be one of the following setting parameters: brightness, contrast, focus, white balance, ISO, exposure, zoom, aspect ratio, resolution.

[0012] Any of the above parameters may be forced to change according to a magnitude, a direction and a timing defined by corresponding alteration command. Magnitude may refer to how much or to what extent the setting parameter is (to be) changed. Direction may indicate towards where in a N-dimensional space of possible values the setting parameter is (to be) changed. For example, in a one-dimensional space, the direction may be up or down or, in other words, increase or decrease of the parameter to be changed. Timing may refer to when the change is to be effected, with starting time and with or without ending time. A change with ending time may refer to that parameter which has been changed may be returned, at the ending time, to its previous state just before the starting time.

[0013] Expected impact by possible alterations may be predetermined experimentally or theoretically by competent staff for all the setting parameters that are alterable in the context of determining methods according to present disclosure. How indicia of expected impact may be determined is thus an aspect known by the skilled person and, therefore, no details on it are provided in present disclosure. Indicia of expected impact may include, e.g., calculation or estimation of a score quantifying whether the expected impact has been reasonably determined, so as to confirm or discard detection of the expected impact. This is also an experimental or theoretical question and, hence, no details on it are provided herein. For example, many known machine learning techniques or approaches may be used to train, e.g., a classifier or any kind of Artificial Intelligence artifact to correlate alteration commands with expected impact and, also, to determine whether the expected impact is present in the capture.

[0014] In some implementations, the person-related entity may include a person’s body part or a person’s identity document. In particular, the person’s body part may include a person’s face or a person’s eye or a person’s finger or fingerprint. Specifically, the person’s identity document may include a person’s national identity document or a person’s passport or a person’s driving license. Any of these person-related entities may be used to prove a personal identity to, e.g., gain access to or use services or places or computer platforms or networks, etc. Thus, identifying whether an injection attack has been suffered when trying to recognise a person through any of said person-related entities is an important concern nowadays. Determining methods according to present disclosure improve prior art manners of doing so.

[0015] In configurations, the providing of the one or more alteration commands to the camera may include selecting the one or more alteration commands from a population of predefined alteration commands and providing, during the video-capture of the person- related entity, the selected one or more alteration commands to the camera. In particular, the selection of the one or more alteration commands may be performed in such a manner that the alteration of the respective setting parameter to be performed by the camera varies randomly from one alteration to the next alteration or from one execution to the next execution of the method. This random variation of the alteration of the respective setting parameter may follow a distribution predefined experimentally and / or theoretically depending on feasibility of the random variation and / or its expectable impact on the capture. With all or part of said manners of selecting the one or more alteration commands, significant unpredictability is provided to determining methods and, accordingly, detectability of injection attacks is improved.

[0016] In exemplary implementations, the method may be performed (or is performable) according to an at-once analysis mode, a step-by-step analysis mode, a local analysis mode, a remote analysis mode, an at-once provision mode, a step-by-step provision mode, an internal provision mode, an external provision mode or any combination thereof.

[0017] The at-once analysis mode may refer to the indicia and the probability of injection attack being determined for all the performed alterations at once, after completion of the capture. The step-by-step analysis mode may refer to the indicia and the probability of injection attack being determined progressively as the alterations are being performed during the video-capture, on a one by one basis. In the step-by-step analysis mode, the determining of the indicia and of the probability of injection attack may be performed step by step until the determined probability denotes that injection attack has been suffered or, otherwise, until the capture is completed. Since the process of determining indicia and probability of injection attack may be terminated before completion of the capture, efficiency is provided to determining methods according to present disclosure.

[0018] The local analysis mode may refer to the indicia and the probability of injection attack being determined at the video capturing device. The remote analysis mode may refer to the indicia and the probability of injection attack being determined at a remote service or computer.

[0019] The at-once provision mode may refer to the one or more alteration commands being provided at once. The step-by-step provision mode may refer to that the one or more alteration commands are provided one by one, at different times during the video-capture. The internal provision mode may refer to that the one or more alteration commands are provided internally at the video capturing device. The external provision mode may refer to that the one or more alteration commands are provided externally from the remote service or computer.

[0020] In the external provision mode, the remote service or computer may transmit the one or more alteration commands to the video capturing device to provide them to the camera during the video-capture of the person-related entity. Each of said alteration commands may be signed using a private key before their transmission, said private key being only known by the remote service or computer. Once received, the video capturing device may validate a signature associated with the signed alteration commands to check that their transmission has been performed securely.

[0021] In a combination of the step-by-step analysis mode and the step-by-step provision mode, the determining method may include several iterations at each of which alteration command(s) are selected and provided to the camera, and impact thereof on the ongoing capture and probability of injection attack are determined. In this scenario, the alteration command(s) may be selected at a present iteration depending on the impact indicia and attack probability determined at previous iteration. For instance, if alteration command(s) selected at previous iteration with, e.g., specific magnitude and direction has / have impacted on the capture not enough conclusively, new alteration command(s) may be selected at present iteration (depending on the impact indicia and attack probability determined at previous iteration) with new magnitude and direction accordingly adjusted to reduce the uncertainty experienced at previous iteration. Determination of whether impact on the capture can be determined sufficiently conclusively may be based on, e.g., comparing obtained impact value(s) to acceptability threshold(s) or range(s) and whether impact value(s) is / are within or outside acceptability threshold(s) or range(s). Said acceptability threshold(s) or range(s) may have been predetermined theoretically and / or empirically depending on each technical scenario. These are known matters to the skilled person, so no further details are provided herein.

[0022] In a combination of the internal provision mode and the remote analysis mode, the one or more alteration commands (provided to the camera) and the capture of the person-related entity may be transmitted from the video capturing device to the remote service / computer for their processing at the remote service / computer. Such a processing may include determining, for each alteration due to the alteration commands (from the video capturing device), indicia of expected impact by the alteration on the capture (from the video capturing device) and determining the probability of injection attack depending on said determined indicia. The one or more alteration commands and the capture of the person- related entity may be signed with a private key secured and stored in the video capturing device before their transmission from the video capturing device to the remote service or computer. Once received, the remote service or computer may validate the corresponding signature associated with the signed alteration commands and the capture, so as to ensure that transmission of the alteration commands and the capture has been performed securely.

[0023] In a combination of the internal provision mode and the step-by-step analysis mode and the remote analysis mode, the video capturing device may stream, as the capture progresses and the alterations instructed by the alteration commands are being performed, the ongoing capture in sync with the alteration commands to the remote service or computer. This way, the remote service or computer may determine, as the streaming of the ongoing capture in sync with the alteration commands is being received, the indicia of expected impact and the probability of injection attack depending on the determined indicia. In this streaming approach, the ongoing capture in sync with the alteration commands may be packaged in data packets, and each of said data packets may be signed with a private key before its streaming to the remote service or computer. Such a private key may reside secured and stored in the video capturing device. As each of said streaming data packets are received by the remote service or computer, a signature associated with the data packet may be validated at the remote service or computer to guarantee that the streaming is being performed in a secure manner. The providing to the camera of the one or more alteration commands may be performed in such a manner that each of the alteration commands causes alteration of the respective setting parameter between start of the capture and start of the preview display on a screen of the video capturing device. In this manner, alterations become substantially imperceptible to the user and, therefore, carrying out an injection attack may be more challenging.

[0024] Additionally or alternatively, the providing to the camera of the one or more alteration commands may be performed in such a manner that each of the alteration commands causes alteration of the respective setting parameter during preview display of the capture along with display of an artifact or preprocessing interfering the preview display. Such an artifact or preprocessing may be or may include, e.g., a display interface element at least partially overlapping the preview display, or a distorting preprocessing of the preview display or a combination thereof. This interfering approach reinforces imperceptibility to the user of alterations deliberately performed to detect injection attack.

[0025] Identity recognition or validation methods may also be provided. These methods (also denominated identifying methods herein) may include performing any of the determining methods described in present disclosure to determine probability of injection attack during video-capture of a person-related entity and, also, performing an identity recognition or validation operation based on the captured person-related entity and the determined probability of injection attack. Such an identity recognition or validation operation may include, e.g., verifying whether the determined probability of injection attack is below or within acceptability threshold or range, in which case the identity recognition or validation operation may be qualified as valid. Otherwise, the identity recognition or validation operation may be qualified as invalid due to an injection attack being detected. In some examples, the determined probability of injection attack may be combined with the determined probability of another type of spoofing attack to calculate a global attack probability. This probability may be used in corresponding identifying method to determine validity of the identity recognition or validation operation.

[0026] In examples, the identity recognition or validation operation may be performed depending on the captured person-related entity and the determined probability of injection attack combined with other data indicating fraud probability from further method or methods of fraud detection. Said further method or methods of fraud detection may include, e.g., a method of detecting presentation attack, a method of fraud evidence integrity, or any combination thereof. In a further aspect, systems are provided for determining probability of injection attack during video-capture of a person-related entity using a video capturing device including a camera. Such systems (also denominated determining systems herein) comprise an alteration module, an indicia module and a probability module. The alteration module is configured to provide to the camera, during the video-capture of the person-related entity, one or more alteration commands each instructing the camera to perform alteration of a respective setting parameter of the camera and defining a magnitude, a direction and a timing of the alteration. The indicia module is configured to determine, for at least one of the performed alterations, indicia of expected impact on the capture by the performed alteration. The probability module is configured to determine the probability of injection attack depending on the determined indicia.

[0027] Determining systems are suitable or configured to perform determining methods proposed herein, so same or similar functional principles and advantages as the ones described with reference to determining methods may be attributed to such determining systems.

[0028] In a still further aspect, computer programs are provided comprising program instructions for causing a system or computer system to perform determining methods of determining probability of injection attack, such as those described in other parts of the disclosure. These computer programs (also denominated determining computer programs herein) may be embodied on a storage medium and / or carried on a carrier signal. Determining computer programs are suitable or configured to perform determining methods proposed herein, so same or similar functional fundamentals and advantages as the ones described with reference to determining methods may be attributed to such determining computer programs.

[0029] In a yet further aspect, computing systems are provided for determining probability of injection attack, said computing systems comprising a memory and a processor, embodying instructions stored in the memory and executable by the processor, and the instructions comprising functionality or functionalities to execute methods of determining probability of injection attack such as those described in other parts of the disclosure. Such computing systems (also denominated determining computing systems herein) are suitable or configured to perform determining methods described herein, so same or similar functional principles and advantages as the ones described with reference to determining methods may be attributed to such determining computing systems.

[0030] In some implementations, video capturing devices may also be provided including any of the determining systems or determining computing systems disclosed herein. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Non-limiting examples of the disclosure will be described in the following, with reference to the appended drawings, in which:

[0032] Figure 1 is a block diagram schematically illustrating determining systems for determining probability of injection attack, according to examples in which the determining system is fully included in the video capturing device.

[0033] Figure 2a is a block diagram schematically illustrating determining systems for determining probability of injection attack, according to examples in which the determining system is fully included in the remote service or computer.

[0034] Figures 2b and 2c are block diagrams schematically illustrating determining systems for determining probability of injection attack, according to examples in which the determining system is distributed between video capturing device and remote service or computer.

[0035] Figure 3 is a flow chart schematically illustrating determining methods of determining probability of injection attack according to examples.

[0036] DETAILED DESCRIPTION OF EXAMPLES

[0037] Determining methods disclosed herein may be performed at / by determining systems according to present disclosure in different execution modes depending on four execution dimensions. First dimension refers to whether indicia and probability of injection attack are determined at once or step by step. Second dimension refers to where indicia and probability of injection attack are determined. Third dimension refers to whether alteration commands are generated or provided at once or step by step. Fourth dimension refers to where alteration commands are generated or from where they are provided.

[0038] In the first dimension, two execution modes may be distinguished: at-once analysis mode and step-by-step analysis mode. At-once analysis mode refers to the indicia and the probability of injection attack being determined for all the performed alterations at once, after completion of the capture. Step-by-step analysis mode refers to the indicia and the probability of injection attack being determined progressively as the alterations are being performed during the video-capture, on a one by one basis. Such at-once and step-by- step analysis modes are mutually exclusive.

[0039] In the second dimension, two execution modes may be distinguished: local analysis mode and remote analysis mode. Local analysis mode refers to that the indicia and the probability of injection attack are determined at / by the video capturing device. Remote analysis mode refers to that the indicia and the probability of injection attack are determined at / by remote service or computer. Such local analysis mode and remote analysis mode are mutually exclusive.

[0040] In the third dimension, two execution modes may be distinguished: at-once provision mode and step-by-step provision mode. At-once provision mode refers to that the one or more alteration commands are provided at once. Step-by-step provision mode refers to that the one or more alteration commands are provided one by one, at different times during the video-capture. Such at-once provision mode and step-by-step provision mode are mutually exclusive.

[0041] In the fourth dimension, two execution modes may be distinguished: internal provision mode and external provision mode. Internal provision mode refers to that the one or more alteration commands are provided internally at the video capturing device. External provision mode refers to that the one or more alteration commands are provided externally from remote service or computer. Such internal provision mode and external provision mode are mutually exclusive.

[0042] Figure 1 is a block diagram schematically illustrating determining systems 100 for determining probability of injection attack during video-capture of a person-related entity using a video capturing device 101 including a camera 102, according to examples. In these examples, the determining system 100 is fully included in the video capturing device 101. Hence, the determining system 100 may perform determining methods in any of the aforementioned execution modes except in the remote analysis mode and the external provision mode.

[0043] The person-related entity may be or may include a body part of the person or an identity document of the person or any other entity related to the person from whose video capture the person is (biometrically) identifiable or recognizable. The body part of the person may be or may include person’s face or person’s eye or person’s finger or fingerprint or any other body part of the person whose video capture may be processed biometrically to identify or recognize the person. The identity document of the person may be or may include person’s national identity document or person’s passport or person’s driving license or any other legal document of the person whose video capture may be processed to identify or recognize the person. As shown in the figure, determining systems 100 may include an alteration module 103, an indicia module 104 and a probability module 105.

[0044] Alteration module 103 may be configured to provide to the camera 102, during the videocapture of the person-related entity, alteration commands 106 each instructing the camera 102 to alter a setting parameter of the camera 102. Each of the alteration commands 106 may define the setting parameter to be altered and a magnitude, a direction and a timing of the alteration. The concepts of magnitude, direction and timing are defined in some other part of the disclosure. Settings parameters alterable or to be altered by the camera 102 may include, e.g., brightness, contrast, focus, white balance, ISO, exposure, zoom, aspect ratio, resolution or any combination thereof.

[0045] The providing of the alteration commands 106 to the camera 102 may be performed by selecting alteration commands 106 from a (prestored) population of predefined alteration commands and providing the selected alteration commands 106 to the camera 102. Such a selection may be performed based on, e.g., selection rules to make the camera to perform alterations in randomly varying manner from one alteration to next alteration or from one execution to next execution of determining method. This manner of selecting the alteration commands 106 may make the determining system 100 highly unpredictable and, therefore, injection attacks may be more difficult to carry out and / or easier to detect. Such a random variation of the alterations to be performed by the camera 102 may follow a distribution predefined experimentally and / or theoretically depending, e.g., on how feasible each of the random variations is or may be and / or on its expectable impact on the capture 107. Implementation of these features is regarded as known by the skilled person and, hence, no details on it is provided herein.

[0046] Indicia module 104 may be configured to determine, for all or part of the performed alterations, indicia of expected impact on the capture by said alterations. Indicia module 104 may receive (from alteration module 103) alteration commands 108 that have been provided to the camera 102 during the video-capture and, also, may receive (from the camera 102) the capture itself 107. Indicia module 104 may then verify whether the received alteration commands 108 have influenced or not influenced as expected the capture 107. Indicia of (reasonably detectable or undetectable) impact 109 on the capture 107 by alterations due to the alteration commands 108 may thus be determined and outputted by indicia module 104. As commented in some other part of the disclosure, it is known how alteration commands and their expected impact may be theoretically and / or experimentally correlated and, therefore, how the determination of indicia of expected impact 109 may be implemented through, e.g., machine learning to train an Al artifact to implement such a functionality. Hence, no details on these matters are provided in present disclosure.

[0047] Since determining systems 100 are completely included in the video capturing device 101 , said systems 100 always operate in both local analysis mode and internal provision mode. Apart from said two permanent or inalterable execution modes, determining systems 100 may operate further according to either at-once analysis mode or step-by-step analysis mode and to either at-once provision mode or step-by-step provision mode.

[0048] In the at-once analysis mode, indicia module 104 may receive the capture 107 (from camera 102) once it has been completed and alteration commands 108 (from alteration module 103) to determine indicia of expected impact 109. Alteration commands 108 may be received all together at same time in the at-once provision mode, or progressively as they are provided to the camera 102 in the step-by-step provision mode. The determined indicia 109 may then be provided to probability module 105 which may determine the probability of injection attack 110 depending on said indicia 109. Both determinations of indicia of expected impact and of probability of injection attack may thus be performed at once or almost at once, after completion of the capture, in the at-once analysis mode.

[0049] In the step-by-step analysis mode, indicia module 104 may receive the capture 107 (from camera 102) progressively as it is being performed and alteration commands 108 (from alteration module 103). Alteration commands 108 may be received all together at same time in the at-once provision mode, or progressively as they are provided to the camera 102 in the step-by-step provision mode. Indicia module 104 may process the capture 107 along with alteration commands 108 also in progressive manner, as the capture 107 is being received, to determine indicia of expected impact 109. Probability module 105 may process the indicia 109 also in progressive manner as they are being received, to determine the probability of injection attack 110 depending on the progressively received indicia 109. This progressive determination of the probability of injection attack 110 may be terminated before completion of the capture when, e.g., the determined probability 110 denotes that injection attack has been suffered or, otherwise, until the capture 107 is completed. Both determinations of indicia of expected impact and of probability of injection attack may thus be performed progressively as the capture and alterations with expected impact are being performed, on a one by one basis, in the step-by-step analysis mode.

[0050] The providing to the camera 102 of the alteration commands 106 may be performed in such a manner that alterations presumably caused by the alteration commands 106 result imperceptible to the user of the video capturing device 101 . This imperceptibility may be achieved in different manners. For example, alteration commands 106 may be provided to the camera 102 in a manner that corresponding alterations are performed by the camera 102 between start of the capture 107 and start of preview display of the capture 107. Alternatively or additionally, alteration commands 106 may be provided to the camera 102 in a manner that corresponding alterations are performed by the camera 102 during preview display of the capture 107 along with display of an interface element at least partially overlapping the preview display, or along with distortion of the preview display, or a combination thereof.

[0051] Figure 2a is a block diagram schematically illustrating determining systems 100 for determining probability of injection attack, according to examples in which determining system 100 is fully included in remote service or computer 200. In this figure, elements that are same or similar as elements of Figure 1 may be designated with same number references. Since determining system 100 is fully included in remote service or computer 200, it may operate in any of the available execution modes except in the local analysis mode and the internal provision mode. Same or similar functional descriptions as those provided regarding determining systems 100 according to Figure 1 may be attributed to determining systems 100 according to Figure 2a, except those affected by the fact that determining system 100 is fully included in remote service or computer 200 instead of in video capturing device 101.

[0052] A difference of Figure 2a with respect to Figure 1 is that an interface module 201 in video capturing device 101 may intermediate between remotely located determining system 100 and camera 102. Any sending or transmission of data from video capturing device 101 to determining system 100, or vice versa, may be performed through a communication network (e.g., Internet) with interface module 201 acting as a bridge or intermediary. Interface module 201 may be configured to, e.g., stream the capture 107 from video capturing device 101 to remote service or computer 200 in the step-by-step analysis mode. Alteration commands 106 may be transmitted (either at once in the at-once provision mode or progressively in the step-by-step provision mode) from alteration module 103 to interface module 201 which, in turn, may provide them to the camera 102 through connection 202. Capture 107 may be provided (either at once in the at-once analysis mode or progressively in the step-by-step analysis mode) by camera 102 to interface module 201 through connection 202 and interface module 201 may transmit the capture 107 to indicia module 104 (either at once in the at-once analysis mode or progressively in the step-by-step analysis mode).

[0053] Probability of injection attack 110 may be transmitted from probability module 105 to interface module 201 which, in turn, may provide it to another module (not shown) to perform, e.g., an identity recognition or validation operation depending on the captured person-related entity 107 and the probability of injection attack 110. The remaining provisions of data may be performed in similar manner as described with reference to Figure 1. For example, alteration commands 108 may be provided from alteration module 103 to indicia module 104 locally but at the remote service or computer 200 (instead of at the video capturing device 101). Similarly, indicia of expected impact 109 may be provided from indicia module 104 to probability module 105 also locally but at the remote service or computer 200 (instead of at the video capturing device 101).

[0054] Interface module 201 may be included in determining system 100, in which case determining system 100 may be considered as a distributed determining system 100 with a first part (alteration module 103, indicia module 104 and probability module 105) at the remote service or computer 200 and a second part (interface module 201) at the video capturing device 201 .

[0055] Figure 2b is a block diagram schematically illustrating determining systems 100 for determining probability of injection attack, according to examples in which determining system 100 is distributed between video capturing device 101 and remote service or computer 200, in particular with alteration module 103 at remote service or computer 200 and with indicia module 104 and probability module 105 at video capturing device 101. In this figure, elements that are same or similar as elements of Figure 2a may be designated with same number references. Taking into account how determining system 100 is distributed between video capturing device 101 and remote service or computer 200, it may operate in any of the available execution modes except in the internal provision mode and the remote analysis mode. Same or similar functional descriptions as those provided regarding determining systems 100 according to Figure 2a may be attributed to determining systems 100 according to Figure 2b, except those affected by the fact that alteration module 103 is at remote service or computer 200 and indicia module 104 and probability module 105 are at video capturing device 101.

[0056] A difference of Figure 2b with respect to Figure 2a is that indicia module 104 and probability module 105 are at video capturing device 101. Alteration commands 106 may be sent (either at once in the at-once provision mode or progressively in the step-by-step provision mode) from alteration module 103 to camera 102 through interface module 201 in the same or similar manner as described in relation to Figure 2a. Alteration module 103 may send (either at once in the at-once provision mode or progressively in the step-by- step provision mode) alteration commands 108 to indicia module 104 through interface module 201 as illustrated in the figure (instead of locally at remote service or computer 200). Camera 102 may provide (either at once in the at-once analysis mode or progressively in the step-by-step analysis mode) the capture 107 to indicia module 104 locally at the video capturing device 101 (instead of from video capturing device 101 to remote service or computer 200). Indicia module 104 may provide (either at once in the at-once analysis mode or progressively in the step-by-step analysis mode) the determined indicia 109 to probability module 105 locally at the video capturing device 101 (instead of at the remote service or computer 200). Probability module 105 may output probability of injection attack 110 locally at the video capturing device 101 (instead of from the remote service or computer 200 to video capturing device 101).

[0057] Determining system 100 according to Figure 2b may be considered as a distributed determining system 100 with one part (alteration module 103) at the remote service or computer 200 and another part (interface module 201 , indicia module 104 and probability module 105) at the video capturing device 201.

[0058] Figure 2c is a block diagram schematically illustrating determining systems 100 for determining probability of injection attack, according to examples in which determining system 100 is distributed between video capturing device 101 and remote service or computer 200, specifically with indicia module 104 and probability module 105 at remote service or computer 200 and with alteration module 103 at video capturing device 101. In this figure, elements that are same or similar as elements of Figure 2a may be designated with same number references. Taking into account how determining system 100 is distributed between video capturing device 101 and remote service or computer 200, it may operate in any of the available execution modes except in the external provision mode and the local analysis mode. Same or similar functional descriptions as those provided regarding determining systems 100 according to Figure 2a may be attributed to determining systems 100 according to Figure 2c, except those affected by the fact that indicia module 104 and probability module 105 are at remote service or computer 200 and alteration module 103 is at video capturing device 101.

[0059] A difference of Figure 2c with respect to Figure 2a is that alteration module 103 is at video capturing device 101. Taking this into account, alteration commands 106 may be provided (either at once in the at-once provision mode or progressively in the step-by-step provision mode) by alteration module 103 to camera 102 locally at the video capturing device 101 (instead of from remote service or computer 200 to video capturing device 101). Alteration module 103 may send (either at once in the at-once provision mode or progressively in the step-by-step provision mode) alteration commands 108 to indicia module 104 through interface module 201 as illustrated in the figure (instead of locally at remote service or computer 200). Camera 102 may provide (either at once in the at-once analysis mode or progressively in the step-by-step analysis mode) the capture 107 to indicia module 104 through interface module 201 in same or similar manner as described in relation to Figure 2a. Indicia module 104 may provide (either at once in the at-once analysis mode or progressively in the step-by-step analysis mode) the determined indicia 109 to probability module 105 locally at the remote service or computer 200 in the same or similar manner as described in relation to Figure 2a. Probability module 105 may send probability of injection attack 110 to video capturing device 101 through interface module 201 in the same or similar manner as described in relation to Figure 2a.

[0060] Determining system 100 according to Figure 2b may be considered as a distributed determining system 100 with one part (indicia module 104 and probability module 105) at the remote service or computer 200 and another part (interface module 201 and alteration module 103) at the video capturing device 201.

[0061] In any or some of the determining systems disclosed herein, provisions / transmissions of data, either locally (within video capturing device 101 or within remote service or computer 200) or remotely (from video capturing device 101 to remote service or computer 200, or vice versa), may be performed in synchronized manner. For instance, transmissions / provisions of alteration commands 108 and of the capture 107 may be synchronized with each other to make that each command 108 is received at indicia module 104 sufficiently earlier than pertinent part of the capture 107 presumably impacted by alteration caused by said alteration command 108. These syncs may be implemented in many manners that are known in the technical field at hand and, therefore, no details on this matter are provided herein. In some examples, as described in other parts of the disclosure, the capture 107 may be streamed in sync with alteration commands 108 such that each of the commands 108 is received at indicia module 104 sufficiently earlier than corresponding part of the capture 107 presumably impacted by alteration provoked by said alteration command 108.

[0062] In any or some of the determining systems disclosed herein, provisions / transmissions of any kind of data from video capturing device 101 to remote service or computer 200, or vice versa, may be performed once the data has been signed with a private key, thereby associating a secure signature to the data to be transmitted. If transmission is performed by / from the video capturing device 101 , such a private key may reside secured and stored in the video capturing device 101. Upon reception of the signed data, the remote service or computer 200 may validate its associated signature to check that transmission has been performed in secure manner. If transmission is performed by / from the remote service or computer 200, such a private key may reside secured and stored in the remote service or computer 200. Upon reception of the signed data, the video capturing device 101 may validate its associated signature to check that transmission has been performed securely. As described in other parts of the disclosure, in stream-based examples, streaming may be performed by progressively packaging ongoing capture 107, in sync with (or without) alteration commands 108, in data packets each of which may be signed as explained in this paragraph before its streaming.

[0063] In examples in which key(s) is / are stored in the video capturing device 101 , to either sign data to be sent or validate signature of signed received data, the storing and use of the key(s) at the video capturing device 101 may be avoided to prevent that corresponding app running in the device 101 that uses the key(s) can be cracked and the key(s) found out by an attacker.

[0064] Instead of attributing to the video capturing device 101 (or app running therein) key-related functions (of, e.g., signing or signature validation), validations aimed at securing transmissions may be all performed at the remote service or computer 200. A manner of doing so may be based on, e.g., JSON Web Signature (abbreviated JWS) and corresponding JWS token(s).

[0065] For example, client app running at the video capturing device 101 may request to the remote service or computer 200 a JWS token including a challenge. Upon reception of the JWS token, the data to be sent may be generated or processed as required (e.g., the capture may be performed) at the video capturing device 101 . Once such a generation or processing has been completed (e.g., the capture has been generated), resulting data (e.g., the capture 107) may be sent to the remote service or computer 200 along with the previously received JWS token. With such a sending to the remote service or computer 200 of the generated or processed data (e.g., the capture 107) and the JWS token together, the validation of the generated or processed data (e.g., the capture 107) is delegated to the remote service or computer 200.

[0066] With this JWS-based solution, the client app running at the video capturing device 101 does not need any key, provided that the JWS data is sent unencrypted. The main concept behind this solution is that the remote service or computer 200 must receive something that only the remote service or computer 200 has generated, i.e., the JWS token. The transmission, from the video capturing device 101 to the remote service or computer 200, is then validated based on that the JWS token in the received data must have been generated by / at the remote service or computer 200. A JWS expiration time (of, e.g., N minutes) may exist or may be defined at the remote service or computer 200, after which or once it has elapsed, the JWS token is rejected and the transmission is thus determined invalid. The transmission may therefore need, to be determined as valid, that the JWS token is exactly the one that was generated previously by / at the remote service or computer 200 and that its JWS expiration time has not elapsed / expired.

[0067] In implementations described in other parts of the disclosure, the probability of injection attack 110 determined by the probability module 105 is shown outputted by said module 105 either with no destination module (e.g., Figures 1 and 2b) or sent to interface module 201 within the video capturing device 101 (e.g., Figures 2a and 2c). Alternatively to said destinations (i.e., either no module or interface module 201), the probability of injection attack 110 may be sent to a validation module (not shown) along with the capture 107. The validation module may reside at the remote service / computer 200 or at a further system or server (not shown). The validation module may be configured to authenticate or validate or recognize person’s identity depending on the capture 107 and the probability of injection attack 110. Such a process of validating the person’s identity may thus be performed more securely in comparison to the process being performed at the video capturing device (e.g., smartphone) 101. Performing the validation process (of validating the person’s identity) at the video capturing device (e.g., smartphone) may generate some evidence(s) that an attacker could use to learn how to bypass security arrangements. However, performing the validation process at the remote service / computer 200 or the further system or server, attacks may be much more difficult to design and carry out. Additionally, stronger security conditions may be defined at the remote service / computer 200 or further system or server in comparison to the video capturing device (e.g., smartphone) 101.

[0068] The validation module may perform the validation process by, e.g., verifying whether the probability of injection attack 110 is below or within acceptability threshold or range, in which case the person’s identity validation may be qualified as “genuine”. Otherwise, the person’s identity validation may be qualified as “attack” because an injection attack has been detected. In case of “genuine”, the validation module may determine whether the person’s identity is valid or invalid depending on the capture 107. In case of “attack”, the validation module may determine that the person’s identity is invalid because an injection attack has been suffered.

[0069] Data representing the validity or invalidity output by the validation module may be sent to an end system to cause operational response to the determined validity or invalidity. The end system may include the video capturing device (e.g., smartphone, biometric access terminal) 101 to notify its user whether the person’s identity has been authenticated or validated or recognized successfully (validity) or unsuccessfully (invalidity). The end system may further or only include an access system configured to provide (in case of validity) or deny (in case of invalidity) to the identified person some kind of access or authorization to do something. Such an access may be physical access to a building or room or space, access to a banking application, unlocking of a smartphone or laptop or any kind of computer or computing device, etc.

[0070] As used herein, the term “module” may be understood to refer to software, firmware, hardware and / or various combinations thereof. It is noted that the modules are exemplary. The modules may be combined, integrated, separated, and / or duplicated to support various applications. Also, a function described herein as being performed by a particular module may be performed by one or more other modules and / or by one or more other devices instead of or in addition to the function performed by the described particular module.

[0071] The modules may be implemented across multiple devices, associated or linked to corresponding determining systems proposed herein, and / or to other components that may be local or remote to one another. Additionally, the modules may be moved from one device and added to another device, and / or may be included in both devices, associated to corresponding determining systems proposed herein. Any software implementations may be tangibly embodied in one or more storage media, such as, e.g., a memory device, a floppy disk, a compact disk (CD), a digital versatile disk (DVD), or other devices that may store computer code.

[0072] Determining systems according to present disclosure may be implemented by computing means, electronic means or a combination thereof. The computing means may be a set of instructions (e.g., a computer program) and then determining systems may comprise a memory and a processor, embodying said set of instructions stored in the memory and executable by the processor. These instructions may comprise functionality or functionalities to execute corresponding determining methods such as those described in other parts of the disclosure.

[0073] In determining systems implemented only by electronic means, a controller of the system may be, for example, a CPLD (Complex Programmable Logic Device), an FPGA (Field Programmable Gate Array) or an ASIC (Application-Specific Integrated Circuit). In determining systems that are a combination of electronic and computing means, the computing means may be a set of instructions (e.g., a computer program) and the electronic means may be any electronic circuit capable of implementing corresponding steps of determining methods proposed herein, such as those described in other parts of the disclosure.

[0074] The computer program(s) may be embodied in / on a storage medium (for example, a CD- ROM, a DVD, a USB drive, a computer memory or a read-only memory) or carried on a carrier signal (for example, on an electrical or optical carrier signal).

[0075] The computer program(s) may be in the form of source code, object code, a code intermediate source and object code such as in partially compiled form, or in any other form suitable for use in implementing determining methods according to present disclosure. The carrier may be any entity or device capable of carrying the computer program(s).

[0076] For example, the carrier may comprise a storage medium, such as a ROM, for example a CD ROM or a semiconductor ROM, or a magnetic recording medium, for example a hard disk. Further, the carrier may be a transmissible carrier such as an electrical or optical signal, which may be conveyed via electrical or optical cable or by radio or other means.

[0077] When the computer program(s) is / are embodied in a signal that may be conveyed directly by a cable or other device or means, the carrier may be constituted by such cable or other device or means. Alternatively, the carrier may be an integrated circuit in which the computer program(s) is / are embedded, the integrated circuit being adapted for performing, or for use in the performance of, determining methods proposed herein.

[0078] Figure 3 is a flow chart schematically illustrating determining methods of determining probability of injection attack, according to examples. Since determining methods according to Figure 3 are performable by or at determining systems according to Figure 1 and Figures 2a - 2c, number references from said Figures 1 and 2a - 2c may be reused in following description of Figure 3. As commented in other parts of the disclosure, such determining methods are performable or performed during video-capture of a person- related entity using a video capturing device 101 including a camera 102.

[0079] Determining methods may be initiated (e.g., at block 300) upon detection of a starting condition such as, e.g., a user request to start the determining method. Determining methods may further include (e.g., at method-block 301) providing to the camera 102, during the video-capture of the person-related entity, alteration commands

[0080] 106 each instructing the camera 102 to perform alteration of a setting parameter of the camera 102. Each of the alteration commands 106 may define the setting parameter to be altered, and a magnitude, a direction and a timing of the alteration. In determining systems 100 with alteration module 103 at remote service or computer 200, this provision to the camera 102 may be performed through interface module 201 as explained in other parts of the disclosure. This altering functionality implemented or implementable at method-block 301 may be performed by, e.g., alteration module 103, possibly in cooperation with interface module 201 , as previously described with reference to Figures 1 and 2a - 2c. Functional details and considerations explained about alteration module 103 (and, in some examples, interface module 201) may thus be similarly attributed or attributable to method-block 301 .

[0081] Determining methods may still further include (e.g., at method-block 302) determining, for at least one of the performed alterations, indicia of expected impact 109 on the capture

[0082] 107 by the performed alteration. This indicia functionality implemented or implementable at method-block 302 may be performed by, e.g., indicia module 104 previously described with reference to Figures 1 and 2a - 2c. Functional details and considerations explained about indicia module 104 may thus be similarly attributed or attributable to method-block 302.

[0083] Determining methods may yet further include (e.g., at method-block 303) determining the probability of injection attack 110 depending on the determined indicia 109. This probability functionality implemented or implementable at method-block 303 may be performed by, e.g., probability module 105 previously described with reference to Figures 1 and 2a - 2c. Functional details and considerations explained about probability module 105 may thus be similarly attributed or attributable to method-block 303.

[0084] Determining methods may furthermore include (at a method-block not shown) evaluating the determined probability of injection attack 110 to conclude whether injection attack has been or has not been suffered. Said evaluation may include, e.g., verifying whether the determined probability of injection attack 110 is above or below (or equal to) a predefined attack threshold. If above (or equal), it may be concluded that injection attack has been suffered. If below (or equal), it may be concluded that injection attack has not been suffered. Determining methods may additionally include (e.g., at decision block 304) verifying whether an ending condition occurs or is satisfied. In case of positive or true result Y of said verification, determining method may proceed to method-block 305 to terminate execution of the method. Otherwise N, determining method may loop back to previous method-block 301 to perform a new iteration to, e.g., determine probability of injection attack in a new capture as described before regarding combination of altering function 301 , indicia function 302 and probability function 303.

[0085] Although only a number of examples have been disclosed herein, other alternatives, modifications, uses and / or equivalents thereof are possible. Furthermore, all possible combinations of the described examples are also covered. Thus, the scope of the disclosure should not be limited by particular examples, but it should be determined only by a fair reading of the claims that follow.

Claims

CLAIMS1 . Method of determining probability of injection attack during video-capture of a person- related entity using a video capturing device including a camera, the method comprising: providing to the camera, during the video-capture of the person-related entity, one or more alteration commands each instructing the camera to perform alteration of a respective setting parameter of the camera and defining a magnitude, a direction and a timing of the alteration; determining, for at least one of the performed alterations, indicia of expected impact on the capture by the performed alteration; and determining the probability of injection attack depending on the determined indicia.

2. Method of determining probability of injection attack according to claim 1 , wherein the respective setting parameter to be altered by the camera is one of following setting parameters: brightness, contrast, focus, white balance, ISO, exposure, zoom, aspect ratio, resolution.

3. Method of determining probability of injection attack according to any of claims 1 or 2, wherein the person-related entity includes a person’s body part or a person’s identity document.

4. Method of determining probability of injection attack according to claim 3, wherein the person’s body part includes person’s face or person’s eye or person’s finger or fingerprint.

5. Method of determining probability of injection attack according to claim 3, wherein the person’s identity document includes person’s national identity document or person’s passport or person’s driving license.

6. Method of determining probability of injection attack according to any of claims 1 to 5, wherein the providing of the one or more alteration commands to the camera includes selecting the one or more alteration commands from a population of predefined alteration commands; and providing to the camera, during the video-capture of the person-related entity, the selected one or more alteration commands.

7. Method of determining probability of injection attack according to claim 6, wherein the selection of the one or more alteration commands is performed in such a manner that the alteration of the respective setting parameter to be performed by the camera variesrandomly from one alteration to next alteration or from one execution to next execution of the method.

8. Method of determining probability of injection attack according to claim 7, wherein the random variation of the alteration of the respective setting parameter follows a distribution predefined experimentally and / or theoretically depending on feasibility of the random variation and / or its expectable impact on the capture.

9. Method of determining probability of injection attack according to any of claims 1 to 8, wherein the method is performable according to at least one of following modes: at-once analysis mode which refers to determining the indicia and the probability of injection attack for all the performed alterations at once, after completion of the capture, step-by-step analysis mode which refers to determining the indicia and the probability of injection attack as the alterations are being performed during the video-capture, on a one-by-one basis, local analysis mode which refers to determining the indicia and the probability of injection attack at / by the video capturing device, remote analysis mode which refers to determining the indicia and the probability of injection attack at / by a remote service or computer, at-once provision mode which refers to provisioning the one or more alteration commands at once, step-by-step provision mode which refers to provisioning the one or more alteration commands one by one, at different times during the video-capture, internal provision mode which refers to provisioning the one or more alteration commands internally at the video capturing device, and external provision mode which refers to provisioning the one or more alteration commands externally from the remote service or computer.

10. Method of determining probability of injection attack according to claim 9 wherein, in the step-by-step analysis mode, the determining of the indicia and of the probability of injection attack are being performed until the determined probability denotes that injection attack has been suffered or, otherwise, until the capture is completed.

11. Method of determining probability of injection attack according to any of claims 9 or 10 wherein, in a combination of the internal provision mode and the remote analysis mode, the video capturing device transmits the one or more alteration commands provided to the camera and the capture of the person-related entity to the remote service or computer, andthe remote service or computer determines, for each alteration due to the transmitted alteration commands, indicia of expected impact by the alteration on the transmitted capture and determines the probability of injection attack depending on the determined indicia.

12. Method of determining probability of injection attack according to claim 11 , wherein the one or more alteration commands and the capture of the person-related entity are signed with a private key before their transmission to the remote service or computer, the private key being secured and stored in the video capturing device.

13. Method of determining probability of injection attack according to any of claims 9 to 12 wherein, in a combination of the internal provision mode and the step-by-step analysis mode and the remote analysis mode, the video capturing device streams, as the capture progresses and the alterations instructed by the alteration commands are being performed, the ongoing capture in sync with the alteration commands to the remote service or computer, and the remote service or computer determines, as the streaming of the ongoing capture in sync with the alteration commands is being received at the remote service or computer, the indicia of expected impact and the probability of injection attack depending on the determined indicia.

14. Method of determining probability of injection attack according to claim 13, wherein the ongoing capture in sync with the alteration commands are packaged in data packets and each of said data packets is signed with a private key before its streaming to the remote service or computer, the private key being secured and stored in the video capturing device.

15. Method of determining probability of injection attack according to any of claims 9 or 10 wherein, in the external provision mode, the remote service or computer transmits the one or more alteration commands to the video capturing device to provide them to the camera during the video-capture of the person-related entity.

16. Method of determining probability of injection attack according to claim 15, wherein the one or more alteration commands are signed with a private key that is only known by the remote service or computer before its / their transmission to the video capturing device.

17. Method of determining probability of injection attack according to claim 15, whereinthe one or more alteration commands are transmitted, by the remote service or computer to the video capturing device, encoded and signed within a JWS token, and the video capturing device provides to the camera, during the video-capture of the person-related entity, said alteration commands and retransmits the JWS token along with the performed video-capture to the remote service or computer for its validation.

18. Method of determining probability of injection attack according to any of claims 1 to 17, wherein the providing to the camera of the one or more alteration commands is performed in such a manner that each of the alteration commands causes alteration of the respective setting parameter between start of the capture and start of preview display of the capture, such that said alteration results substantially imperceptible to the user.

19. Method of determining probability of injection attack according to any of claims 1 to 17, wherein the providing to the camera of the one or more alteration commands is performed in such a manner that each of the alteration commands causes alteration of the respective setting parameter during preview display of the capture along with display of an interface element at least partially overlapping the preview display, or along with distortion of the preview display or a combination thereof, such that said alteration results substantially imperceptible to the user.

20. Identity recognition or validation method including: performing a method according to any of claims 1 to 19 of determining probability of injection attack during video-capture of a person-related entity; and performing an identity recognition or validation operation depending on the captured person-related entity and the determined probability of injection attack.

21. Identity recognition or validation method according to claim 20, wherein the performing of the identity recognition or validation operation depending on the captured person- related entity and the determined probability of injection attack includes performing the identity recognition or validation operation depending on the captured person-related entity and a combination of the determined probability of injection attack and other data indicating fraud probability from further method or methods of fraud detection.

22. Identity recognition or validation method according to claim 21 , wherein the further method or methods of fraud detection includes at least one of a method of detecting presentation attack, a method of fraud evidence integrity.

23. Identity recognition or validation method according to any of claims 20 to 22, wherein the performing of the identity recognition or validation operation depending on the captured person-related entity and the determined probability of injection attack includes verifying whether the determined probability of injection attack is below or within acceptability threshold or range, in which case the identity recognition or validation operation is qualified as performable depending on the captured person-related entity and, otherwise, the identity recognition or validation operation is qualified as unperformable due to injection attack.

24. Computer program comprising program instructions for causing a computing system to perform a method according to any of claims 1 to 23.

25. System for determining probability of injection attack during video-capture of a person- related entity using a video capturing device including a camera, the system comprising: an alteration module configured to provide to the camera, during the video-capture of the person-related entity, one or more alteration commands each instructing the camera to perform alteration of a respective setting parameter of the camera and defining a magnitude, a direction and a timing of the alteration; an indicia module configured to determine, for at least one of the performed alterations, indicia of expected impact on the capture by the performed alteration; and a probability module configured to determine the probability of injection attack depending on the determined indicia.

26. Video capturing device including a system according to claim 25 for determining probability of injection attack.

Images