Human-machine verification method and apparatus, and device and medium

Abstract

The present application belongs to the field of data processing. Disclosed are a human-machine verification method and apparatus, and a device and a medium. The method comprises: in response to at least one randomly output screen partition of a display screen, collecting image frames of the display screen within a verification duration; on the basis of change information of a facial image in the image frames, obtaining a first-type verification parameter; in response to at least one randomly output facial region of the facial image, acquiring a touch input of a user on the display screen; on the basis of the touch input, obtaining a second-type verification parameter; on the basis of the first-type verification parameter and the second-type verification parameter, obtaining a comprehensive verification parameter; and when the comprehensive verification parameter meets a preset user determination condition, determining that the subject who executes the present verification operation is a legitimate user.

Description

Human-machine verification method, device, equipment and medium

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to Chinese patent application 202410071077.1, filed on January 17, 2024, entitled “Human-machine verification method, device, equipment and medium,” and the entire contents of that application are incorporated herein by reference. Technical Field

[0003] The present application relates to the field of data processing, and in particular to a human-machine verification method, device, equipment and medium. Background Art

[0004] With the continuous development of electronic information technology, more and more businesses are being implemented through applications (APPs). Applications require user input to access business resources. However, during business execution, some criminals use machine scripts to perform batch operations, resulting in the illicit acquisition of large amounts of business resources, making it difficult for legitimate users to access business resources.

[0005] To distinguish legitimate users from machine scripts, verification codes such as slider verification codes and graphic verification codes can be set to complete user authentication, thereby distinguishing legitimate users from machine scripts. However, slider verification codes can be cracked by machine scripts that include image processing and comparison and simulated linear movement, and graphic verification codes can be cracked by machine scripts using optical character recognition (OCR) technology. Using these verification code methods, it is difficult to accurately distinguish between legitimate user and machine script operations, and the accuracy of human-machine verification is poor. Summary of the Invention

[0006] The embodiments of the present application provide a human-machine verification method, apparatus, device, and medium, which can improve the accuracy of human-machine verification.

[0007] In the first aspect, an embodiment of the present application provides a human-machine verification method, including: in response to at least one screen partition of a randomly output display screen, collecting image frames of the display screen within the verification time; obtaining a first type of verification parameter based on change information of the face image in the image frame; in response to at least one facial area of ​​the randomly output face image, obtaining the user's touch input on the display screen; based on the touch input, obtaining a second type of verification parameter; according to the first type of verification parameter and the second type of verification parameter, obtaining a comprehensive verification parameter; when the comprehensive verification parameter meets the preset user judgment condition, determining that the object performing this verification operation is a legitimate user.

[0008] In the second aspect, an embodiment of the present application provides a human-machine verification device, including: an image data acquisition module, used to collect image frames of the display screen within the verification time in response to at least one screen partition of the randomly output display screen; a first parameter acquisition module, used to obtain a first type of verification parameter based on change information of the face image in the image frame; a touch data acquisition module, used to obtain the user's touch input on the display screen in response to at least one facial area of ​​the randomly output face image; a second parameter acquisition module, used to obtain a second type of verification parameter based on the touch input; a user judgment module, used to obtain a comprehensive verification parameter based on the first type of verification parameter and the second type of verification parameter; and, when the comprehensive verification parameter meets the preset user judgment condition, determining that the object performing this verification operation is a legitimate user.

[0009] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the human-machine verification method of the first aspect is implemented.

[0010] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, the human-machine verification method of the first aspect is implemented.

[0011] Embodiments of the present application provide a human-machine verification method, apparatus, device, and medium. These methods can randomly output at least one screen partition of a display screen during verification to instruct a user to move a facial image on the screen to the randomly output screen partition, thereby capturing image frames of the display screen during verification and obtaining first-category verification parameters based on changes in the facial image within the image frames. During verification, at least one facial region of a facial image can also be randomly output to instruct the user to perform touch input on the randomly output facial region, thereby obtaining second-category verification parameters based on the touch input. The randomness of the screen partitions and facial regions, as well as the user-related operations during verification, are difficult to simulate or crack by machine scripts. The first-category verification parameter can reflect the likelihood that the verification operation is being performed by a legitimate user, based on changes in the facial image. The second-category verification parameter can reflect the likelihood that the verification operation is being performed by a legitimate user, based on touch input. A comprehensive verification parameter is derived based on the first and second-category verification parameters. This comprehensive verification parameter accurately distinguishes legitimate user operations from those performed by machine scripts, thereby improving the accuracy of human-machine verification. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0013] FIG1 is a flow chart of a human-machine verification method provided by an embodiment of the present application;

[0014] FIG2 is a schematic diagram of an example of screen partitioning in a display screen provided by an embodiment of the present application;

[0015] FIG3 is a schematic diagram of another example of screen partitioning in a display screen provided by an embodiment of the present application;

[0016] FIG4 is a schematic diagram of an example of three screen partitions randomly outputted according to an embodiment of the present application;

[0017] FIG5 is a schematic diagram of an example of a facial region in a face image provided by an embodiment of the present application;

[0018] FIG6 is a schematic diagram of another example of a facial region in a face image provided by an embodiment of the present application;

[0019] FIG7 is a flow chart of a human-machine verification method provided by another embodiment of the present application;

[0020] FIG8 is a schematic diagram of an example of a three-dimensional space coordinate system provided in an embodiment of the present application;

[0021] FIG9 is a flowchart of a human-machine verification method provided by another embodiment of the present application;

[0022] FIG10 is a schematic diagram of the structure of a human-machine verification device provided in one embodiment of the present application;

[0023] FIG11 is a schematic structural diagram of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0024] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is merely to provide a better understanding of the present application by illustrating examples of the present application. It should be noted that the acquisition, storage, use, processing, etc. of information and data in the embodiments of the present application are authorized by the user or relevant agencies and comply with the relevant provisions of national laws and regulations.

[0025] With the continuous development of electronic information technology, more and more businesses are being implemented through applications. Applications require user input to access business resources. However, during business execution, some criminals use machine scripts to perform batch operations, resulting in the illegal acquisition of large amounts of business resources and the difficulty for legitimate users to access these resources. For example, a service provider may offer coupons that require user authentication to claim. Using machine scripts to claim coupons can allow a large number of coupons to be obtained in a short period of time, significantly reducing the number of coupons available to legitimate users who perform the operations themselves, or even eliminating them altogether. To distinguish legitimate users from machine scripts, verification codes such as slider verification codes and graphic verification codes can be implemented. These verification codes can be used to perform human-machine verification, thereby distinguishing legitimate users from machine scripts. However, slider verification codes can be cracked by machine scripts using techniques such as image processing and comparison and simulated linear movement, while graphic verification codes can be cracked by machine scripts using optical character recognition technology. These verification methods make it difficult to accurately distinguish between legitimate users and machine scripts, resulting in poor human-machine verification accuracy.

[0026] The present application provides a human-machine verification method, apparatus, device, and medium that can perform human-machine verification based on facial images, allowing users to move their facial images so that they appear in randomly output screen partitions, and allowing users to touch randomly output facial areas in facial images. Based on the information collected during the user's execution of the above operations, the object performing the verification operation is verified and evaluated, and a comprehensive verification parameter that can characterize the possibility that the object performing the verification operation is a legitimate user is obtained. Based on the comprehensive verification parameter, it is determined whether the object performing the verification operation is a legitimate user. The human-machine verification method, apparatus, device, and medium provided in the present application can effectively determine the operation of a real person, the person himself, and the randomness of the screen partitions, the randomness of the facial areas, and the operations related to the facial image that the user needs to perform are difficult to be cracked by machine scripts, thereby accurately distinguishing between the operations of legitimate users and machine scripts, and improving the accuracy of human-machine verification. The above verification operations are related to facial images and do not require the graphic design of the verification code itself, making it easier to maintain and easier to operate. The human-machine verification method is also more adaptable to the elderly population and provides a better user experience.

[0027] The human-machine verification method, device, equipment and medium provided in this application are described below.

[0028] In a first aspect, the present application provides a human-machine verification method that can be applied to business resource acquisition scenarios, such as e-voucher acquisition, as an enhanced verification method. The human-machine verification method can be performed by a human-machine verification device, equipment, etc. The human-machine verification device, equipment, etc. can include a terminal device and / or a server, but is not limited here. Figure 1 is a flowchart of the human-machine verification method provided in one embodiment of the present application. As shown in Figure 1, the human-machine verification method can include steps S101 to S106.

[0029] In step S101 , in response to at least one screen partition of a display screen output randomly, image frames of the display screen within a verification time period are collected.

[0030] The display screen is a display screen of a user-facing device, and the display screen may include multiple screen partitions. In some examples, two adjacent screen partitions in the multiple screen partitions may partially overlap, and adjacent here refers to adjacent in position. The display screen may be divided into multiple screen partitions according to the size of the display screen. In some examples, a preset size range may be set in advance, and the preset size range may be set according to scenarios, requirements, experience, etc., and is not limited here. For example, the preset size range may be less than or equal to 6.1 inches. When the size of the display screen exceeds the preset size range, the number of screen partitions into which the display screen is divided may be greater than the number of screen partitions into which the display screen is divided when the size of the display screen is within the preset size range. Specifically, when the size of the display screen exceeds the preset size range, the display screen includes N1 screen partitions, N1 is a positive integer greater than 1; when the size of the display screen is within the preset size range, the display screen includes N2 screen partitions, N2 is a positive integer greater than 1 and N2 < N1.

[0031] For example, Figure 2 is a schematic diagram of an example of screen partitions in a display screen provided in an embodiment of the present application. The size of the display screen in Figure 2 exceeds the preset size range. As shown in Figure 2, the display screen may include 9 screen partitions, and the 9 screen partitions have the same size. The 9 screen partitions are A10A12A5A17, A11A13A6A4, A12A14A18A5, A15A2A8A19, A1A3A9A7, A2A16A18A5, A17A5A23A21, A4A6A24A22 and A5A18A25A23, and the center points of these 9 screen partitions are point A1 to point A9, respectively.

[0032] For another example, FIG3 is a schematic diagram of another example of screen partitions in a display screen provided by an embodiment of the present application. The size of the display screen in FIG3 is within a preset size range. As shown in FIG3 , the display screen may include five screen partitions, each of which has the same size. The five screen partitions are B1B2B12B11, B3B6B16B13, B5B8B18B15, B9B10B20B19, and B4B6B16B14. The screen partitions in FIG3 can be viewed as retaining the screen partitions centered on points A2, A4, A5, A6, and A8 from the screen partitions in FIG2 , and moving the screen partitions centered on points A2, A4, A6, and A8 toward point A5 by a certain distance, such as 1 / 8 of the length or 1 / 8 of the width of the screen partition, thereby forming the five screen partitions shown in FIG3 .

[0033] At least one screen partition of the display screen can be randomly displayed to prompt the user to move their head or a camera device on the display screen so that the user's facial image moves to the randomly displayed screen partition. During the movement of the user's facial image, image frames of the display screen can be captured, including the facial image and other images. The verification duration can be a set duration, determined based on the scenario, needs, experience, etc. The verification duration can also be the time required for the facial image to move from an initial position to the last randomly displayed screen partition. The difficulty of the screen partition verification can be set based on the number of randomly displayed screen partitions. The greater the difficulty of the screen partition verification, the greater the number of randomly displayed screen partitions. For example, screen partition verification difficulty M1 corresponds to the random output of one screen partition; screen partition verification difficulty M2 corresponds to the random output of two screen partitions; and screen partition verification difficulty M3 corresponds to the random output of three screen partitions. Difficulty M3 is higher than difficulty M2, which is higher than difficulty M1. If one screen partition is randomly displayed, the facial image must be moved from the initial position to the randomly displayed screen partition. If more than two screen partitions are randomly output, the facial image needs to be moved from the starting position to the first screen partition randomly output, and then from the first screen partition randomly output to the second screen partition randomly output. In order to increase the difficulty of cracking the machine script, when the number of randomly output screen partitions is more than three, the line connecting the center points of the three or more randomly output screen partitions is a broken line, that is, the center points of the three or more randomly output screen partitions are not on the same straight line, which increases the difficulty of the machine script in predicting the randomly output screen partitions in the embodiment of the present application. For example, Figure 4 is a schematic diagram of an example of three randomly output screen partitions provided in an embodiment of the present application. As shown in Figure 4, the center points of the three screen partitions are O1, O2 and O3 respectively. During the verification process, the user can move the facial image from the screen partition where the center point O1 is located to the screen partition where the center point O2 is located, and then from the screen partition where the center point O2 is located to the screen partition where the center point O3 is located. The line connecting the center points O1, O2 and O3 is a broken line.

[0034] The random output of screen partitions can reduce the possibility of attackers using automated operations to crack human-machine verification by using database collisions.

[0035] In step S102, a first type of verification parameter is obtained based on change information of the face image in the image frame.

[0036] During the movement of the user's facial image, multiple image frames may be acquired, and the facial images in different image frames may change. For example, the position of the facial image in the image frame may change, and the movement of the facial image in the image frame may change. The first type of verification parameter may characterize the possibility that the object of this verification operation is a legitimate user in terms of the change of the facial image. The higher the first type of verification parameter, the higher the possibility that the object of this verification operation is a legitimate user in terms of the change of the facial image. The first type of verification parameter may be positively correlated with the degree of closeness between the change information of the facial image in the image frame and the change standard, that is, the closer the change information of the facial image in the image frame is to the change standard, the larger the value of the first type of verification parameter.

[0037] In step S103, in response to at least one facial region of the randomly outputted face image, a touch input of the user on the display screen is acquired.

[0038] A facial image can be divided into multiple facial regions based on the distribution of facial organs. In some examples, the feature points of the facial organs can be obtained through a face detection algorithm, that is, the facial image may include the feature points of the identified facial organs. According to the coordinates of the feature points of the facial organs, the facial region in the facial image is obtained. The facial region may include the forehead region, eyebrow region, eye region, nose region, chin region, etc. For example, Figure 5 is a schematic diagram of an example of a facial region in a facial image provided by an embodiment of the present application. As shown in Figure 5, the solid circles are the feature points of the facial organs, and the area defined by the dotted box is the facial region corresponding to the facial organ. The facial region corresponding to a facial organ can be determined based on the feature point with the largest coordinate value and the feature point with the smallest coordinate value among the feature points of the panel organ. For example, the coordinates may include horizontal and vertical coordinates. After obtaining the feature points of a facial organ, the feature point with the largest horizontal coordinate value, the feature point with the smallest horizontal coordinate value, the feature point with the largest vertical coordinate value, and the feature point with the smallest vertical coordinate value may be selected from the feature points. A straight line 1 parallel to the vertical axis is drawn from the feature point with the largest horizontal coordinate value, a straight line 2 parallel to the vertical axis is drawn from the feature point with the smallest horizontal coordinate value, a straight line 3 parallel to the horizontal axis is drawn from the feature point with the largest vertical coordinate value, and a straight line 4 parallel to the horizontal axis is drawn from the feature point with the smallest vertical coordinate. The area defined by straight line 1, straight line 2, straight line 3, and straight line 4 is the facial area corresponding to the facial organ.

[0039] In some examples, in order to improve the tolerance of human-machine verification for touch input of legitimate users, the facial area can appropriately add a tolerance area based on the area determined by the feature point with the largest coordinate value and the feature point with the smallest coordinate value. For example, the feature points of the facial organs can be shifted, and offset values ​​are added to the horizontal and vertical coordinates of each feature point. For example, if the coordinates of the feature point are (Xi, Yi) and the offset value is a, the coordinates of the shifted feature point are (Xi+a, Yi+a). It should be noted that the offset value of the horizontal coordinate and the offset value of the vertical coordinate can be different, and are not limited here; that is, the facial area including the tolerance area can be determined based on the feature point with the largest coordinate value and the feature point with the smallest coordinate value after adding the offset value. For example, Figure 6 is a schematic diagram of another example of the facial area in the face image provided by an embodiment of the present application. As shown in Figure 6, the solid circles are the feature points of the facial organs, and the area defined by the dotted box is the facial area corresponding to the facial organs. According to the comparison between Figure 5 and Figure 6, the facial area in Figure 6 is enlarged relative to the facial area in Figure 5, which can improve the fault tolerance of the touch input of legitimate users, thereby improving the pass rate of legitimate users in human-machine verification.

[0040] At least one facial region may be randomly output to prompt the user to touch the randomly output facial region in the face image on the display screen according to the output touch method. The user's touch input on the display screen is collected. The touch input may be determined based on the number of randomly output facial regions and the touch method. For example, the touch input may include any of the following: a single-finger touch on a facial region; multiple fingers clustered together to touch a facial region; or multiple fingers simultaneously touching multiple facial regions. The number of randomly output facial regions corresponding to a single-finger touch on a facial region and multiple fingers clustered together to touch a facial region is one; the number of randomly output facial regions corresponding to multiple fingers simultaneously touching multiple facial regions is multiple; the touch method corresponding to a single-finger touch on a facial region is a single-finger touch; the touch method corresponding to multiple fingers clustered together to touch a facial region is a multi-finger cluster touch; and the touch method corresponding to multiple fingers simultaneously touching multiple facial regions is a multi-finger separate touch.

[0041] The random output of the facial area can reduce the possibility of attackers using automated operations such as database collision to crack human-machine verification.

[0042] In step S104, a second type of verification parameter is obtained based on the touch input.

[0043] Different randomly output facial regions and different objects performing touch input result in different touch input characteristics. The characteristics of touch input may include, but are not limited to, positional characteristics and behavioral characteristics. The second type of verification parameter may characterize the likelihood that the object of this touch input verification operation is a legitimate user. The higher the second type of verification parameter, the higher the likelihood that the object of this touch input verification operation is a legitimate user. The second type of verification parameter may be positively correlated with the degree of proximity between the characteristics of the touch input and the characteristic standard, that is, the closer the characteristics of the touch input are to the characteristic standard, the larger the value of the second type of verification parameter.

[0044] In step S105, a comprehensive verification parameter is obtained according to the first type of verification parameter and the second type of verification parameter.

[0045] The first and second verification parameters each have a weighting factor. A weighted algorithm can be used to calculate the combined verification parameters based on the first verification parameters, their corresponding weighting factors, and the second verification parameters. This combined verification parameter comprehensively assesses the likelihood that the user being verified is legitimate, based on both facial image changes and touch input. This assessment is more accurate.

[0046] The first type of verification parameters may include one verification parameter or more than two verification parameters. If the first type of verification parameters includes more than two verification parameters, each verification parameter in the first type of verification parameters each corresponds to a weight coefficient. Similarly, the second type of verification parameters may include one verification parameter or more than two verification parameters. If the second type of verification parameters includes more than two verification parameters, each verification parameter in the second type of verification parameters each corresponds to a weight coefficient. For example: the first type of verification parameters includes one verification parameter, which is P1, and the second type of verification parameters includes three verification parameters, which are P2, P3 and P4 respectively; the weight coefficient corresponding to the verification parameter P1 is W1, the weight coefficient corresponding to the verification parameter P2 is W2, the weight coefficient corresponding to the verification parameter P3 is W3, and the weight coefficient corresponding to the verification parameter P4 is W4, then the comprehensive verification parameter P can be calculated according to the following formula (1): P=P1×W1+P2×W2+P3×W3+P4×W4 (1)

[0047] The larger the value of the comprehensive verification parameter, the more likely it is that the user performing the verification operation is a legitimate user. In other words, the likelihood that the user performing the verification operation is a real person is higher. The smaller the value of the comprehensive verification parameter, the more likely it is that the user performing the verification operation is a machine script. For example, if P∈[0,1], the closer P is to 1, the more likely it is that the user performing the verification operation is a legitimate user. The closer P is to 0, the more likely it is that the user performing the verification operation is a machine script.

[0048] In step S106, when the comprehensive verification parameters meet the preset user determination conditions, it is determined that the object performing this verification operation is a legitimate user.

[0049] The user determination condition includes conditions for determining whether the user performing the verification operation is a legitimate user. A legitimate user is a real person. In some examples, the user determination condition may include a comprehensive verification parameter being greater than or equal to a preset verification threshold. That is, if the comprehensive verification parameter is greater than or equal to the preset verification threshold, the user performing the verification operation is determined to be a legitimate user; if the comprehensive verification parameter is less than the preset verification threshold, the user performing the verification operation is determined to be a machine script.

[0050] In an embodiment of the present application, during human-machine verification, at least one screen partition of the display screen can be randomly output to instruct the user to move the facial image on the screen to the randomly output screen partition, thereby capturing image frames of the display screen during the process. Based on the changes in the facial image within the image frames, a first type of verification parameter can be obtained. During human-machine verification, at least one facial region of the facial image can also be randomly output to instruct the user to perform touch input on the randomly output facial region, thereby obtaining a second type of verification parameter based on the touch input. The randomness of the screen partitions and facial regions, as well as the required user-related operations during the human-machine verification process, are difficult to simulate and crack by machine scripts. The first type of verification parameter can reflect the likelihood that the verification operation is being performed by a legitimate user based on changes in the facial image, while the second type of verification parameter can reflect the likelihood that the verification operation is being performed by a legitimate user based on touch input. Based on the first and second type of verification parameters, a comprehensive verification parameter is obtained. This comprehensive verification parameter can accurately distinguish between legitimate user operations and machine script operations, thereby improving the accuracy of human-machine verification. Moreover, the above-mentioned human-machine verification process is mainly implemented through facial images, and there is no need for graphic design of the verification code itself. It is easier to maintain and easier to operate. It is more adaptable to the elderly and visually impaired people, and the user experience is better.

[0051] In some embodiments, the change information of the facial image in the image frame may include the change trajectory information of the facial image in the image frame, and the change trajectory information may include relevant information about the change trajectory of the position of the facial image. The first type of verification parameter can be obtained based on the degree to which the change trajectory information conforms to the standard trajectory information. Figure 7 is a flowchart of the human-machine verification method provided by another embodiment of the present application. The difference between Figure 7 and Figure 1 is that step S102 in Figure 1 can be specifically refined into steps S1021 to S1023 in Figure 7.

[0052] In step S1021 , the change trajectory information of the face image in the image frame and the standard trajectory information of the face image from the initial position to the randomly output screen partition are obtained.

[0053] When a facial image moves to a randomly output screen partition, the position of the facial image in the captured image frame changes. Change trajectory information can be obtained based on the change in the position of the facial image. Standard trajectory information includes trajectory information of the facial image from its initial position to the randomly output screen partition. For example, if the randomly output screen partition includes the screen partition centered on point O1 and the screen partition centered on point O2 in FIG4 , the standard trajectory information can represent the straight line trajectory from the initial position of the facial image to the screen partition centered on point O1 and the straight line trajectory from the screen partition centered on point O1 to the screen partition centered on point O2. The standard trajectory information can be implemented as the slope of each straight line trajectory in the total trajectory. The change trajectory information of the facial image in the image frame can represent the connection trajectory formed by connecting the facial images in multiple image frames. The change trajectory information can represent the slope of the straight line from the facial image position to the initial position and the slope of the straight line from the facial image position to the screen partition. For ease of understanding, the change trajectory information and the standard trajectory information are explained below using an example. For example, the change trajectory information is obtained based on the position information of the center point of the face image and the position information of the center point of the randomly output screen partition. The user needs to move the face image from the screen partition with point O1 as the center point to the screen partition with point O2 as the center point shown in FIG4. The standard trajectory information k12 corresponding to the screen partition with point O1 as the center point moving to the screen partition with point O2 as the center point can be obtained according to the following formula (2). The change trajectory information k1i of the face image can be obtained according to the following formula (3): k12=(Y2-Y1)÷(X2-X1) (2) k1i=(Yi-Y1)÷(Xi-X1) (3)

[0054] The coordinates of point O1 are (X1, Y1), the coordinates of point O2 are (X2, Y2), and the coordinates of the center of the face image are (Xi, Yi).

[0055] In step S1022 , an offset parameter between the changed trajectory information and the standard trajectory information is obtained according to the changed trajectory information and the standard trajectory information.

[0056] The offset parameter can reflect the degree to which the change trajectory of the facial image deviates from the standard trajectory. In some examples, the more the offset parameter exceeds the offset threshold range, the more the change trajectory of the facial image deviates from the standard trajectory. For example, when the change trajectory information and the standard trajectory information can be represented by a slope, the offset parameter can be determined by the ratio of the change trajectory information to the standard estimation information. For example, the offset parameter can be obtained according to the following formula (4): θi=k1i÷k12 (4)

[0057] Wherein, θi is an offset parameter, and the description of k1i and k12 can be found in the relevant description of the above formulas (2) and (3), which will not be repeated here.

[0058] In step S1023, a first type of verification parameter is obtained according to the offset parameter.

[0059] The more image frames whose corresponding offset parameters fall within the offset threshold range, the closer the trajectory generated by the facial image movement operation is to the standard trajectory. A first number of image frames containing facial images whose offset parameters fall within the offset threshold range can be obtained, and a first ratio of the first number to the total number of image frames can be calculated. Based on the first ratio, a first verification parameter can be obtained, and the first verification parameter and the first ratio are positively correlated. Based on the first verification parameter, a first type of verification parameter can be obtained.

[0060] The offset threshold range is the range of offset parameters that determines whether the trajectory of a facial image's change is close to the standard trajectory. If the offset parameter is within the offset threshold range, it indicates that the trajectory of the facial image in that image frame is close to the standard trajectory. The first number is the number of image frames containing facial images whose trajectory information has an offset parameter between the trajectory information and the standard trajectory information within the offset threshold range. A first ratio of the first number to the total number of image frames can represent the proportion of image frames whose trajectory of facial images is close to the standard trajectory. A correspondence between the first ratio and the first verification parameter can be preset, and a first verification parameter corresponding to the first ratio can be determined based on the correspondence and the first ratio. In some examples, a ratio threshold can be pre-set. If the first ratio is below the ratio threshold, the first verification parameter is Pa1. If the first ratio is greater than or equal to the ratio threshold, the first verification parameter is Pa2, where Pa2 > Pa1. Pa1 can be a fixed value or a variable value that changes with the first ratio below the ratio threshold. Similarly, Pa2 can be a fixed value or a variable value that changes with the first ratio above or equal to the ratio threshold.

[0061] In some examples, the first verification parameter may be determined as a first type of verification parameter.

[0062] In other examples, the liveness detection result of the facial image in the image frame can be obtained, and the second verification parameter can be obtained based on the liveness detection result; the first type of verification parameter can be obtained based on the first verification parameter and the second verification parameter. Liveness detection can be performed by the movement changes of the facial image in the image frame. For example, whether the eye image of the facial image in the image frame changes between opening and closing the eyes can be used to determine whether the facial image is a real facial image, so as to distinguish real facial images from static photos. The liveness detection in the embodiment of the present application is imperceptible to the user, and can improve the user experience on the basis of improving the accuracy of human-computer verification. The second verification parameter corresponding to the liveness detection result indicating that the facial image is a real facial image is greater than the second verification parameter corresponding to the liveness detection result indicating that the facial image is not a real facial image. Weight coefficients can be set for the first verification parameter and the second verification parameter respectively. According to the first verification parameter, the weight coefficient corresponding to the first verification parameter, the second verification parameter and the weight coefficient corresponding to the second verification parameter, a weighted algorithm is used to obtain the first type of verification parameter.

[0063] In the above embodiment, the method for obtaining the first type of verification parameters can be determined based on the difficulty mode corresponding to the current human-machine verification. The difficulty mode can indicate the difficulty, and the difficulty can be related to the security risk of the business corresponding to the human-machine verification, customer needs, etc. For example, the higher the security risk of the business corresponding to the human-machine verification, the higher the difficulty indicated by the difficulty mode corresponding to the human-machine verification; the faster the verification speed required by the business corresponding to the human-machine verification, the lower the difficulty mode corresponding to the human-machine verification. The correspondence between the difficulty mode and the human-machine verification can be set in advance, and the difficulty mode corresponding to the current human-machine verification can be determined based on the correspondence between the current human-machine verification, the difficulty mode and the human-machine verification, so as to obtain the first type of verification parameters according to the correspondence between the difficulty mode. For example, when the security risk corresponding to this human-machine verification is low and the verification speed requirement is relatively high, that is, the difficulty indicated by the difficulty mode corresponding to this human-machine verification is low, the first verification parameter can be determined as the first type of verification parameter; when the security risk corresponding to this human-machine verification is high and the verification speed requirement is relatively low, that is, the difficulty indicated by the difficulty mode corresponding to this human-machine verification is high, a liveness detection process can be added to obtain a second verification parameter based on the liveness detection result, and the first type of verification parameter is obtained according to the first verification parameter and the second verification parameter.

[0064] In some embodiments, the preview of facial images can be optimized. Within the image frame, a facial image located in a randomly output screen partition can be captured; if the facial image meets valid facial image conditions, the facial image is determined to be valid. Only if the facial image is valid will change information be acquired and liveness detection performed on the facial image. The valid facial image conditions may include: the area ratio of the largest facial image in the randomly output screen partition to the screen partition exceeds a preset ratio, and / or the pitch angle of the facial image in the randomly output screen partition is within a preset pitch angle range. In some cases, a screen partition may contain multiple facial images. Only the largest facial image is captured. A face detection algorithm can be used to calculate the position and angle of the largest facial image in the screen partition, and the largest face can be framed (i.e., a bounding box operation). The area ratio of the face frame framing the largest face to the screen partition can be determined as the area ratio of the largest facial image to the screen partition. The preset ratio can be set based on the scenario, needs, experience, etc., and can be, for example, 70%. When the area ratio exceeds the preset ratio, the collected facial image, i.e., the maximum facial image, is considered valid. Since different screen partitions are located at different positions on the display screen, the pitch angles of facial images in different screen partitions will also be different. Figure 8 is a schematic diagram of an example of a three-dimensional space coordinate system provided in an embodiment of the present application. As shown in Figure 8, the three-dimensional space coordinate system involves pitch angles, yaw angles, and roll angles. The pitch angle of the collected facial image must be within the preset pitch angle range in order for the facial image to be considered valid. However, since the pitch angle of the facial image in the screen partition located below the display screen is often larger, in order to improve the pass rate of the effective judgment of the facial image, the head-up scene in which the facial image is located in the screen partition below the display screen is optimized, and the upper limit value of the preset pitch angle range corresponding to the screen partition located below the display screen may be higher than the upper limit value of the preset pitch angle range corresponding to the screen partition not located below the display screen. For example, the preset pitch angle range corresponding to the screen partition that is not located below the display screen, such as located above the display screen or in the middle of the display screen, is (-20°, 20°), and the preset pitch angle range corresponding to the screen partition located below the display screen can be (-20°, 40°).

[0065] In some embodiments, the second type of verification parameters may include one or more of a third verification parameter, a fourth verification parameter, and a fifth verification parameter. The third verification parameter is related to the position of the touch input, the fourth verification parameter is related to the behavioral characteristics of the touch input, and the fifth verification parameter is related to the recognition of the face image collected during the touch input process. The following is an example of the second type of verification parameters including the third verification parameter, the fourth verification parameter, and the fifth verification parameter. Figure 9 is a flowchart of a human-machine verification method provided by another embodiment of the present application. The difference between Figure 9 and Figure 1 is that step S104 in Figure 1 can be specifically refined into steps S1041 to S1047 in Figure 9.

[0066] In step S1041, a third verification parameter is determined based on the positional relationship between the touch input and the randomly output facial area.

[0067] The location of the touch input can be obtained through the application programming interface (API) of the system of the device where the display screen is located. The location of the touch input can be compared with the locations of each facial region to determine whether the touch input is within the facial region. In some examples, it can be first determined whether the location of the touch input is within the facial region. If the location of the touch input is not within the facial region, it can be determined whether the location of the touch input is within the facial region including the error tolerance region. For example, the facial region is a rectangular region, the coordinates of the upper left vertex of the rectangular region are (left, top), the coordinates of the lower right vertex of the rectangular region are (right, bottom), and the coordinates of the location of the touch input are (x, y). If x-left>0, x-right<0, y-bottom>0, and y-top<0, it can be determined that the location of the touch input is within the facial region. The distance between the location of the touch input and the center point of the facial region closest to the location of the touch input can be calculated; the closer the distance, the larger the third verification parameter obtained; the farther the distance, the smaller the third verification parameter obtained. In other words, the third verification parameter is negatively correlated with the distance between the touch input and the center point of the facial region.

[0068] In step S1042 , behavior information of the touch input is acquired, and behavior features are extracted according to the behavior information.

[0069] The behavioral information of touch input can be used to describe the behavior of touch input. For example, the behavioral information of touch input may include but is not limited to one or more of the following: the time interval from outputting the facial area to receiving the touch input, the location of the touch input, the pressure of the touch input, the gesture speed of the touch input, the gesture direction of the touch input, and the duration of the touch input. If the human-machine verification method is executed by a terminal device, the terminal device with a display screen can directly obtain the behavioral information of the touch input. If the human-machine verification method is executed by a human-machine verification platform such as a server, the terminal device with a display screen obtains the behavioral information and then uploads the behavioral information to the human-machine verification platform.

[0070] Behavioral features can be extracted from behavioral information. Behavioral features can include multi-dimensional behavioral features and / or discrete data on touch pressure changes. Multi-dimensional behavioral features can comprehensively characterize the characteristics of multiple pieces of behavioral information described above, with each dimension of the multi-dimensional behavioral features representing the characteristics of one piece of behavioral information described above. Because touch input is not an instantaneous action and lasts for a period of time, the pressure value of the touch input will change during the duration of the touch input. The pressure at multiple time points within the duration of the touch input can be collected to form discrete data on pressure changes.

[0071] In step S1043 , based on the behavior feature and the pre-built historical behavior feature of the legitimate user, a target similarity parameter between the behavior feature and the historical behavior feature of the legitimate user is obtained.

[0072] The behavior information of legitimate users in the historical time period can be subjected to cluster analysis, normalization processing, fitting processing and other processing operations in advance to obtain the historical behavior characteristics of the legitimate users. Corresponding to the behavioral characteristics including multi-dimensional behavioral characteristics and / or discrete data of touch pressure changes, the historical behavior characteristics of legitimate users may include the historical multi-dimensional behavioral characteristics of legitimate users and / or the relationship between touch pressure changes of legitimate users. The historical multi-dimensional behavior characteristics of legitimate users can be extracted through cluster analysis and normalization processing based on the behavioral information of legitimate users in the historical time period. For example, the behavioral information such as the time interval from outputting the facial area to receiving the touch input, the position of the touch input, the pressure of the touch input, the gesture speed of the touch input, the gesture direction of the touch input, and the duration of the touch input can be clustered and normalized to extract the historical multi-dimensional behavior data of the legitimate users. The relationship between touch pressure changes of legitimate users can be obtained through fitting processing based on the pressure of the touch input of the legitimate users in the historical time period. For example, based on the relationship between the pressure and time of the touch input of the legitimate user in the historical time, the touch pressure change relationship of the legitimate user can be obtained through polynomial fitting. The touch pressure change relationship of the legitimate user can characterize the relationship between the pressure and time of the touch input of the legitimate user. The touch pressure change relationship of the legitimate user can be implemented as a pressure change curve or a pressure change function.

[0073] The target similarity parameter can reflect the similarity between the behavior characteristics and the historical behavior characteristics of legitimate users.

[0074] In some examples, when the behavioral characteristics include multi-dimensional behavioral characteristics and the legitimate user's historical behavioral characteristics include the legitimate user's historical multi-dimensional behavioral characteristics, the target similarity parameter includes a similarity parameter between the multi-dimensional behavioral characteristics and the legitimate user's historical multi-dimensional behavioral characteristics. For example, the multi-dimensional behavioral characteristics and the legitimate user's historical behavioral characteristics can be represented by feature vectors. The similarity parameter between the multi-dimensional behavioral characteristics and the legitimate user's historical multi-dimensional behavioral characteristics can be obtained by calculating the similarity between the feature vector corresponding to the multi-dimensional behavioral characteristics and the feature vector corresponding to the legitimate user's historical behavioral characteristics.

[0075] In some examples, when the behavior characteristics include discrete data of touch pressure changes and the legitimate user's historical behavior characteristics include the relationship between touch pressure changes of the legitimate user, the target similarity parameter includes a correlation coefficient between the discrete data of touch pressure changes and the relationship between touch pressure changes of the legitimate user. For example, the correlation coefficient between the discrete data of touch pressure changes and the relationship between touch pressure changes of the legitimate user can be obtained according to the following formula (5):

[0076] Where m is the unit time of delay; R xy (m) is the cross-correlation coefficient between x(n) and y(n) under m unit time delays. The value of the cross-correlation coefficient can be between -1 and 1. The larger the value of the cross-correlation coefficient, the higher the waveform similarity between the relationship between the discrete data of touch pressure changes and the relationship between the touch pressure changes of the legitimate user; x(n) is the discrete data of touch pressure changes corresponding to the current touch input; y(n+m) is the discrete data of historical touch pressure changes corresponding to the relationship between the touch pressure changes of the legitimate user.

[0077] In step S1044, a fourth verification parameter is determined according to the target similarity parameter.

[0078] The fourth verification parameter is positively correlated with the similarity indicated by the target similarity parameter, that is, the higher the similarity indicated by the target similarity parameter, the larger the value of the fourth verification parameter. In some examples, when the behavioral characteristics include multidimensional behavioral characteristics and discrete data of touch pressure changes, the fourth verification parameter can be calculated based on the similarity parameters between the multidimensional behavioral characteristics and the historical multidimensional behavioral characteristics of the legitimate user, and the mutual correlation coefficient between the relationship between the discrete data of touch pressure changes and the touch pressure changes of the legitimate user. For example, a weight coefficient can be set for the similarity parameter between the multidimensional behavioral characteristics and the historical multidimensional behavioral characteristics of the legitimate user, and a weight coefficient can be set for the mutual correlation coefficient between the relationship between the discrete data of touch pressure changes and the touch pressure changes of the legitimate user. The fourth verification parameter is calculated using a weighted algorithm based on the similarity parameter, the weight coefficient corresponding to the similarity parameter, the mutual correlation coefficient, and the weight coefficient corresponding to the mutual correlation coefficient.

[0079] In step S1045 , a face image during the touch input process is acquired.

[0080] Real-time panoramic facial images can be collected. If the human-machine verification method is executed by a terminal device with a display screen, the terminal device will collect the facial image and perform subsequent processing on the facial image. If the human-machine verification method is executed by a human-machine verification platform such as a server, the terminal device will collect the facial image, upload the facial image to the human-machine verification platform, and the human-machine verification platform will perform subsequent processing on the facial image.

[0081] In step S1046, the collected facial image is matched with the facial sample image of the legitimate user to obtain a matching result.

[0082] This verification operation is performed after logging in as a legitimate user. The captured facial image is matched against the legitimate user's facial sample image to determine if the operation is being performed by the user. The matching result indicates the degree of match between the captured facial image and the legitimate user's facial sample image. The higher the match, the more likely it is that the user performing the verification operation is the legitimate user. The lower the match, the less likely it is that the user performing the verification operation is the legitimate user.

[0083] In step S1047, a fifth verification parameter is obtained according to the matching result.

[0084] The fifth verification parameter is positively correlated with the matching degree represented by the matching result, that is, the higher the matching degree represented by the matching result, the greater the value of the fifth verification parameter.

[0085] The execution order of the above-mentioned steps S1041, S1042 to S1044, and S1045 to S1047 is not limited here. The process of determining the third verification parameter, the process of determining the fourth verification parameter, and the process of determining the fifth verification parameter can be independent of each other, can be performed in a certain order, or can be performed simultaneously.

[0086] In the above embodiment, the second type of verification parameters includes which one or more of the third verification parameters, the fourth verification parameters, and the fifth verification parameters, which can be set according to the corresponding business, scenario, demand, etc. of the human-machine verification. A correspondence between the difficulty mode of the human-machine verification and the content included in the second type of verification parameters can be established in advance. The difficulty mode of the human-machine verification can indicate the difficulty of the human-machine verification, and the difficulty of the human-machine verification can be related to the security risk of the human-machine verification, customer needs, etc. In some examples, the higher the security risk of the human-machine verification, the higher the difficulty of the human-machine verification. For example, the security risk of the human-machine verification of the transfer business is higher than the security risk of the human-machine verification of the information query. Correspondingly, the difficulty of the human-machine verification of the transfer business will be higher than the difficulty of the human-machine verification of the information query.

[0087] In the subsequent process, the difficulty mode of the human-machine verification can be determined based on this human-machine verification, and then the content contained in the second type of verification parameters corresponding to the difficulty mode of the human-machine verification can be determined. The corresponding verification parameters can be obtained according to the content contained in the second type of verification parameters. The higher the difficulty indicated by the difficulty mode of the human-machine verification, the more content is contained in the second type of verification parameters. For example, the difficulty indicated by the difficulty mode of the human-machine verification of the transfer business is higher than the difficulty indicated by the difficulty mode of the human-machine verification of the information query; correspondingly, the second type of verification parameters corresponding to the human-machine verification of the transfer business include the third verification parameter, the fourth verification parameter, and the fifth verification parameter, that is, it is necessary to determine the positional relationship between the touch input and the randomly output facial area, compare the behavioral characteristics with the historical behavioral characteristics of the legitimate user, and match the collected facial image with the facial sample image of the legitimate user to obtain the third verification parameter, the fourth verification parameter, and the fifth verification parameter; the second type of verification parameters corresponding to the human-machine verification of the information query include the third verification parameter and the fourth verification parameter, that is, it is necessary to determine the positional relationship between the touch input and the randomly output facial area and compare the behavioral characteristics with the historical behavioral characteristics of the legitimate user, but it is not necessary to match the collected facial image with the facial sample image of the legitimate user.

[0088] The human-machine verification method in the embodiment of the present application can effectively determine the operation of a real person, enable legitimate users to complete verification more quickly, and make it difficult for machine scripts to crack the verification, so it can accurately distinguish between the operations of legitimate users and the operations of machine scripts. When it is determined that the object performing this verification operation is a legitimate user, the object performing this verification operation is allowed to obtain business resources; when it is determined that the object performing this verification operation is a machine script, the object performing this verification operation is denied access to business resources. The human-machine verification method in the embodiment of the present application can replace the common liveness detection through shaking the head, nodding, and opening the mouth, and can better determine whether it is the operation of a legitimate user.

[0089] A second aspect of the present application provides a human-machine verification device. FIG10 is a schematic diagram of the structure of a human-machine verification device provided in one embodiment of the present application. As shown in FIG10 , the human-machine verification device 200 may include an image data acquisition module 201, a first parameter acquisition module 202, a touch data acquisition module 203, a second parameter acquisition module 204, and a user determination module 205.

[0090] The image data acquisition module 201 may be configured to acquire image frames of the display screen within a verification time period in response to at least one screen partition of the display screen output randomly.

[0091] In some examples, if the size of the display screen exceeds a preset size range, the display screen includes N1 screen partitions, where N1 is a positive integer greater than 1. If the size of the display screen is within the preset size range, the display screen includes N2 screen partitions, where N2 is a positive integer greater than 1 and N2 < N1. Two adjacent screen partitions partially overlap. If the number of randomly output screen partitions is three or more, the line connecting the center points of the three or more randomly output screen partitions is a broken line.

[0092] The first parameter acquisition module 202 may be configured to obtain a first type of verification parameter based on change information of a facial image in an image frame.

[0093] The touch data acquisition module 203 may be configured to acquire a user's touch input on a display screen in response to at least one facial region of a randomly outputted facial image.

[0094] In some examples, the touch input includes any of the following: a single finger touching a facial area; multiple fingers gathered together to touch a facial area; multiple fingers touching multiple facial areas simultaneously.

[0095] In some examples, a facial image includes feature points of identified facial organs, and a facial area corresponding to a facial organ is determined based on a feature point with a maximum coordinate value and a feature point with a minimum coordinate value among the feature points of the facial organ.

[0096] The second parameter acquisition module 204 may be configured to obtain a second type of verification parameter based on the touch input.

[0097] The user determination module 205 can be used to obtain a comprehensive verification parameter based on the first type of verification parameters and the second type of verification parameters; and, when the comprehensive verification parameter meets the preset user determination conditions, determine that the object performing this verification operation is a legitimate user.

[0098] In an embodiment of the present application, during human-machine verification, at least one screen partition of the display screen can be randomly output to instruct the user to move the facial image on the screen to the randomly output screen partition, thereby capturing image frames of the display screen during the process. Based on the changes in the facial image within the image frames, a first type of verification parameter can be obtained. During human-machine verification, at least one facial region of the facial image can also be randomly output to instruct the user to perform touch input on the randomly output facial region, thereby obtaining a second type of verification parameter based on the touch input. The randomness of the screen partitions and facial regions, as well as the required user-related operations during the human-machine verification process, are difficult to simulate and crack by machine scripts. The first type of verification parameter can reflect the likelihood that the verification operation is being performed by a legitimate user based on changes in the facial image, while the second type of verification parameter can reflect the likelihood that the verification operation is being performed by a legitimate user based on touch input. Based on the first and second type of verification parameters, a comprehensive verification parameter is obtained. This comprehensive verification parameter can accurately distinguish between legitimate user operations and machine script operations, thereby improving the accuracy of human-machine verification. Moreover, the above-mentioned human-machine verification process is mainly implemented through facial images, and there is no need for graphic design of the verification code itself. It is easier to maintain and easier to operate. It is more adaptable to the elderly and visually impaired people, and the user experience is better.

[0099] In some embodiments, the change information of the facial image in the image frame includes the change trajectory information of the facial image in the image frame. The first parameter acquisition module 202 can be specifically configured to: obtain the change trajectory information of the facial image in the image frame and the standard trajectory information of the facial image from the initial position to the randomly output screen partition; obtain an offset parameter between the change trajectory information and the standard trajectory information based on the change trajectory information and the standard trajectory information; and obtain the first type of verification parameter based on the offset parameter.

[0100] In some examples, the change trajectory information is obtained based on the position information of the center point of the facial image and the position information of the center point of the randomly output screen partition. The first parameter acquisition module 202 can be specifically configured to: obtain a first number of image frames containing facial images whose offset parameters fall within an offset threshold range, and calculate a first ratio of the first number to the total number of image frames; obtain a first verification parameter based on the first ratio, wherein the first verification parameter is positively correlated with the first ratio; and obtain a first type of verification parameter based on the first verification parameter.

[0101] In some examples, the first parameter acquisition module 202 can be specifically used to: obtain the liveness detection result of the face image in the image frame, and obtain the second verification parameter based on the liveness detection result; obtain the first type of verification parameter based on the first verification parameter and the second verification parameter.

[0102] In some embodiments, the human-machine verification device may further include a preview tuning module. The preview tuning module may be used to: in an image frame, collect a face image located in a randomly output screen partition; and determine that the face image is valid if the face image meets a face image validity condition. The face image validity condition includes: the area ratio of the largest face image in the face images in the randomly output screen partition to the screen partition exceeds a preset ratio, and / or the pitch angle of the face image in the randomly output screen partition is within a preset pitch angle range, wherein the upper limit value of the preset pitch angle range corresponding to the screen partition located below the display screen is higher than the upper limit value of the preset pitch angle range corresponding to the screen partition not located below the display screen.

[0103] In some embodiments, the second type of verification parameter includes a third verification parameter. The second parameter acquisition module 204 can be specifically configured to determine the third verification parameter based on a positional relationship between the touch input and the randomly output facial region, wherein the third verification parameter is negatively correlated with a distance between the touch input and a center point of the facial region.

[0104] In some embodiments, the second type of verification parameters includes a fourth verification parameter. The second parameter acquisition module 204 can be specifically configured to: acquire touch input behavior information, extract behavior features based on the behavior information; obtain a target similarity parameter between the behavior features and pre-built historical behavior features of legitimate users; and determine a fourth verification parameter based on the target similarity parameter, where the fourth verification parameter is positively correlated with the similarity indicated by the target similarity parameter.

[0105] In some examples, when the behavior characteristics include multidimensional behavior characteristics and the legitimate user's historical behavior characteristics include the legitimate user's historical multidimensional behavior characteristics, the target similarity parameter includes a similarity parameter between the multidimensional behavior characteristics and the legitimate user's historical multidimensional behavior characteristics. When the behavior characteristics include discrete data of touch pressure changes and the legitimate user's historical behavior characteristics include the legitimate user's touch pressure change relationship, the target similarity parameter includes a correlation coefficient between the discrete data of touch pressure changes and the legitimate user's touch pressure change relationship.

[0106] In some embodiments, the second type of verification parameters includes a fifth verification parameter. The second parameter acquisition module 204 can be specifically configured to: acquire a facial image during touch input; match the acquired facial image with a sample facial image of a legitimate user to obtain a matching result; and obtain a fifth verification parameter based on the matching result, wherein the fifth verification parameter is positively correlated with a matching degree represented by the matching result.

[0107] In a third aspect, the present application further provides an electronic device. FIG11 is a schematic diagram of the structure of an electronic device provided in one embodiment of the present application. As shown in FIG11 , the electronic device 300 includes a memory 301, a processor 302, and a computer program stored in the memory 301 and executable on the processor 302.

[0108] In some examples, the processor 302 may include a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0109] The memory 301 may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Therefore, generally, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the human-machine authentication method according to the embodiment of the present application.

[0110] The processor 302 reads the executable program code stored in the memory 301 to run a computer program corresponding to the executable program code, so as to implement the human-machine authentication method in the above embodiment.

[0111] In some examples, the electronic device 300 may further include a communication interface 303 and a bus 304. As shown in FIG11 , the memory 301, the processor 302, and the communication interface 303 are connected via the bus 304 and communicate with each other.

[0112] The communication interface 303 is mainly used to implement communication between the modules, devices, units and / or equipment in the embodiment of the present application. Input devices and / or output devices can also be connected through the communication interface 303.

[0113] Bus 304 includes hardware, software, or both that couples components of electronic device 300 to each other. By way of example, and not limitation, bus 304 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-E) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of the above. Bus 304 may include one or more buses, where appropriate. Although embodiments herein describe and illustrate a particular bus, this application contemplates any suitable bus or interconnect.

[0114] In a fourth aspect, the present application further provides a computer-readable storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, the human-machine verification method in the above-mentioned embodiment can be implemented, and the same technical effect can be achieved. To avoid repetition, the above-mentioned computer-readable storage medium may include a non-transitory computer-readable storage medium, such as a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc., which is not limited here.

[0115] An embodiment of the present application provides a computer program product. When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the human-machine verification method in the above embodiment and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0116] It should be understood that the various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. For device embodiments, equipment embodiments, and computer-readable storage medium embodiments, the relevant parts can be referred to the description part of the method embodiment. This application is not limited to the specific steps and structures described above and shown in the figures. Those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of this application. In addition, for the sake of brevity, a detailed description of known method technologies is omitted here.

[0117] Aspects of the present application have been described above with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine so that these instructions executed via the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. This processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit. It is also understood that each box in the block diagram and / or the flowchart and the combination of the boxes in the block diagram and / or the flowchart can also be implemented by the dedicated hardware that performs the specified function or action, or can be implemented by the combination of dedicated hardware and computer instructions.

[0118] Those skilled in the art should understand that the above embodiments are illustrative rather than restrictive. Different technical features appearing in different embodiments can be combined to achieve beneficial effects. Based on a study of the drawings, the specification and the claims, those skilled in the art should be able to understand and implement other variations of the disclosed embodiments. In the claims, the term "comprising" does not exclude other devices or steps; the quantifier "one" does not exclude a plurality; the terms "first" and "second" are used to identify names rather than to indicate any specific order. Any figure marks in the claims should not be understood as limiting the scope of protection. The functions of multiple parts appearing in the claims can be implemented by a separate hardware or software module. The fact that certain technical features appear in different dependent claims does not mean that these technical features cannot be combined to achieve beneficial effects.

Claims

1. A human-machine verification method, comprising: Collecting an image frame of the display screen within a verification duration in response to at least one screen partition of the randomly output display screen; Obtaining a first type of verification parameter based on the change information of the face image in the image frame; Obtaining a touch input of the user on the display screen in response to at least one facial region of the randomly output face image; Obtaining a second type of verification parameter based on the touch input; Obtaining a comprehensive verification parameter according to the first type of verification parameter and the second type of verification parameter; Determining that the object performing the current verification operation is a legitimate user when the comprehensive verification parameter meets a preset user determination condition.

2. The method according to claim 1, wherein The change information of the face image in the image frame includes the change trajectory information of the face image in the image frame. The obtaining a first type of verification parameter based on the change information of the face image in the image frame includes: Obtaining the change trajectory information of the face image in the image frame and the standard trajectory information of the face image from the initial position to the randomly output screen partition; Obtaining an offset parameter between the change trajectory information and the standard trajectory information according to the change trajectory information and the standard trajectory information; Obtaining the first type of verification parameter according to the offset parameter.

3. The method according to claim 2, wherein, The change trajectory information is obtained according to the position information of the center point of the face image and the position information of the center point of the randomly output screen partition. The obtaining the first type of verification parameter according to the offset parameter includes: Obtaining a first quantity of the image frames where the face images with the offset parameter within the offset threshold range are located, and calculating a first ratio of the first quantity to the total quantity of the image frames; Obtaining a first verification parameter according to the first ratio, and the first verification parameter has a positive correlation with the first ratio; Obtaining the first type of verification parameter based on the first verification parameter.

4. The method according to claim 3, wherein, The obtaining the first type of verification parameter based on the first verification parameter includes: Obtaining a live detection result of the face image in the image frame, and obtaining a second verification parameter according to the live detection result; Obtaining the first type of verification parameter according to the first verification parameter and the second verification parameter.

5. The method according to claim 1, wherein When the size of the display screen exceeds a preset size range, the display screen includes N1 screen partitions, and N1 is a positive integer greater than 1; When the size of the display screen is within the preset size range, the display screen includes N2 screen partitions, and N2 is a positive integer greater than 1 and N2 < N1; Wherein, two adjacent screen partitions partially overlap; When the number of randomly output screen partitions is more than three, the connection line of the center points of the more than three randomly output screen partitions is a broken line.

6. The method according to claim 1, further comprising: Collecting a face image located in the randomly output screen partition in the image frame; Determining that the face image is valid when the face image meets the face image valid condition; Wherein, the face image valid condition includes: The ratio of the area of the largest face image in the face images in the randomly output screen partition to the area of the screen partition exceeds a preset ratio, and / or, the pitch angle of the face image in the randomly output screen partition is within a preset pitch angle range. Wherein, the upper limit value of the preset pitch angle range corresponding to the screen partition located below the display screen is higher than the upper limit value of the preset pitch angle range corresponding to the screen partition not located below the display screen.

7. The method according to claim 1, wherein The second type of verification parameter includes a third verification parameter. Obtaining the second type of verification parameter based on the touch input includes: Determining a third verification parameter according to the positional relationship between the touch input and the randomly output facial area, and the third verification parameter has a negative correlation with the distance between the touch input and the center point of the facial area.

8. The method according to claim 1, wherein The second type of verification parameter includes a fourth verification parameter. Obtaining the second type of verification parameter based on the touch input includes: Obtaining the behavior information of the touch input, and extracting behavior features according to the behavior information; Obtaining a target similarity parameter between the behavior features and the pre-constructed legal user historical behavior features according to the behavior features; Determining the fourth verification parameter according to the target similarity parameter, and the fourth verification parameter has a positive correlation with the similarity indicated by the target similarity parameter.

9. The method according to claim 8, wherein In the case where the behavior features include multi-dimensional behavior features and the legal user historical behavior features include legal user historical multi-dimensional behavior features, the target similarity parameter includes the similarity parameter between the multi-dimensional behavior features and the legal user historical multi-dimensional behavior features; In the case where the behavior features include discrete data of touch pressure change and the legal user historical behavior features include the touch pressure change relationship of the legal user, the target similarity parameter includes the cross-correlation coefficient between the discrete data of touch pressure change and the touch pressure change relationship of the legal user.

10. The method according to claim 1, wherein The second type of verification parameter includes a fifth verification parameter. Obtaining the second type of verification parameter based on the touch input includes: Obtaining a face image during the process of the touch input; Matching the collected face image with the face sample image of the legal user to obtain a matching result; Obtaining the fifth verification parameter according to the matching result, and the fifth verification parameter has a positive correlation with the matching degree characterized by the matching result.

11. The method according to claim 1, wherein, The touch input includes any one of the following: Single-finger touching a facial area; Multiple fingers gathering and touching a facial area; Multiple fingers simultaneously touching multiple facial areas respectively.

12. The method according to claim 1, wherein The face image includes feature points of the recognized facial organs, and the facial area corresponding to one facial organ is determined based on the feature point with the largest coordinate value and the feature point with the smallest coordinate value among the feature points of the one facial organ.

13. A human-machine verification device, comprising: An image data acquisition module, configured to collect image frames of the display screen during a verification duration in response to at least one screen partition of the randomly output display screen. A first parameter acquisition module, configured to obtain a first type of verification parameter based on change information of a face image in the image frame; A touch data acquisition module, configured to obtain a touch input of a user on the display screen in response to at least one facial region of a randomly output face image; A second parameter acquisition module, configured to obtain a second type of verification parameter based on the touch input; A user determination module, configured to obtain a comprehensive verification parameter according to the first type of verification parameter and the second type of verification parameter; and determine that the object performing the current verification operation is a legitimate user when the comprehensive verification parameter meets a preset user determination condition.

14. An electronic device, comprising: A processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the human-machine verification method according to any one of claims 1 to 12 is implemented.

15. A computer-readable storage medium, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the human-machine verification method according to any one of claims 1 to 12 is implemented.

Images