Online examination behavior monitoring method and system based on privacy protection

Abstract

The invention discloses an online examination behavior monitoring method and system based on privacy protection, and belongs to the technical field of online education, and the method comprises the steps: S10, obtaining related data of a target examinee, the related data comprising a head image of the target examinee and an image of an object associated with an online examination of the target examinee, and storing the head image of the target examinee and the image of the object associated with the online examination of the target examinee; the image comprises an image of the computer display screen, acquiring four corner points and a central axis of the computer display screen based on the image of the computer display screen, and generating a two-dimensional image about the head image, the four corner points and the central axis of the target examinee. According to the method, the posture abnormity and plagiarism behaviors of the examinee are accurately detected by combining real-time monitoring of the head image of the examinee and the image of the computer display screen, meanwhile, the accuracy and reliability of behavior recognition are effectively improved by utilizing multi-dimensional data analysis and scientific calculation, and while the privacy of the examinee is protected, the accuracy and reliability of behavior recognition are improved. And the fairness and the seriousness of the online examination are ensured.

Description

Technical Field

[0001] This invention relates to the field of online education technology, and in particular to a method and system for monitoring online examination behavior based on privacy protection. Background Technology

[0002] With the popularization of online education, video streams, screen streams, and audio streams from online examination devices can accurately capture the actions and behaviors of individual candidates, thereby enabling the detection of whether candidates are suspected of cheating in the examination room.

[0003] Regarding this research, application CN202311610864.0 provides a method, device, equipment, and storage medium for monitoring and preventing cheating in online examinations. The technical solution includes: when a candidate's terminal receives a candidate's operation event, determining whether the event is a sensitive event; if the event is sensitive, determining whether it is a compliant event based on the operation request information; if it is compliant, responding to the event; and if it is not compliant, prohibiting response. This technical solution provides real-time monitoring of the human-computer interaction process between the candidate and their terminal during online examinations, enabling comprehensive monitoring of high-incidence cheating methods in online examination scenarios, thus ensuring the security and fairness of the examination.

[0004] Another application, CN202510206240.5, provides an online examination behavior detection method based on edge computing and multimodal fusion. This technical solution includes: identifying the initial group of suspected group cheating based on the co-occurrence of cheating behaviors at different times within the examination room; constructing a reachability matrix with each target candidate in the initial group as rows and columns; if any row has visual reachability to any target candidate in any column, setting the matrix elements of that row and column to 1; identifying rows or columns with all empty elements, and for each empty row or column: defining a neighborhood A within the examination room centered on the current target candidate; assigning different first weights to each candidate within the visual reachability neighborhood A, and assigning second weights to each candidate based on the nearest reachability distance between each candidate within neighborhood A and other target candidates in the reachability matrix; this technical solution can automatically identify group cheating behavior.

[0005] However, the aforementioned technical solutions collect data through full-screen monitoring and camera recording throughout the process, which may lead to the leakage of sensitive data such as candidates' biometric characteristics (face, voiceprint), posing a risk of privacy breach. Summary of the Invention

[0006] In view of the problems existing in the field of online education technology, the present invention is proposed.

[0007] Therefore, one of the objectives of this invention is to provide a privacy-protected online examination behavior monitoring method and system. By combining real-time monitoring of the examinee's head image and the computer screen image, it accurately detects abnormal posture and plagiarism behavior of the examinee. At the same time, by utilizing multi-dimensional data analysis and scientific calculation, it effectively improves the accuracy and reliability of behavior recognition, ensuring the fairness and seriousness of online examinations while protecting the examinee's privacy.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution: On the one hand, this invention provides a privacy-protected online examination behavior monitoring method, comprising the following steps: S10: Obtain relevant data about the target candidate, including a head image of the target candidate and images of objects associated with the online exam of the target candidate, including an image of a computer screen. Based on the image of the computer screen, obtain the four corner points and the central axis of the computer screen, and generate a two-dimensional image of the target candidate's head image, the four corner points, and the central axis. S20: Obtain the height of the central axis and perform real-time monitoring of the target examinee based on the two-dimensional image. The real-time monitoring includes real-time monitoring of vision and screen. In the real-time monitoring of vision, the height change of the target examinee's head is collected based on the central axis. The collection steps are as follows: The monitoring height range is divided based on the middle of the central axis, including the height range corresponding to 10cm above the middle of the central axis. Within the monitored height range, the height change of the target candidate's head is calculated in a collection period of 1 to 2 seconds. If the height of the target candidate's head moves toward the middle of the central axis within 3 consecutive collection periods, the target candidate is determined to be in an abnormal posture state and an early warning is issued; otherwise, no judgment is made. S30: When it is determined that the target examinee is in an abnormal posture state, the changes in the page display on the computer screen are obtained through real-time monitoring of the screen. The steps for obtaining this information are as follows: When it is determined that the target candidate is in an abnormal posture state, the changes in the page display on the computer screen are collected. The page display changes include the questions displayed on the computer screen when it is determined that the target candidate is in an abnormal posture state, and the questions include unanswered questions. Obtain the question type corresponding to the question. When the height of the target candidate's head moves towards the middle away from the central axis, collect the page display changes on the computer screen. The page display changes include whether the target candidate answers the question. Mark the question type as a reference question type. S40: If the target candidate answers the question while moving their head away from the center axis, it is determined that the target candidate has committed plagiarism and a warning is issued; otherwise, no judgment is made.

[0009] In a preferred embodiment of the present invention, the height change of the target examinee's head is calculated in the monitored height range at 1-2 second intervals, and the result is obtained according to the following formula: ; In the formula, Indicates the first The change in head height of the target examinee between the previous and current data collection periods; Indicated at the The height of the target examinee's head during each data collection period; Indicated at the -1 The height of the target candidate's head during the sampling period.

[0010] In a preferred embodiment of the present invention, the following formula is also included: ; In the formula, Indicates the first The change in head height of the target examinee between the previous and current data collection periods; Indicates the number of consecutive data collection periods; express Average height change rate during each data collection period.

[0011] In a preferred embodiment of the present invention, the following is determined: based on the calculation results, the rate at which the target examinee's head moves towards the midpoint of the central axis during the first acquisition time period is obtained during three consecutive acquisition time periods, and is calculated using the following formula: ; In the formula, This indicates the rate at which the target examinee's head moves towards the middle of the central axis during the first data collection period; This indicates the height of the target examinee's head from the midpoint of the central axis at the start of the first data collection period; This indicates the height of the target examinee's head from the midpoint of the central axis at the end of the first data collection period; This indicates the duration of the first data collection period.

[0012] In a preferred embodiment of the present invention, the calculated rate is marked as a reference rate. When a question type corresponding to the reference question type is obtained on the computer screen at a future time, the height change of the target candidate's head is obtained. If the height change is the same as the reference rate, the target candidate is determined to be in an abnormal posture state and an early warning is issued; otherwise, no determination is made.

[0013] In a preferred embodiment of the present invention, the four corner points are divided into left and right corner points based on the central axis. When the target candidate is not determined to be in an abnormal posture state, the change of the target candidate's head is obtained. The change includes calculating the distance between the target candidate's head and the left and right corner points. When the distance between the target candidate's head and the left or right corner point shows a shortening trend, it is determined that the target candidate's head is deviating towards the corresponding corner point, and the target candidate is determined to be in an abnormal posture state, and a warning is issued; otherwise, no determination is made.

[0014] In a preferred embodiment of the present invention, when it is determined that the target examinee's head is deviating from the corner of the corresponding side, relevant distance data is collected. The distance data is divided into Class I data and Class II data. The Class I data is collected according to the reference rate. The collection steps are as follows: Based on the height of the target candidate's head from the midpoint of the central axis at the start of the first collection period, the height is divided into at least 3 height points at 0.5 seconds per time node, and the distance between adjacent height points is obtained. The steps for collecting Type II data are as follows: When the distance between the target candidate's head and the left or right corner point is decreasing, obtain the maximum range of movement of the target candidate's head toward the left or right corner point during this process. When the height change of the target candidate's head is acquired at a future moment, given an initial acquisition time point, the height change is acquired at the initial acquisition time point. If the height change is the same as either the maximum movement amplitude or the distance between the first and second height points, it is determined that the target candidate's head is deflected to the corner point on the corresponding side or the target candidate's head is moved towards the middle of the central axis.

[0015] In a preferred embodiment of the present invention, when obtaining the height change of the target candidate's head, if the target candidate's head deviates to the corner point of the corresponding side but does not reach the maximum movement range, the distance of the target candidate's head deviating to the corner point of the corresponding side is calculated. If the distance exceeds two-thirds of the maximum movement range, it is determined that the target candidate's head deviates to the corner point of the corresponding side.

[0016] On the other hand, the present invention provides a system for use in a privacy-preserving online examination behavior monitoring method as described above, comprising: The data acquisition module is used to acquire relevant data about the target candidate. The relevant data includes a head image of the target candidate and images of objects associated with the online exam of the target candidate. The images include images of computer screens. Based on the images of the computer screens, the module acquires the four corner points and the central axis of the computer screen and generates two-dimensional images of the target candidate's head image, the four corner points, and the central axis. The data processing module includes an acquisition unit and a calculation unit; The acquisition unit is used to obtain the height of the central axis and to perform real-time monitoring of the target examinee based on the two-dimensional image. The real-time monitoring includes real-time monitoring of vision and screen. In the real-time monitoring of vision, the height change of the target examinee's head is acquired based on the central axis. The acquisition steps are as follows: The monitoring height range is divided based on the middle of the central axis, including the height range corresponding to 10cm above the middle of the central axis. The calculation unit is used to calculate the height change of the target candidate's head in the monitoring height range at 1 to 2 seconds per collection period. If the height of the target candidate's head moves toward the middle of the central axis within 3 consecutive collection periods, the target candidate is determined to be in an abnormal posture state and an early warning is issued; otherwise, no determination is made. The display monitoring module is used to capture changes in the computer screen's display in real-time when the target examinee is determined to be in an abnormal posture state. The steps for capturing these changes are as follows: When it is determined that the target candidate is in an abnormal posture state, the changes in the page display on the computer screen are collected. The page display changes include the questions displayed on the computer screen when it is determined that the target candidate is in an abnormal posture state, and the questions include unanswered questions. Obtain the question type corresponding to the question. When the height of the target candidate's head moves towards the middle away from the central axis, collect the page display changes on the computer screen. The page display changes include whether the target candidate answers the question. Mark the question type as a reference question type. The warning module is used to determine that the target candidate has committed plagiarism and issue a warning when the target candidate answers the question while the height of the target candidate's head is moving away from the center axis; otherwise, no judgment is made.

[0017] Beneficial effects: 1. By monitoring the candidate's head image and the computer screen image, the examination behavior can be monitored while protecting the candidate's privacy. This avoids the privacy infringement issues caused by direct monitoring and improves the acceptance and compliance of the examination system. At the same time, by acquiring the four corner points and the central axis of the computer screen and generating two-dimensional images for real-time monitoring, the visual and screen behavior of the candidates can be efficiently captured, ensuring the fairness and seriousness of the examination. 2. By monitoring changes in the height of the examinee's head, especially the movement of the head toward the midpoint of the central axis, the system can accurately determine whether the examinee is in an abnormal posture state. For example, when the examinee's head moves toward the midpoint of the central axis for three consecutive collection periods, the system will issue an early warning, effectively preventing the examinee from cheating. Furthermore, by calculating the rate at which the examinee's head moves toward the midpoint of the central axis and comparing the rate of head height change when the same question type occurs in the future, the accuracy of abnormal posture detection is further improved. 3. When the system determines that the candidate is in an abnormal posture, it will further capture the changes in the computer screen display, including unanswered questions and question types. By comparing the changes in the candidate's head height with the answers to the questions, the system can intelligently identify whether the candidate has engaged in plagiarism. 4. By calculating the distance between the candidate's head and the left and right corners of the computer screen, the system can monitor the candidate's head tilting behavior in real time. When the distance between the candidate's head and a certain corner point shows a shortening trend, the system will determine that the candidate's head is tilted towards the corresponding corner point and issue an early warning, effectively preventing the candidate from cheating due to tilting behavior. 5. The system divides the collected distance data into Class I and Class II data, and collects and analyzes them based on the reference rate and maximum movement amplitude, respectively. This data classification and refined analysis method enables the system to more accurately judge the candidate's behavior pattern and improve the accuracy and reliability of cheating behavior identification. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a schematic diagram of the modular structure of the privacy-protected online examination behavior monitoring system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the process structure of an embodiment of the present invention; The diagram is labeled as follows: 110 - Data acquisition module; 120 - Data processing module; 1201 - Acquisition unit; 1202 - Calculation unit; 130 - Display and monitoring module; 140 - Early warning module. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0020] Because existing technologies collect data through full-screen monitoring and camera recording, they are prone to leaking sensitive data such as candidates' biometric features (face, voiceprint), posing a risk of privacy breaches.

[0021] Based on this, the present invention proposes an online examination behavior monitoring method and system based on privacy protection. By combining real-time monitoring of the candidate's head image and the computer screen image, it can accurately detect abnormal posture and plagiarism behavior of the candidate, thus protecting the candidate's privacy while ensuring the fairness and seriousness of the online examination.

[0022] The present solution will be further described in detail below through embodiments and in conjunction with the accompanying drawings.

[0023] Reference Figures 1 to 2 As one embodiment of the present invention, this embodiment provides a privacy-protected online examination behavior monitoring method, comprising the following steps: S10: Obtain relevant data about the target candidate, including the head image of the target candidate and images of objects associated with the online exam of the target candidate, including images of the computer screen. Based on the images of the computer screen, obtain the four corner points and the central axis of the computer screen, and generate two-dimensional images of the head image, the four corner points and the central axis of the target candidate. In this embodiment, by acquiring only the necessary image information, the privacy of the examinee is protected by avoiding direct monitoring of the examinee's face or the exam content. It should be noted that the head image of the target candidate was obtained from the back of the target candidate, not from the front; This provides foundational data and a reference framework for subsequent monitoring, ensuring the accuracy and effectiveness of the monitoring. S20: Obtain the height of the central axis and perform real-time monitoring of the target examinee based on the two-dimensional image. Real-time monitoring includes real-time monitoring of vision and screen. In the real-time monitoring of vision, the height change of the target examinee's head is collected based on the central axis. The collection steps are as follows: The monitoring height range is divided based on the middle of the central axis, including the height range corresponding to 10cm above the middle of the central axis. Within the monitored height range, the height change of the target candidate's head is calculated in a collection period of 1 to 2 seconds. If the height of the target candidate's head moves toward the middle of the central axis within 3 consecutive collection periods, the target candidate is determined to be in an abnormal posture state and an early warning is issued; otherwise, no judgment is made. In this embodiment, if the height of the target candidate's head moves toward the middle of the central axis during three consecutive collection periods, it is considered that the target candidate is looking down at an object, thereby triggering an early warning. By analyzing height changes over consecutive time periods, the accuracy of judging abnormal posture can be improved, which can also promptly detect abnormal behavior of candidates and prevent cheating; at the same time, it sets clear judgment criteria (three consecutive time periods) to reduce the false judgment rate. S30: When it is determined that the target examinee is in an abnormal posture state, the changes in the page display on the computer screen are obtained through real-time monitoring of the screen. The steps for obtaining this information are as follows: When it is determined that the target candidate is in an abnormal posture state, the changes in the page display on the computer screen are collected. The page display changes include the questions displayed on the computer screen when it is determined that the target candidate is in an abnormal posture state, including unanswered questions. Obtain the question type corresponding to the question. When the height of the target candidate's head moves towards the middle away from the central axis, collect the changes in the page display on the computer screen. The page display changes include whether the target candidate answers the question. Mark the question type as a reference question type. S40: If the target candidate answers the question while moving their head away from the center line, it is determined that the target candidate has committed plagiarism and a warning is issued; otherwise, no judgment is made. In this embodiment, when an abnormal posture is determined, changes in the computer screen display are captured (e.g., unanswered questions); and the question type corresponding to the question is obtained. When the head height moves away from the center of the central axis, changes in the page display are captured (e.g., whether the question has been answered) and marked as a reference question type. If the question is answered while the head is moving away from the center of the axis, it is judged as plagiarism and a warning is issued. This method, which links changes in head height to changes in screen display, improves the accuracy of cheating detection. It can also effectively identify candidates using abnormal postures to copy, thus maintaining the fairness of the exam.

[0024] Within the monitored height range, the change in the target examinee's head height is calculated in 1-2 second intervals, using the following formula: ; In the formula, Indicates the first The change in head height of the target examinee between the first and previous data collection periods (the first data collection period) -1 data collection period, when When =1, set an initial height. (For reference only) Indicated at the The height of the target examinee's head during each data collection period; Indicated at the - The height of the target examinee's head during one data collection period; Furthermore, it also includes calculations based on the following formula: ; In the formula, Indicates the first The change in head height of the target examinee between the previous and current data collection periods; This indicates the number of consecutive data collection periods (here) =3); express Average height change rate during each data collection period; It provides a formula for calculating changes in head height, which quantifies the height change between each data collection period and the previous period. This converts the height change into a specific value, facilitating subsequent analysis and judgment. It can also provide a scientific basis for judging abnormal posture and plagiarism, improving the objectivity and accuracy of the judgment. Calculating the average height change rate and understanding the overall trend of changes in the examinee's head height provides auxiliary information for judging abnormal posture and plagiarism, improving the comprehensiveness and accuracy of the judgment.

[0025] Based on the above, and according to the calculation results, the rate at which the target examinee's head moves towards the midpoint of the central axis is obtained in the first collection period out of three consecutive collection periods, and is calculated using the following formula: ; In the formula, This indicates the rate at which the target examinee's head moves towards the middle of the central axis during the first data collection period; This indicates the height of the target examinee's head from the midpoint of the central axis at the start of the first data collection period; This indicates the height of the target examinee's head from the midpoint of the central axis at the end of the first data collection period; This indicates the duration of the first data collection period (given that each data collection period is 1-2 seconds, therefore...). Take a fixed value, for example =1 second or 2 seconds); The calculated rate is marked as the reference rate. When a question type corresponding to the reference question type is obtained on the computer screen at a future time, the height change of the target candidate's head is obtained. If the height change is the same as the reference rate, the target candidate is determined to be in an abnormal posture state and an early warning is issued; otherwise, no judgment is made. In this embodiment, the calculation rate is marked as the reference rate for comparison of head height changes when the same question type appears in the future. By calculating the head movement rate, we can understand how fast the examinee's head moves. By comparing it with a reference rate, we can promptly detect abnormal behavior and issue warnings, thereby improving the sensitivity and accuracy of monitoring.

[0026] Based on the above, the four corner points are divided into left and right corner points according to the central axis. When the target candidate is not determined to be in an abnormal posture state, the changes in the target candidate's head are obtained. The changes include calculating the distance between the target candidate's head and the left and right corner points. When the distance between the target candidate's head and the left or right corner point shows a shortening trend, it is determined that the target candidate's head is deviating towards the corresponding corner point, and the target candidate is determined to be in an abnormal posture state, and an early warning is issued; otherwise, no judgment is made. Furthermore, when it is determined that the target examinee's head is facing a corner to the corresponding side, relevant distance data is collected. The distance data is divided into Class I data and Class II data. Class I data is collected according to the reference rate. The collection steps are as follows: Based on the height of the target candidate's head from the midpoint of the central axis at the start of the first collection period, the height is divided into at least 3 height points at 0.5 seconds per time node, and the distance between adjacent height points is obtained. The steps for collecting Type II data are as follows: When the distance between the target candidate's head and the left or right corner point is decreasing, obtain the maximum range of movement of the target candidate's head toward the left or right corner point during this process. When the height change of the target candidate's head is acquired at a future moment, given an initial acquisition time point, the height change is acquired at the initial acquisition time point. If the height change is the same as either the maximum movement amplitude or the distance between the first and second height points, it is determined that the target candidate's head is deflected to the corner point on the corresponding side or the target candidate's head is moved towards the middle of the central axis. In this embodiment, the behavior of candidates tilting their heads to one side can be effectively identified to prevent cheating due to tilting. At the same time, the accuracy and reliability of behavior recognition can be improved through data classification and refined analysis. Furthermore, verification is carried out by combining multi-dimensional information such as changes in head height and changes in corner spacing to reduce false alarm and false negative rates.

[0027] Furthermore, when obtaining the height change of the target candidate's head, if the target candidate's head deviates to the corner point on the corresponding side but does not reach the maximum movement range, the distance of the target candidate's head deviating to the corner point on the corresponding side is calculated. If the distance exceeds two-thirds of the maximum movement range, it is determined that the target candidate's head deviates to the corner point on the corresponding side. When the head deviates slightly to the side of the corresponding corner but does not reach the maximum range of movement, the offset distance is calculated. This can effectively prevent candidates from escaping monitoring due to slight deviation and maintain the fairness of the examination.

[0028] As can be seen from the above, this application, by combining real-time monitoring of the candidate's head image and the computer screen image, accurately detects abnormal posture and plagiarism behavior of the candidate, thus protecting the candidate's privacy while ensuring the fairness and seriousness of the online examination.

[0029] This embodiment, in conjunction with the above-mentioned privacy-preserving online examination behavior monitoring method, also proposes a working system applied to this method, as follows: The data acquisition module 110 is used to acquire relevant data about the target candidate. The relevant data includes the head image of the target candidate and the image of the object associated with the online exam of the target candidate. The image includes the image of the computer screen. Based on the image of the computer screen, the module acquires the four corner points and the central axis of the computer screen and generates a two-dimensional image of the head image, the four corner points and the central axis of the target candidate. Data processing module 120, which includes acquisition unit 1201 and calculation unit 1202; The acquisition unit 1201 is used to acquire the height of the central axis and to perform real-time monitoring of the target examinee based on the two-dimensional image. The real-time monitoring includes real-time monitoring of vision and screen. In the real-time monitoring of vision, the height change of the target examinee's head is acquired based on the central axis. The acquisition steps are as follows: The monitoring height range is divided based on the middle of the central axis, including the height range corresponding to 10cm above the middle of the central axis. The calculation unit 1202 is used to calculate the height change of the target candidate's head in the monitoring height range at 1 to 2 seconds as a collection period. If the height of the target candidate's head moves toward the middle of the central axis in 3 consecutive collection periods, it is determined that the target candidate is in an abnormal posture state and an early warning is issued; otherwise, no judgment is made. The display monitoring module 130 is used to obtain changes in the computer screen display in real-time monitoring when it is determined that the target examinee is in an abnormal posture state. The steps for obtaining the changes are as follows: When it is determined that the target candidate is in an abnormal posture state, the changes in the page display on the computer screen are collected. The page display changes include the questions displayed on the computer screen when it is determined that the target candidate is in an abnormal posture state, including unanswered questions. Obtain the question type corresponding to the question. When the height of the target candidate's head moves towards the middle away from the central axis, collect the changes in the page display on the computer screen. The page display changes include whether the target candidate answers the question. Mark the question type as a reference question type. The warning module 140 is used to determine that the target candidate has committed plagiarism and issue a warning when the target candidate answers the question while the height of the target candidate's head is moving away from the center axis; otherwise, no judgment is made.

[0030] In summary, by combining real-time monitoring of candidates' head images with images on computer screens, abnormal postures and plagiarism can be accurately detected. Furthermore, by utilizing multi-dimensional data analysis and scientific calculation formulas, the accuracy and reliability of behavior recognition are effectively improved, ensuring the fairness and seriousness of online examinations while protecting candidates' privacy.

[0031] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A privacy-preserving method for monitoring online examination behavior, characterized in that, Includes the following steps: S10: Obtain relevant data about the target candidate, including a head image of the target candidate and images of objects associated with the online exam of the target candidate, including an image of a computer screen. Based on the image of the computer screen, obtain the four corner points and the central axis of the computer screen, and generate a two-dimensional image of the target candidate's head image, the four corner points, and the central axis. S20: Obtain the height of the central axis and perform real-time monitoring of the target examinee based on the two-dimensional image. The real-time monitoring includes real-time monitoring of vision and screen. In the real-time monitoring of vision, the height change of the target examinee's head is collected based on the central axis. The collection steps are as follows: The monitoring height range is divided based on the middle of the central axis, including the height range corresponding to 10cm above the middle of the central axis. Within the monitored height range, the height change of the target candidate's head is calculated in a collection period of 1 to 2 seconds. If the height of the target candidate's head moves toward the middle of the central axis within 3 consecutive collection periods, the target candidate is determined to be in an abnormal posture state and an early warning is issued; otherwise, no judgment is made. S30: When it is determined that the target examinee is in an abnormal posture state, the changes in the page display on the computer screen are obtained through real-time monitoring of the screen. The steps for obtaining this information are as follows: When it is determined that the target candidate is in an abnormal posture state, the changes in the page display on the computer screen are collected. The page display changes include the questions displayed on the computer screen when it is determined that the target candidate is in an abnormal posture state, and the questions include unanswered questions. Obtain the question type corresponding to the question. When the height of the target candidate's head moves towards the middle away from the central axis, collect the page display changes on the computer screen. The page display changes include whether the target candidate answers the question. Mark the question type as a reference question type. S40: If the target candidate answers the question while moving their head away from the center axis, it is determined that the target candidate has committed plagiarism and a warning is issued. Conversely, no judgment is made.

2. The method for monitoring online examination behavior based on privacy protection as described in claim 1, characterized in that, Within the monitored height range, the change in the target examinee's head height is calculated in 1-2 second intervals, using the following formula: ； In the formula, Indicates the first The change in head height of the target examinee between the previous and current data collection periods; Indicated at the The height of the target examinee's head during each data collection period; Indicated at the -1 The height of the target candidate's head during the sampling period.

3. The method for monitoring online examination behavior based on privacy protection as described in claim 2, characterized in that, It also includes calculations based on the following formula: ； In the formula, Indicates the first The change in head height of the target examinee between the previous and current data collection periods; Indicates the number of consecutive data collection periods; express Average height change rate during each data collection period.

4. A privacy-protected online examination behavior monitoring method as described in any one of claims 2 to 3, characterized in that, Based on the calculation results, the rate at which the target examinee's head moves towards the midpoint of the central axis during the first acquisition time period is obtained in the first acquisition time period out of three consecutive acquisition time periods, and is calculated according to the following formula: ； In the formula, This indicates the rate at which the target examinee's head moves towards the middle of the central axis during the first data collection period; This indicates the height of the target examinee's head from the midpoint of the central axis at the start of the first data collection period; This indicates the height of the target examinee's head from the midpoint of the central axis at the end of the first data collection period; This indicates the duration of the first data collection period.

5. The method for monitoring online examination behavior based on privacy protection as described in claim 4, characterized in that, The calculated rate is marked as the reference rate. When a question type corresponding to the reference question type is obtained on the computer screen at a future time, the height change of the target candidate's head is obtained. If the height change is the same as the reference rate, the target candidate is determined to be in an abnormal posture state and an early warning is issued; otherwise, no determination is made.

6. The method for monitoring online examination behavior based on privacy protection as described in claim 5, characterized in that, Based on the central axis, the four corner points are divided into left corner point and right corner point. When the target candidate is not determined to be in an abnormal posture state, the change of the target candidate's head is obtained. The change includes calculating the distance between the target candidate's head and the left corner point and the right corner point. When the distance between the target candidate's head and the left corner point or the right corner point shows a shortening trend, it is determined that the target candidate's head is deviating towards the corresponding corner point, and the target candidate is determined to be in an abnormal posture state, and an early warning is issued. Conversely, no judgment is made.

7. The method for monitoring online examination behavior based on privacy protection as described in claim 6, characterized in that, When it is determined that the target candidate's head is deviating from the corner on the corresponding side, relevant distance data is collected. The distance data is divided into Class I data and Class II data. Class I data is collected according to the reference rate. The collection steps are as follows: Based on the height of the target candidate's head from the midpoint of the central axis at the start of the first collection period, the height is divided into at least 3 height points at 0.5 seconds per time node, and the distance between adjacent height points is obtained. The steps for collecting Type II data are as follows: When the distance between the target candidate's head and the left or right corner point is decreasing, obtain the maximum range of movement of the target candidate's head toward the left or right corner point during this process. When the height change of the target candidate's head is acquired at a future moment, given an initial acquisition time point, the height change is acquired at the initial acquisition time point. If the height change is the same as either the maximum movement amplitude or the distance between the first and second height points, it is determined that the target candidate's head is deflected to the corner point on the corresponding side or the target candidate's head is moved towards the middle of the central axis.

8. The method for monitoring online examination behavior based on privacy protection as described in claim 7, characterized in that, When obtaining the height change of the target candidate's head, if the target candidate's head deviates to the corner point on the corresponding side but does not reach the maximum movement range, the distance of the target candidate's head deviating to the corner point on the corresponding side is calculated. If the distance exceeds two-thirds of the maximum movement range, it is determined that the target candidate's head deviates to the corner point on the corresponding side.

9. A system applied to the privacy-protected online examination behavior monitoring method as described in claim 1, characterized in that, include: The data acquisition module is used to acquire relevant data about the target candidate. The relevant data includes a head image of the target candidate and images of objects associated with the online exam of the target candidate. The images include images of computer screens. Based on the images of the computer screens, the module acquires the four corner points and the central axis of the computer screen and generates two-dimensional images of the target candidate's head image, the four corner points, and the central axis. The data processing module includes an acquisition unit and a calculation unit; The acquisition unit is used to obtain the height of the central axis and to perform real-time monitoring of the target examinee based on the two-dimensional image. The real-time monitoring includes real-time monitoring of vision and screen. In the real-time monitoring of vision, the height change of the target examinee's head is acquired based on the central axis. The acquisition steps are as follows: The monitoring height range is divided based on the middle of the central axis, including the height range corresponding to 10cm above the middle of the central axis. The calculation unit is used to calculate the height change of the target candidate's head in the monitoring height range at 1 to 2 seconds per collection period. If the height of the target candidate's head moves toward the middle of the central axis within 3 consecutive collection periods, the target candidate is determined to be in an abnormal posture state and an early warning is issued; otherwise, no determination is made. The display monitoring module is used to capture changes in the computer screen's display in real-time when the target examinee is determined to be in an abnormal posture state. The steps for capturing these changes are as follows: When it is determined that the target candidate is in an abnormal posture state, the changes in the page display on the computer screen are collected. The page display changes include the questions displayed on the computer screen when it is determined that the target candidate is in an abnormal posture state, and the questions include unanswered questions. Obtain the question type corresponding to the question. When the height of the target candidate's head moves towards the middle away from the central axis, collect the page display changes on the computer screen. The page display changes include whether the target candidate answers the question. Mark the question type as a reference question type. The early warning module is used to determine that the target candidate has committed plagiarism and issue an early warning when the target candidate answers the question while the height of the target candidate's head is moving away from the center axis. Conversely, no judgment is made.

Images