Identity verification method and device

Abstract

The invention provides an identity verification method and device, and relates to the technical field of data processing, and the method comprises the steps: obtaining the eye and head movement of a first user in each glancing time period to form a glancing feature set, and the eye and head movement in each staring time period to form a staring feature set in the process of inputting a password through the eye and head cooperative movement of the first user; the glancing feature set and the staring feature set are input into corresponding pre-trained random forest classifiers, output results of the random forest classifiers are summarized, a first authentication result is obtained through a voting mechanism, and the random forest classifiers are constructed according to the corresponding feature sets obtained when a target user registers; matching authentication is carried out according to a password indicated by the eye movement track and a password when the target user registers, and a second authentication result is obtained; and determining whether the identity verification of the first user is successful or not according to the first authentication result and the second authentication result. Therefore, the time cost can be reduced, and the user experience and safety can be improved.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and in particular, to an identity authentication method and device. Background Art

[0002] Currently, the need for user identity authentication has become increasingly important. There are some inconveniences in current authentication methods. For example, password authentication is vulnerable to shoulder surfing attacks; biometric authentication (such as fingerprint, iris) may raise users' concerns about privacy issues, and manually entering passwords or fingerprints requires manual operations, especially when there are items in hand and the items need to be put down, which affects the user experience. Summary of the Invention

[0003] In view of this, this application provides an identity authentication method and device, aiming to reduce costs and improve user experience and security.

[0004] In a first aspect, this application provides an identity authentication method, and the method includes:

[0005] During the process of obtaining the password input by the coordinated movement of the eyes and head of the first user, the eye and head movements in each saccade time period form a saccade feature set, and the eye and head movements in each fixation time period form a fixation feature set;

[0006] Input the saccade feature set and the fixation feature set into the corresponding pre-trained random forest classifier, summarize the output results of each random forest classifier, and obtain a first authentication result through a voting mechanism. The random forest classifier is a classifier constructed according to the corresponding feature set obtained during the registration of the target user;

[0007] Perform matching authentication on the password indicated by the eye movement trajectory and the password of the target user during registration to obtain a second authentication result;

[0008] Determine whether the first user is successfully authenticated according to the first authentication result and the second authentication result.

[0009] Optionally, during the process of obtaining the password input by the coordinated movement of the eyes and head of the first user, the eye and head movements in each saccade time period form a saccade feature set, and the eye and head movements in each fixation time period form a fixation feature set, including:

[0010] Collect the eye movement trajectory and head movement trajectory during the process of the first user inputting the password by the coordinated movement of the eyes and head;

[0011] Generate an eye time series according to the eye movement trajectory, and generate a head time series according to the head movement trajectory;

[0012] Extract the saccade feature sets of the eye time series and the head time series in each saccade time period; extract the fixation feature sets of the eye time series and the head time series in each fixation time period.

[0013] Optionally, when the eyes and head move coordinately to input a password, it includes multiple saccade time periods and multiple fixation time periods, and the saccade time periods and fixation time periods are arranged alternately.

[0014] Optionally, the step of extracting the saccade feature sets of the eye time series and the head time series in each saccade time period; extracting the fixation feature sets of the eye time series and the head time series in each fixation time period includes:

[0015] Extract the eye saccade features corresponding to each saccade time period and the eye fixation features corresponding to each fixation time period in the eye time series; extract the head saccade features corresponding to each saccade time period and the head fixation features corresponding to each fixation time period in the head time series;

[0016] Combine the corresponding eye saccade features and head saccade features in the same saccade time period into a saccade feature set; combine the corresponding eye fixation features and head fixation features in the same fixation time period into a fixation feature set.

[0017] Optionally, the training method of the random forest classifier includes:

[0018] When collecting the target user's registration, collect the eye movement trajectory and head movement trajectory when the eyes and head move coordinately to input a password, and generate the eye registration time series of the eye movement trajectory and the head registration time series of the head movement trajectory;

[0019] Extract the saccade registration feature sets in each saccade time period from the eye registration time series and the head registration time series, and extract the fixation registration feature sets in each fixation time period from the eye registration time series and the head registration time series;

[0020] Perform data augmentation on the saccade registration feature sets of each saccade time period by adding Gaussian noise to obtain a saccade augmented feature set; perform data augmentation on the fixation registration feature sets of each fixation time period by adding Gaussian noise to obtain a fixation augmented feature set;

[0021] Construct the random forest classifier corresponding to the saccade time period based on the saccade augmented feature sets corresponding to each saccade time period, and each saccade time period corresponds to a random forest classifier;

[0022] Construct the random forest classifier corresponding to the gaze time period based on the gaze enhancement feature set corresponding to each gaze time period, and each gaze time period corresponds to a random forest classifier.

[0023] Optionally, the inputting the saccade feature set and the gaze feature set into the corresponding pre-trained random forest classifier includes:

[0024] Input the saccade feature set corresponding to each saccade time period into the pre-trained random forest classifier corresponding to the saccade time period, and output a first authentication sub-result; wherein, when the probability that the random forest classifier analyzes that the first user is the target user based on the saccade feature set exceeds a first threshold, the first authentication sub-result is authentication passed; when the probability that the random forest classifier analyzes that the first user is the target user based on the saccade feature set does not exceed the first threshold, the first authentication sub-result is authentication failed;

[0025] Input the gaze feature set corresponding to each gaze time period into the random forest classifier corresponding to the gaze time period, and output a second authentication sub-result; wherein, when the probability that the random forest classifier analyzes that the first user is the target user based on the gaze feature set exceeds a second threshold, the second authentication sub-result is authentication passed; when the probability that the random forest classifier analyzes that the first user is the target user based on the gaze feature set does not exceed the second threshold, the second authentication sub-result is authentication failed.

[0026] Optionally, the collecting the eye movement trajectory and the head movement trajectory during the process of the first user inputting the password with the coordinated movement of the eyes and the head includes:

[0027] When it is detected that the first user blinks unnaturally, start to execute the collection of the eye movement trajectory and the head movement trajectory during the coordinated movement of the eyes and the head, and end the collection when it is detected again that the first user blinks unnaturally.

[0028] Optionally, the user authentication interface layouts a 3×3 grid array at a 60-degree field of view angle to facilitate the user to input the password by gazing at the 3×3 grid array.

[0029] Optionally, the determining whether the first user passes the authentication according to the first authentication result and the second authentication result includes:

[0030] When both the first authentication result and the second authentication result are authentication passed, it is determined that the first user's identity verification is successful.

[0031] In a second aspect, the present application provides an identity verification device, and the device includes:

[0032] An acquisition unit, configured to acquire a saccade feature set formed by the eye and head movements during each saccade time period and a fixation feature set formed by the eye and head movements during each fixation time period when the first user inputs a password through the coordinated movement of the eyes and head.

[0033] A classification unit, configured to input the saccade feature set and the fixation feature set into corresponding pre-trained random forest classifiers, summarize the output results of each random forest classifier, and obtain a first authentication result through a voting mechanism. The random forest classifier is a classifier constructed according to the corresponding feature set obtained when the target user registers.

[0034] A judgment unit, configured to perform matching authentication on the password indicated by the eye movement trajectory and the password of the target user when registering, and obtain a second authentication result.

[0035] A result unit, configured to determine whether the first user is successfully authenticated according to the first authentication result and the second authentication result.

[0036] In a third aspect, the present application provides a device, which includes a memory and a processor. The memory is used to store instructions or codes, and the processor is used to execute the instructions or codes so that the device executes an authentication method according to any one of the foregoing first aspects.

[0037] In a fourth aspect, the present application provides a computer storage medium, in which codes are stored. When the codes are run, the device running the codes implements an authentication method according to any one of the foregoing first aspects.

[0038] The present application provides an authentication method and apparatus. When executing the method, during the process of obtaining the password by the coordinated movement of the eyes and head of the first user, the eye and head movements during each saccade time period form a saccade feature set, and the eye and head movements during each fixation time period form a fixation feature set; the saccade features and the fixation features are input into the corresponding trained random forest classifier, and the output results of each random forest classifier are summarized, and the first authentication result is obtained through a voting mechanism. The random forest classifier is a classifier constructed according to the corresponding feature set obtained during the registration of the target user; the password indicated by the eye movement trajectory is matched and authenticated with the password of the target user during registration to obtain a second authentication result; it is determined whether the first user is successfully authenticated according to the first authentication result and the second authentication result. In this way, the coordinated movement of the eyes and head to input the password combines two biometric information of eye movement and head movement, and a decision analysis is performed through a random forest classifier to determine whether it is the registered target user. At the same time, the password formed by the coordinated movement of the head and eyes is used for password verification. In this way, multiple information verifications are performed by combining the password and two biometric information, improving security, not being vulnerable to shoulder surfing attacks, and not requiring the user to provide information with unique identification such as fingerprints and irises, and only requiring the coordinated movement of the head and eyes without manual operation, enhancing the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] To more clearly illustrate the technical solutions in the embodiments or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0040] Figure 1 It is a schematic flowchart of an authentication method provided by an embodiment of the present application;

[0041] Figure 2 It is a schematic diagram of the process of password input provided by an embodiment of the present application;

[0042] Figure 3 It is a schematic diagram of the extraction of head-eye coordination movement features provided by an embodiment of the present application;

[0043] Figure 4 It is a schematic flowchart of the pre-training of a random forest classifier based on the biometric characteristics of the registered target user provided by an embodiment of the present application;

[0044] Figure 5 It is a schematic structural diagram of an authentication device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] With the rapid development of virtual reality (VR) technology in fields such as gaming, e-commerce, and social networks, using traditional input devices (such as keyboards and touchpads) to enter passwords (such as character passwords and pattern passwords) will destroy the immersion of VR and is vulnerable to shoulder surfing attacks. Using biometric recognition (such as fingerprints and irises) is likely to raise users' privacy concerns. In addition, manually operating to enter passwords or fingerprints affects users' VR immersive experience.

[0046] Based on the above problems, this application provides an authentication method and device. The combination of eye and head movement for password input combines two types of biometric information, eye movement and head movement, and uses a random forest classifier for decision analysis to determine whether the user is a registered user. At the same time, password verification is also performed based on the password formed by the coordinated movement of the head and eyes. In this way, the combination of two types of biometric information and password matching is achieved, and multiple information combinations are used for verification, improving security, being less vulnerable to shoulder surfing attacks, and only requiring the coordinated movement of the head and eyes without manual operation, thus enhancing the user experience.

[0047] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with the relevant laws, regulations, and standards of relevant countries and regions.

[0048] In order to be able to understand the characteristics and technical content of the embodiments of the present disclosure in more detail, the implementation of the embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. The accompanying drawings are only for reference and explanation purposes and are not used to limit the embodiments of the present disclosure. In the following technical description, for the sake of explanation, multiple details are provided to provide a full understanding of the disclosed embodiments. However, one or more embodiments can still be implemented without these details. In other cases, well-known structures and devices can be shown in a simplified manner to simplify the drawings.

[0049] In the embodiments of the present disclosure, the terms "first", "second", etc. in the specification, claims, and the above-mentioned drawings are used to distinguish similar objects and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present disclosure described here can be implemented. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion.

[0050] Unless otherwise stated, the term "plurality" means two or more.

[0051] In the embodiments of the present disclosure, the character " / " indicates that the objects before and after are in an "or" relationship. For example, A / B means: A or B.

[0052] The term "and / or" describes the relationship between objects and indicates that there can be three relationships. For example, A and / or B means: A or B, or A and B.

[0053] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the following will clearly and completely describe the technical solutions in the embodiments of this application in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this application.

[0054] See Figure 1 , Figure 1 is a schematic flowchart of an identity authentication method provided for an embodiment of this application. An identity authentication method includes:

[0055] S101. During the process of obtaining the input password through the coordinated movement of the eyes and head of the first user, a saccade feature set formed by the eye and head movements in each saccade time period, and a fixation feature set formed by the eye and head movements in each fixation time period are obtained.

[0056] Optionally, a dot matrix can be provided on the user login registration interface to facilitate obtaining the head movement trajectory and eye movement trajectory of the user while obtaining the input password through the coordinated movement of the eyes and head when the user gazes at the dot matrix.

[0057] Optionally, the coordinated movement of the user's head and eyes corresponds to drawing a password on the dot matrix in the login registration interface. During the process of drawing a password between any two adjacent dots, the eyes will have a fixation time period between the first dot and the second dot among any two adjacent dots, and a saccade time period will occur during the process of drawing from the first dot to the second dot. Thus, a fixation feature set for the fixation time period is obtained based on the coordinated movement trajectory of the eyes and head in each fixation time period, and a saccade feature set for the saccade time period is obtained based on the coordinated movement trajectory of the eyes and head in each saccade time period. Exemplarily, the obtained features may include position, speed, spectral features, and frequency domain features (such as energy, central frequency) extracted using Fourier transform.

[0058] Optionally, the head and eye movements are recorded by an IMU and an eye tracker, and the data is sampled at a frequency of 50 Hz.

[0059] Optionally, a dot matrix with a 3×3 layout and a 60-degree field of view can be provided on the user login registration interface.

[0060] S102. Input the saccade feature set and the fixation feature set into the corresponding pre-trained random forest classifier, summarize the output results of each random forest classifier, and obtain the first authentication result through a voting mechanism. The random forest classifier is a classifier constructed based on the corresponding feature set obtained when the target user registers.

[0061] When the user registers and logs in, they need to input the password through head-eye coordinated movement. It can be understood that the trajectories drawn when inputting the password during registration and login are the same. For example, if the password is 12369, then the trajectories drawn by the head-eye coordinated movement during registration and login are the same, and both form the figure connected by 12369.

[0062] In a specific implementation manner, obtaining the first authentication result through a voting mechanism means inputting the features of each time period into the corresponding random forest classifier for biometric authentication matching to obtain the output results of each random forest classifier. If a target number of random forest classifiers consider the currently logged-in user to be the registered target user, then the biometric authentication passes; otherwise, the biometric authentication fails.

[0063] The random forest classifier is a classifier constructed based on the corresponding feature set obtained when the target user registers. It can be, when the user registers, based on the head movement trajectory and eye movement trajectory formed by inputting the password through head-eye coordinated movement, to extract the saccade feature set formed by the eye and head movements in each saccade time period, and the fixation feature set formed by the eye and head movements in each fixation time period. Based on this, to construct the random forest model classifier corresponding to each time period. Optionally, a random forest model classifier can be pre-trained for each saccade feature set, and a random forest model classifier can be pre-trained for each fixation feature set.

[0064] S103. Perform matching authentication on the password indicated by the eye movement trajectory and the password of the target user when registering to obtain the second authentication result.

[0065] Match the password drawn by the user's head-eye coordinated movement with the password pre-stored by the registered target user. If the password match passes, then the password authentication passes; otherwise, the authentication fails.

[0066] S104. Determine whether the first user is successfully authenticated according to the first authentication result and the second authentication result.

[0067] In a possible implementation manner, when both the first authentication result and the second authentication result are authentication passed, it is determined that the first user is successfully authenticated. Of course, it is also possible to select one of the authentication results as needed to determine the authentication result.

[0068] Based on the above steps S101 - S104, it can be known that the password input method combining eye and head coordinated movements in this application combines two types of biological information, namely eye movement and head movement, and uses a random forest classifier for decision - making analysis to determine whether the user is a registered user. At the same time, password verification is also performed according to the password formed by the head - eye coordinated movement. In this way, the combination of two types of biological information and password matching is achieved, and multiple - information combination verification is carried out, which improves security, is not vulnerable to shoulder - surfing attacks, and only requires head - eye coordinated movement without manual operation, thus enhancing the user experience.

[0069] In the embodiment of this application, the above Figure 1 There are multiple possible implementation methods for the step S101, which will be introduced separately below. It should be noted that the implementation methods given in the following introduction are only for illustrative purposes and do not represent all the implementation methods of the embodiment of this application.

[0070] In a specific implementation method, refer to Figure 2 A schematic diagram of a password - input process as shown. During the process of obtaining the password input by the coordinated movement of the eyes and head of the first user, the eye and head movements in each saccade time period form a saccade feature set, and the eye and head movements in each fixation time period form a fixation feature set. The implementation steps can be as follows:

[0071] First, collect the eye movement trajectory and head movement trajectory during the process of the first user inputting the password by the coordinated movement of the eyes and head.

[0072] Optionally, the login / registration interface adopts a grid design of a 3×3 dot matrix, and the user completes password drawing input by the coordinated movement of the head and eyes to select the target dot.

[0073] Optionally, when it is detected that the first user blinks unnaturally, start collecting the eye movement trajectory and head movement trajectory during the coordinated movement of the eyes and head, and end the collection when it is detected again that the first user blinks unnaturally. In this way, refer to Figure 2 , use "unnatural blinking" (lasting more than 0.3 seconds) as the trigger mechanism for starting and ending the collection of head - eye movement trajectories to reduce false triggers.

[0074] Then, generate an eye time series according to the eye movement trajectory and generate a head time series according to the head movement trajectory.

[0075] Optionally, use the IMU sensor and eye tracker of the VR headset to record the movement trajectories of the head and eyes and generate time - series data.

[0076] Exemplarily, refer to Figure 2, the password for head-eye coordinated movement input is 12369. Among them, the interface adopts a 3×3 dot array layout with a 60-degree field of view. Specifically, the process of drawing the input password through head-eye coordination can be as follows: First, the user closes their eyes and then re-opens them after 0.3 seconds to initiate the authentication process, as shown in (a) of Figure 2 . Second, to input the "1" in the password, the user rotates their head and eyes and focuses their line of sight on the upper left dot in the 3×3 dot array of the target test area. Third, the blue arrow indicates the orientation of the head, and the red arrow indicates the direction of eye fixation, as shown in (b) of Figure 2 . Furthermore, the user continues to rotate their head and eyes and moves their line of sight to the center point of the first row in the target test area to input "2", as shown in (c) of Figure 2 . Fourth, by analogy with the head-eye coordination movement, after completing the drawing and input of the entire password pattern, the user closes their eyes again and re-opens them to end the authentication process, as shown in (d) of Figure 2 .

[0077] Finally, extract the saccade feature sets of the eye time series and the head time series in each saccade time period; extract the fixation feature sets of the eye time series and the head time series in each fixation time period.

[0078] During the process of inputting the password, the eyes will exhibit an eye fixation state corresponding to each dot of the password, and an eye saccade state will occur during the process of the eyes saccading to connect adjacent dots of the password. Therefore, refer to the schematic diagram of head-eye coordination movement feature extraction shown in Figure 3 . The eye time series and the head time series formed by head-eye coordinated movement can be sequentially segmented into multiple time periods based on eye fixation and eye saccades. Each time period corresponds to a segment of the eye time series and a segment of the head time series. When the eyes and the head are coordinated to input the password, there are multiple saccade time periods and multiple fixation time periods, and the saccade time periods and the fixation time periods are arranged alternately. Furthermore, features can be extracted based on the time series corresponding to each time period to form a feature set.

[0079] Exemplarily, as shown in Figure 3, the time axis is divided into multiple segments. Two states of eye movement (Fixation and Saccade) during the head-eye coordinated movement are alternately presented in the segments, forming an alternation between the Fixation time period and the Saccade time period. Among them, Fixation is when the eyes focus relatively statically at a certain position, and Saccade is when the eyes quickly move from one fixation point (grid point) to another fixation point (grid point). Furthermore, the Time Series lists the parameters of the Head and Eye in the time dimension for each time period, which can include Yaw(t), Pitch(t), Roll(t), spatial position coordinates X(t), Y(t), Z(t), etc., and is used to describe the movement states and positions of the head and eyes at different times. Finally, the Features can be extracted for each parameter in the time dimension for each time period, and the extracted features can include two parts: Raw Deviation and Spectrum. The Raw Deviation part can include the maximum value max, the minimum value min, the average value mean, the median median, and the standard deviation variance; the Spectrum part contains the spectral energy energy, the spectral centroid centroid, the spectral bandwidth spectralbandwidth, and the peak frequency position peak frequency position. The above features can be used for the analysis and description of eye movement data.

[0080] Therefore, extracting the saccade feature sets of the eye time series and the head time series in each saccade time period; extracting the fixation feature sets of the eye time series and the head time series in each fixation time period can include:

[0081] First, extract the eye saccade features corresponding to each saccade time period and the eye fixation features corresponding to each fixation time period in the eye time series; extract the head saccade features corresponding to each saccade time period and the head fixation features corresponding to each fixation time period in the head time series; then, combine the corresponding eye saccade features and head saccade features in the same saccade time period into a saccade feature set; combine the corresponding eye fixation features and head fixation features in the same fixation time period into a fixation feature set.

[0082] In this way, the password for password matching is obtained. At the same time, the eye time series based on the eye movement trajectory and the head time series based on the head movement trajectory when the user completes the password input process are also obtained, so as to facilitate the subsequent extraction of the first user biometric features (saccade features and fixation features) for authentication with the random forest classifier pre-trained with the biometric features (saccade enhanced feature set and fixation enhanced feature set) of the target user registered through registration. Among them, the method for the target user registered through registration to obtain biometric features is the same as the method for obtaining the biometric features of the first user logging in described above. Therefore, referring to Figure 4 the schematic flowchart of pre-training a random forest classifier based on the biometric features of the target user registered through registration shown. The method for pre-training a random forest classifier based on the biometric features of the target user registered through registration can be:

[0083] S401. When the target user registers, collect the eye movement trajectory and the head movement trajectory when the eyes and the head move in coordination to input the password, and generate the eye registration time series of the eye movement trajectory and the head registration time series of the head movement trajectory;

[0084] S402. Extract the saccade registration feature set in each saccade time period from the eye registration time series and the head registration time series, and extract the fixation registration feature set in each fixation time period from the eye registration time series and the head registration time series;

[0085] S403. Perform data augmentation on the saccade registration feature set in each saccade time period by adding Gaussian noise to obtain a saccade enhanced feature set; perform data augmentation on the fixation registration feature set in each fixation time period by adding Gaussian noise to obtain a fixation enhanced feature set;

[0086] S404. Construct the random forest classifier corresponding to the saccade time period based on the saccade enhanced feature set corresponding to each saccade time period, with each saccade time period corresponding to a random forest classifier; construct the random forest classifier corresponding to the fixation time period based on the fixation enhanced feature set corresponding to each fixation time period, with each fixation time period corresponding to a random forest classifier.

[0087] In this way, in the registration stage, the target user can input the same password multiple times to generate a behavioral biometric template and pre-train the corresponding random forest classifier, so that when the subsequent user logs in, the biometric features of the first user logging in can be authenticated through the pre-trained random forest classifier, and the output results of multiple classifiers are statistically analyzed through a voting mechanism, so as to verify the user identity through biometric features. In addition, the present application adopts two-factor authentication including password matching and biometric matching to improve the reliability of authentication.

[0088] In a specific implementation manner, inputting the saccade feature set and the fixation feature set into the corresponding pre-trained random forest classifier may include:

[0089] Inputting the saccade feature set corresponding to each saccade time period into the pre-trained random forest classifier corresponding to the saccade time period, and outputting a first authentication sub-result; wherein, when the probability that the random forest classifier analyzes that the first user is the target user based on the saccade feature set exceeds a first threshold, the first authentication sub-result is authentication passed; when the probability that the random forest classifier analyzes that the first user is the target user based on the saccade feature set does not exceed the first threshold, the first authentication sub-result is authentication failed;

[0090] Inputting the fixation feature set corresponding to each fixation time period into the random forest classifier corresponding to the fixation time period, and outputting a second authentication sub-result; wherein, when the probability that the random forest classifier analyzes that the first user is the target user based on the fixation feature set exceeds a second threshold, the second authentication sub-result is authentication passed; when the probability that the random forest classifier analyzes that the first user is the target user based on the fixation feature set does not exceed the second threshold, the second authentication sub-result is authentication failed.

[0091] The first thresholds corresponding to the pre-trained random forests corresponding to any two saccade time periods may be the same or different. Similarly, the second thresholds corresponding to the pre-trained random forest classifiers corresponding to any two fixation time periods may be the same or different from each other.

[0092] It can be understood that in actual applications or testing processes, the false rejection rate (FRR) and the false acceptance rate (FAR) can be balanced through thresholds (such as the first threshold, the second threshold, and the number of classifiers corresponding to the first authentication result being passed when statistically analyzing the output results of multiple random forest classifiers by a voting mechanism, etc.).

[0093] It can be understood that the above embodiments are applicable to all head-mounted display devices supporting IMU and eye movement tracking, including VR and AR devices. By inputting a password based on head-eye coordinated movement to implement two-factor authentication of biometric matching and password matching, additional hardware support is not required, and hand movement is not required, which has broader application potential in restricted scenarios and conforms to the immersive interaction experience of VR.

[0094] The above are some specific implementation manners of a method for an authentication device provided by an embodiment of the present application. Based on this, the present application also provides a corresponding device. Next, the device provided by the embodiment of the present application will be introduced from the perspective of functional modularization.

[0095] SeeFigure 5 Schematic structural diagram of an authentication device shown, an authentication device includes:

[0096] An acquisition unit 501, configured to obtain a saccade feature set formed by eye and head movements during each saccade time period and a fixation feature set formed by eye and head movements during each fixation time period when acquiring the input password through the coordinated movement of the eyes and head of the first user;

[0097] A classification unit 502, configured to input the saccade feature set and the fixation feature set into corresponding pre-trained random forest classifiers, summarize the output results of each random forest classifier, and obtain a first authentication result through a voting mechanism, where the random forest classifier is a classifier constructed according to the corresponding feature set obtained during the registration of the target user;

[0098] A determination unit 503, configured to perform matching authentication on the password indicated by the eye movement trajectory and the password of the target user during registration to obtain a second authentication result;

[0099] A result unit 504, configured to determine whether the first user is successfully authenticated according to the first authentication result and the second authentication result.

[0100] Based on the above device, multiple information verifications are combined with passwords and two biometric information to improve security, not easily vulnerable to shoulder surfing attacks, and there is no need to provide information with unique identification such as fingerprints and irises of users, and only head-eye coordinated movement is required without manual operation, improving the user experience.

[0101] In a possible implementation manner, the acquisition unit 501 is specifically configured to collect the eye movement trajectory and the head movement trajectory during the coordinated movement of the eyes and head of the first user when inputting the password; generate an eye time series according to the eye movement trajectory, and generate a head time series according to the head movement trajectory; extract the saccade feature set of the eye time series and the head time series during each saccade time period; extract the fixation feature set of the eye time series and the head time series during each fixation time period.

[0102] Optionally, when it is detected that the first user blinks unnaturally, start collecting the eye movement trajectory and the head movement trajectory during the coordinated movement of the eyes and head, and end the collection when it is detected again that the first user blinks unnaturally.

[0103] Optionally, when inputting the password through the coordinated movement of the eyes and head, it includes multiple saccade time periods and multiple fixation time periods, and the saccade time periods and the fixation time periods are arranged alternately.

[0104] Optionally, the above-mentioned obtaining unit 501 for obtaining the saccade feature sets of the eye time series and the head time series in each saccade time period, and the fixation feature sets of the eye time series and the head time series in each fixation time period, specifically is configured to:

[0105] Extract the eye saccade features corresponding to each saccade time period and the eye fixation features corresponding to each fixation time period in the eye time series; extract the head saccade features corresponding to each saccade time period and the head fixation features corresponding to each fixation time period in the head time series; combine the corresponding eye saccade features and head saccade features in the same saccade time period into a saccade feature set; combine the corresponding eye fixation features and head fixation features in the same fixation time period into a fixation feature set.

[0106] Optionally, the device further includes a pre-training module, which is configured to collect the eye movement trajectory and the head movement trajectory when the target user inputs a password during registration with the coordinated movement of the eyes and the head, and generate the eye registration time series of the eye movement trajectory and the head registration time series of the head movement trajectory; extract the saccade registration feature sets of the eye registration time series and the head registration time series in each saccade time period, and extract the fixation registration feature sets of the eye registration time series and the head registration time series in each fixation time period; perform data augmentation on the saccade registration feature sets of each saccade time period by adding Gaussian noise to obtain saccade enhanced feature sets; perform data augmentation on the fixation registration feature sets of each fixation time period by adding Gaussian noise to obtain fixation enhanced feature sets; construct the random forest classifiers corresponding to the saccade time periods based on the saccade enhanced feature sets corresponding to each saccade time period, with each saccade time period corresponding to a random forest classifier; construct the random forest classifiers corresponding to the fixation time periods based on the fixation enhanced feature sets corresponding to each fixation time period, with each fixation time period corresponding to a random forest classifier.

[0107] Optionally, the pre-training module is specifically configured to input the saccade feature set corresponding to each saccade time period into the pre-trained random forest classifier corresponding to the saccade time period, and output a first authentication sub-result; wherein, when the probability that the random forest classifier analyzes that the first user is the target user based on the saccade feature set exceeds a first threshold, the first authentication sub-result is authentication passed; when the probability that the random forest classifier analyzes that the first user is the target user based on the saccade feature set does not exceed the first threshold, the first authentication sub-result is authentication failed; input the fixation feature set corresponding to each fixation time period into the random forest classifier corresponding to the fixation time period, and output a second authentication sub-result; wherein, when the probability that the random forest classifier analyzes that the first user is the target user based on the fixation feature set exceeds a second threshold, the second authentication sub-result is authentication passed; when the probability that the random forest classifier analyzes that the first user is the target user based on the fixation feature set does not exceed the second threshold, the second authentication sub-result is authentication failed.

[0108] Optionally, the user authentication interface arranges a 3×3 grid array at a 60-degree field of view angle, so as to facilitate the user to input a password by gazing at the 3×3 grid array.

[0109] The result module 504 is specifically configured to determine that the first user authentication is successful when both the first authentication result and the second authentication result are authentication passed.

[0110] The embodiments of the present application also provide corresponding devices and computer storage media for implementing the solutions provided by the embodiments of the present application.

[0111] Wherein, the device includes a memory and a processor, the memory is used to store instructions or codes, and the processor is used to execute the instructions or codes so that the device executes an identity authentication method according to any embodiment of the present application.

[0112] The computer storage medium stores codes, and when the codes are run, the device running the codes implements an identity authentication method according to any embodiment of the present application.

[0113] In the embodiments of the present application, the "first", "second" (if any) in the names such as "first" and "second" are only used as name identifiers and do not represent the first and second in order.

[0114] As can be seen from the description of the above embodiments, those skilled in the art can clearly understand that all or part of the steps in the above method embodiments can be implemented by means of software plus a general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, and the computer software product can be stored in a storage medium, such as read-only memory (ROM) / RAM, magnetic disk, optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network communication device such as a router) to execute the methods described in various embodiments or some parts of the embodiments of the present application.

[0115] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the device embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiments. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0116] The above is only an exemplary embodiment of the present application and is not used to limit the protection scope of the present application.

Claims

1. An identity authentication method, characterized in that: The method comprises: In the process of obtaining the coordinated eye and head movement of the first user to input the password, the eye and head movement in each scanning time period forms a scanning feature set, and the eye and head movement in each gaze time period forms a gaze feature set; Inputting the scan feature set and the gaze feature set into a corresponding pre-trained random forest classifier, summarizing the output results of each random forest classifier, and obtaining a first authentication result through a voting mechanism, wherein the random forest classifier is a classifier constructed according to the corresponding feature set obtained when the target user registers; Authentication is performed based on matching the password indicated by the eye movement trajectory with the password of the target user when registering, to obtain a second authentication result; Determine whether the first user's identity authentication is successful based on the first authentication result and the second authentication result.

2. The method according to claim 1, characterized in that In the process of obtaining the coordinated eye and head movement of the first user to input the password, the eye and head movements in each scanning time period form a scanning feature set, and the eye and head movements in each gaze time period form a gaze feature set, including: Collecting eye movement trajectories and head movement trajectories of the first user during the process of inputting the password by coordinated eye and head movement; Generate an eye time series according to the eye movement trajectory, and generate a head time series according to the head movement trajectory; Extracting a set of glance features of the eye time series and the head time series in each glance time period; extracting a set of gaze features of the eye time series and the head time series in each gaze time period.

3. The method according to claim 2, characterized in that The coordinated movement of the eyes and the head to input a password includes a plurality of scanning time periods and a plurality of gazing time periods, and the scanning time periods and the gazing time periods are arranged alternately.

4. The method according to claim 3, characterized in that extracting a set of glance features of the eye time series and the head time series in each glance time period; Extracting a gaze feature set of the eye time series and the head time series in each gaze time period includes: Extracting eye scanning features corresponding to each scanning time period in the eye time series and eye gaze features corresponding to each gaze time period; extracting head scanning features corresponding to each scanning time period in the head time series and head gaze features corresponding to each gaze time period; The eye glance features and head glance features corresponding to the same glance time period are combined into a glance feature set; the eye gaze features and head gaze features corresponding to the same gaze time period are combined into a gaze feature set.

5. The method according to any one of claims 1 to 4, characterized in that: The training method of the random forest classifier includes: Collecting the eye movement trajectory and the head movement trajectory of the target user when the eyes and head move in coordination to input the password during registration, and generating an eye registration time series of the eye movement trajectory and a head registration time series of the head movement trajectory; Extracting a set of glance registration features in each glance time period in the eye registration time series and the head registration time series, and extracting a set of gaze registration features in each gaze time period in the eye registration time series and the head registration time series; The scanning registration feature set of each scanning time period is enhanced by adding Gaussian noise to obtain a scanning enhancement feature set; the gaze registration feature set of each gaze time period is enhanced by adding Gaussian noise to obtain a gaze enhancement feature set; Based on the scanning enhancement feature set corresponding to each scanning time period, a random forest classifier corresponding to the scanning time period is constructed, and each scanning time period corresponds to a random forest classifier; A random forest classifier corresponding to each gaze time period is constructed based on the gaze enhancement feature set corresponding to each gaze time period, and each gaze time period corresponds to a random forest classifier.

6. The method according to claim 5, characterized in that The step of inputting the scan feature set and the gaze feature set into a corresponding pre-trained random forest classifier comprises: Inputting a scanning feature set corresponding to each scanning time period into a pre-trained random forest classifier corresponding to the scanning time period, and outputting a first authentication sub-result; wherein, when the probability that the first user is a target user analyzed by the random forest classifier based on the scanning feature set exceeds a first threshold, the first authentication sub-result is authentication passed; when the probability that the first user is a target user analyzed by the random forest classifier based on the scanning feature set does not exceed the first threshold, the first authentication sub-result is authentication failed; The gaze feature set corresponding to each gaze time period is input into the random forest classifier corresponding to the gaze time period, and a second authentication sub-result is output; wherein, when the probability that the first user is the target user analyzed by the random forest classifier based on the gaze feature set exceeds a second threshold, the second authentication sub-result is authentication passed; when the probability that the first user is the target user analyzed by the random forest classifier based on the gaze feature set does not exceed the second threshold, the second authentication sub-result is authentication failed.

7. The method according to claim 2, characterized in that The collecting of the eye movement trajectory and the head movement trajectory of the first user in the process of inputting the password by coordinated eye and head movement includes: When unnatural blinking of the first user is detected, the acquisition of eye movement trajectory and head movement trajectory in the coordinated movement of the eyes and the head is started, and when unnatural blinking of the first user is detected again, the acquisition is terminated.

8. The method according to claim 7, characterized in that The user authentication interface is arranged with a 3×3 grid dot array at a 60-degree viewing angle, so that the user can focus on the 3×3 grid dot array to input a password.

9. The method according to claim 1, characterized in that: The determining whether the first user passes the authentication according to the first authentication result and the second authentication result includes: When both the first authentication result and the second authentication result are authentication passed, it is determined that the first user identity authentication is successful.

10. An identity verification device, characterized in that: The device comprises: An acquisition unit, configured to acquire, during the process of inputting a password by coordinated eye and head movement of the first user, a set of scan features formed by eye and head movement in each scan time period, and a set of gaze features formed by eye and head movement in each gaze time period; A classification unit, configured to input the scan feature set and the gaze feature set into a corresponding pre-trained random forest classifier, summarize the output results of each random forest classifier, and obtain a first authentication result through a voting mechanism, wherein the random forest classifier is a classifier constructed according to the corresponding feature set obtained when the target user registers; A judgment unit, configured to match and authenticate the password indicated by the eye movement track with the password of the target user when registering, to obtain a second authentication result; A result unit is used to determine whether the identity authentication of the first user is successful according to the first authentication result and the second authentication result.

Images