Campus personnel access process optimization management method and system

Abstract

The application relates to the technical field of entrance and exit control management, and discloses a campus personnel entrance and exit process optimization management method and system, which comprises the following steps: determining credit evaluation data of campus personnel based on historical passing behavior records; acquiring congestion state parameters of a passing channel in real time, and determining whether the congestion state parameters satisfy preset trigger conditions; if the trigger conditions are satisfied, a differentiated verification mode is enabled for the passing channel; in the differentiated verification mode, the credit evaluation data of the current campus personnel is used to identify the credit grade corresponding to the campus personnel; the credit grade is used to match a corresponding verification process intensity, the identity of the passing campus personnel is verified according to the matched verification process intensity, and the release state is controlled according to the verification result; and the congestion state parameters are continuously monitored, and when preset recovery conditions are satisfied, the differentiated verification mode is closed. The application greatly improves the passing efficiency during peak periods and effectively relieves congestion under the premise of ensuring safety.

Description

Technical Field

[0001] This application relates to the field of access control and management technology, and more specifically, to a method and system for optimizing the access process of personnel on campus. Background Technology

[0002] The dual pressures of managing pedestrian flow and security verification at school entrances and exits, especially when dealing with sudden surges in visitor numbers, reveal significant technical shortcomings in existing management methods. Conventional entrance and exit control systems rely heavily on fixed schedules or on-site judgment by security personnel to adjust the opening status of passages. This model can barely cope with predictable peak periods for classes, but it becomes sluggish when faced with unexpected surges in visitor numbers caused by temporary events, weather changes, or off-campus traffic conditions. When a large number of people rush to the entrances and exits in a short period of time, the system's uniform and indiscriminate identity verification process quickly becomes a bottleneck, leading to large backlogs and long queues.

[0003] This congestion not only severely impacts traffic efficiency but also creates chaos in security management. On-site personnel are forced to make difficult trade-offs between ensuring rigorous verification and managing the flow of people. Simplifying procedures to facilitate flow could lead to security vulnerabilities such as tailgating and identity theft. The fundamental technical problem lies in the lack of a closed-loop management mechanism in current technology that can perceive the dynamic changes in crowd congestion in real time and automatically and intelligently adjust verification strategies accordingly.

[0004] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this application discloses a method and system for optimizing campus personnel access processes. It aims to solve the technical problem that existing campus access management systems suffer from both low access efficiency and security risks due to the use of fixed verification processes when dealing with sudden peak traffic flows.

[0006] Firstly, this application discloses a method for optimizing campus personnel access management, the steps of which include: Based on the historical access behavior records of campus personnel, determine the credit evaluation data of campus personnel; The system acquires the congestion status parameters of the passageway in real time and determines whether the congestion status parameters meet the preset trigger conditions. If the triggering conditions are met, a differentiated verification mode will be activated for the access channel; under the differentiated verification mode, the corresponding credit rating will be identified based on the current credit evaluation data of campus personnel. Based on the credit rating, a corresponding verification process intensity is matched. The identity of personnel passing through the campus is verified according to the matched verification process intensity, and the release status of the access facilities is controlled based on the verification results. Continuously monitor congestion status parameters, and turn off the differentiated verification mode when the congestion status parameters meet the preset recovery conditions.

[0007] This technical solution establishes a closed-loop management mechanism that can dynamically adjust security strategies based on real-time congestion conditions. When the system detects peak pedestrian flow, it can automatically switch to a differentiated verification mode, employing verification processes of varying intensities based on individual credit ratings. This significantly improves traffic efficiency during peak hours and effectively alleviates congestion while ensuring safety.

[0008] Furthermore, the steps for determining the credit rating data of campus personnel based on their historical access behavior records include: Obtain historical access records of campus personnel within a preset time window; Identify violations in historical access records, including tailgating, impersonation, and access outside of permitted hours; Credit scores are calculated based on the frequency of violations, and personnel credit evaluation data is updated based on these scores.

[0009] This technical solution provides clear and quantifiable data sources and calculation basis for credit evaluation. By identifying and statistically analyzing specific violations such as tailgating and impersonation, it makes credit rating assessment more objective and fair, laying a solid foundation for the accurate implementation of subsequent differentiated verification.

[0010] Furthermore, the steps for obtaining real-time congestion status parameters of the passageway include: Real-time acquisition of the time interval between consecutive campus personnel entering the monitoring area; The rate of change in the influx of people into the campus is calculated based on time intervals and used as a congestion status parameter. The steps to determine whether the congestion status parameters meet the preset trigger conditions include: Determine if the congestion status parameter is greater than the preset acceleration threshold. If it is, then the trigger condition is met.

[0011] This technical solution uses the rate of change in traffic flow, or the "acceleration" of pedestrian flow, as the basis for judging congestion. Compared with monitoring only static indicators such as queue length, it can predict the formation of congestion trends earlier and more sensitively. This allows the system to intervene in advance and start a differentiated verification mode before the congestion worsens, thus improving the predictability and initiative of management.

[0012] Furthermore, the strength of the verification process includes the number of categories of verification factors; The steps involved in matching the strength of the verification process to the credit rating include: When the credit rating is high, the first verification strength is matched, and the verification factor corresponding to the first verification strength is any one of biometric features or identity medium features. When the credit rating is low, a second verification strength is matched. The verification factors corresponding to the second verification strength include biometric features and identity medium features.

[0013] This technical solution concretizes the abstract verification intensity into a combination of verification factors, providing high-credit personnel with a single-factor fast verification channel (such as facial recognition or card swiping only), and setting up a two-factor combination verification (such as facial recognition plus card swiping) for low-credit personnel. This achieves substantial differentiation in the verification process, directly converting credit rating into differences in passage time costs, and effectively incentivizing personnel to regulate their passage behavior.

[0014] Furthermore, the steps for verifying the identity of personnel passing through the campus according to the matched verification process intensity, and controlling the release status of access facilities based on the verification results, include: Obtain verification data provided by personnel waiting to pass through the campus; The verification data is compared using a verification algorithm that matches the intensity of the verification process to obtain the verification results; When the verification result is successful, a release instruction is sent to the passage facility to switch the passage facility to the open state. If the verification result is that the verification fails, keep the passage facility closed and generate a manual verification request.

[0015] This technical solution clarifies the complete execution logic from data collection and comparison to physical access control, ensuring that the differentiated verification strategy can be accurately translated into the specific release action of the gate, and sets up a clear handling process (generating a manual verification request) for verification failures, thus ensuring the integrity of the entire management process and the closed loop of operation.

[0016] Furthermore, after comparing the verification data using a verification algorithm that matches the strength of the verification process to obtain the verification result, the process also includes: When the verification result is not passed, the multi-dimensional auxiliary features of the personnel waiting to pass through the campus are obtained in real time within the preset monitoring area before entering the passage. The multi-dimensional auxiliary features include at least one of the following: physical characteristics, clothing characteristics, and characteristics of the items carried. The multidimensional auxiliary features are compared with the auxiliary reference features associated with the personnel to be admitted to the campus in the registration database; If the comparison result is greater than the preset compensation verification threshold, the verification result is corrected to pass verification; if the comparison result is less than the preset compensation verification threshold, the verification result is maintained as fail verification. Based on the final verification results, a credit compensation record is generated. According to the trigger frequency of the credit compensation record, the credit evaluation data of the personnel to be allowed to pass through the campus is dynamically downgraded.

[0017] This technical solution introduces a compensatory verification mechanism for initial verification failures. It utilizes auxiliary features such as body shape and clothing for secondary comparison, effectively reducing the false rejection rate caused by non-subjective factors such as fingerprint wear and facial occlusion, thus improving the system's fault tolerance and user experience. Simultaneously, by downgrading the credit score of users who frequently trigger the compensation mechanism, it prevents the mechanism from being abused, achieving a balance between humanization and risk control.

[0018] Furthermore, the method also includes: The physical envelope information of campus personnel entering the access passage is acquired in real time. The physical envelope information is used to characterize the volume boundary of campus personnel in physical space. Logically bind the verification results to the corresponding physical envelope information; Real-time monitoring of the edge spacing between two adjacent physical envelopes.

[0019] This technical solution binds the abstract identity verification result to the physical existence (physical envelope) of a person in the physical space, enabling precise tracking and spatial relationship analysis of individuals passing through. It provides accurate physical measurement means to prevent tailgating and other close following behaviors, extending security management from simple identity authentication to the monitoring of physical behavior.

[0020] Furthermore, the steps for controlling the release status of access facilities based on the verification results include: If the edge spacing is less than the preset critical spacing value, the passage facility will be controlled to perform a blocking action to increase the distance between two adjacent physical envelope information. Once the previous physical envelope information has completely left the sensing area of ​​the access facility, the verification permission for the next physical envelope information is released.

[0021] This technical solution provides a proactive physical anti-tailgating strategy. When the system detects that people are too close together, it can proactively intervene by blocking the turnstiles, forcibly creating a safe distance, thus physically eliminating the possibility of tailgating and greatly enhancing the security of the passage.

[0022] Furthermore, the steps for obtaining real-time physical envelope information of campus personnel entering the access passage include: Obtain the motion vector field distribution information within the physical envelope information; Identify independent motion units in the physical envelope information based on the motion vector field distribution information; If the difference in motion vectors between independent motion units in two adjacent physical envelope information is less than a preset consistency deviation threshold, it is determined that there is a load coupling relationship between the two adjacent physical envelope information. The steps for controlling the release status of passage facilities based on the verification results include: When a coupling relationship with the load is determined, the preset critical distance value is dynamically reduced and compensated according to the geometric dimensions of the coupling relationship with the load, and the access facility is kept open until the information of two adjacent physical envelopes completely leaves the sensing area of ​​the access facility.

[0023] This technical solution further enhances the intelligence level of the physical anti-tailgating system. By analyzing motion vectors, it distinguishes between people following each other and people accompanying luggage. It can accurately identify normal passage scenarios such as carrying large suitcases, avoiding passage interruptions caused by luggage being mistakenly identified as tailgating personnel. While ensuring safety, it optimizes the passage experience in special situations.

[0024] Secondly, this application also discloses a campus personnel access process optimization management system, including: The data acquisition module determines the credit evaluation data of campus personnel based on their historical access behavior records. The parameter monitoring module acquires the congestion status parameters of the passage in real time and determines whether the congestion status parameters meet the preset trigger conditions. If the triggering conditions are met, the mode activation module will activate the differentiated verification mode for the access channel; under the differentiated verification mode, the corresponding credit rating will be identified based on the current credit evaluation data of campus personnel. The identity verification module matches the corresponding verification process intensity based on the credit rating, verifies the identity of personnel to be admitted to the campus according to the matched verification process intensity, and controls the release status of the access facilities based on the verification results. The status recovery module continuously monitors congestion status parameters. When the congestion status parameters do not meet the preset recovery conditions, the differentiated verification mode is turned off.

[0025] In summary, this application provides a method and system for optimizing campus access control. This method fundamentally solves this problem by introducing a dynamic, closed-loop intelligent management mechanism. First, it can monitor the acceleration of pedestrian flow in passageways in real time, providing early warnings of congestion trends. Second, upon detecting congestion risk, it automatically activates a differentiated verification mode, the core of which is a credit evaluation system based on historical behavior. For students and faculty with good credit records, the system uses a single-factor, high-speed verification process for rapid passage, quickly absorbing the majority of peak pedestrian flow. For individuals with lower credit ratings or violations, a multi-factor, high-intensity verification process is used to ensure the rigor of security verification. Ultimately, without compromising overall security standards, it significantly improves traffic efficiency during peak hours, effectively alleviates congestion, improves the user experience, and achieves simultaneous improvements in both efficiency and security at campus entrances and exits. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating a campus personnel access management optimization method provided in an embodiment of this application.

[0027] Figure 2 This is a schematic diagram of the structure of a campus personnel access optimization management system provided in an embodiment of this application.

[0028] Labeling Explanation: 210, Data Acquisition Module; 220, Parameter Monitoring Module; 230, Mode Activation Module; 240, Identity Verification Module; 250, Status Recovery Module. Detailed Implementation

[0029] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0030] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0031] In a specific application scenario, university libraries, as areas with high concentrations of people on campus, face significant challenges in managing their entrances and exits. Especially during the morning opening hours of final exam weeks, a large number of students flood in within a short period, hoping to secure seats. Traditional access control systems typically employ uniform verification standards for all personnel, such as requiring everyone to swipe their campus card and undergo facial recognition verification. This fixed, indiscriminate process ensures security when the flow of people is stable, but during peak hours, each student needs several seconds to complete the entire process. This small amount of time accumulates quickly, forming long queues at the entrance and causing severe congestion. The anxious waiting not only affects students' study mood, but the crowded conditions also provide opportunities for tailgating, identity theft, and other violations. Security personnel are caught in a dilemma between maintaining order and strictly verifying credentials, resulting in a decline in both efficiency and security levels.

[0032] In this regard, firstly, referring to Figure 1 This application provides a method for optimizing the management of campus personnel access procedures, the steps of which include: S1. Determine the credit evaluation data of campus personnel based on their historical access behavior records; S2. Obtain the congestion status parameters of the passage in real time and determine whether the congestion status parameters meet the preset trigger conditions. S3. If the triggering conditions are met, a differentiated verification mode will be activated for the access channel. Under the differentiated verification mode, the corresponding credit rating of the current campus personnel will be identified based on their credit evaluation data. S4. Match the corresponding verification process intensity according to the credit rating, verify the identity of the personnel to be passed through the campus according to the matched verification process intensity, and control the release status of the access facilities according to the verification results; S5. Continuously monitor congestion status parameters. When the congestion status parameters meet the preset recovery conditions, turn off the differentiated verification mode.

[0033] Credit rating data refers to a data indicator used to quantify and characterize the compliance of historical access behavior of campus personnel. It is not a simple good or bad label, but a dynamically updated numerical value or level stored in the central database of campus management and linked to each person's unique identifier (such as student ID or faculty / staff ID). Congestion status parameters are dynamic indicators that reflect the degree and growth trend of personnel gathering at the entrance of access channels in real time, aiming to capture early signals of congestion. Differentiated verification mode is a special working state that differs from the conventional uniform verification mode. In this mode, the access control system no longer performs the same verification process for everyone, but intelligently and differentially assigns different verification tasks based on individual credit ratings. Verification process intensity measures the complexity of the verification task; a high-intensity process means more information needs to be verified and it takes longer, while a low-intensity process does the opposite.

[0034] In one specific embodiment, the execution process of the aforementioned campus personnel access control optimization method can be deployed in the entrance gate control system of a university library. This system consists of a management server deployed in the cloud, an edge computing gateway installed at the entrance, a high-definition camera, a multimodal card reader, an infrared beam sensor, and the gate itself.

[0035] The entire process begins with establishing credit evaluation data for each campus staff member. This process is automatically completed periodically by the management server in the background. The server accesses historical access logs stored in the database. These logs record in detail every access event over a period of time (e.g., the last six months), including the access person's ID, access time, access channel, verification method, and abnormal event markers captured by sensors. The server program filters and identifies violations from this massive amount of log data based on a preset rule base. For example, if the log shows that an ID swipes its card to open a door, but then the infrared sensor in the channel detects two physical entities passing through, the system will determine this as a tailgating event and associate it with the previously successfully verified ID. Similarly, if the log shows that an ID swipes its card, but the facial image captured by the camera is compared with a template photo of that ID stored in the database, and the similarity is far below a preset threshold (e.g., below 90%), the system will determine this as a suspected impersonation event. All these identified violations are accumulated and recorded in the corresponding personnel's credit file. Based on these records, the system updates each person's credit score and assigns different credit levels according to the score range. For example, the initial score is 100 points. Ten points are deducted for each instance of tailgating, twenty points are deducted for each instance of identity theft, and five points are added for a month without any violations. A final score of 90 or above is considered a high credit level, below 70 is a low credit level, and the rest are considered medium credit levels. This credit evaluation data forms the basis for all subsequent differentiated strategies.

[0036] In daily operation, the library entrance access control system operates in a standard verification mode, requiring everyone to swipe their card and undergo facial recognition. Simultaneously, the system continuously acquires real-time congestion status parameters of the passageway via a wide-angle camera mounted on the ceiling five meters away from the turnstiles. This process is handled by an edge computing gateway to ensure low latency. The camera captures individuals entering its field of view and uses object detection algorithms to outline each person's silhouette. The system records the timestamp when the center point of each silhouette first crosses a virtual detection line on the ground. By calculating the difference between the timestamps of two consecutive individuals, the system obtains a series of time interval data. If the population is sparse, this time interval may be five to ten seconds; if the flow of people becomes dense, the time interval will shorten to two to three seconds. The system does not simply use this time interval but further calculates the trend of these time interval changes, i.e., the acceleration of the influx of people. Specifically, the system maintains a queue containing the ten most recent time interval values ​​and calculates the rate of decrease of the values ​​in this queue. When a large number of people rush in within a short period of time due to some reason (such as rain causing students to arrive in large numbers), the time interval will rapidly decrease from three seconds to one second or even lower, and the rate of change will increase dramatically. The edge computing gateway uses this calculated rate of change in traffic as a core congestion status parameter and compares it with a preset acceleration threshold. This threshold is set based on the analysis of historical data of the entry point and represents the critical point at which congestion is about to occur. Once the detected rate of change in traffic exceeds this threshold, for example, if the average interval change rate over three consecutive seconds exceeds 50%, the system determines that the preset trigger condition has been met.

[0037] Once the triggering conditions are met, the edge computing gateway immediately sends a mode switching request to the management server and simultaneously enables differentiated verification mode for the access channel locally. At this time, the prompt on the display screen above the entrance changes from "Please swipe your card and perform facial recognition" to "During peak hours, the system has activated fast-pass mode." When a student waiting to pass through approaches the gate, the camera first performs a quick facial capture. Based on the captured facial features, the edge computing gateway retrieves the student's identity ID from its local cache database. Subsequently, the system uses this ID to query the corresponding credit rating data and identify the student's credit level.

[0038] Next, the system will match an appropriate verification process strength for the student based on the identified credit rating. If the student's credit rating is high, the system will match them with the first verification strength. This strength means that only one verification factor needs to be verified. For example, the system can directly use the captured facial recognition result; as long as the result matches the database template, the verification is considered complete, and the student does not need to swipe their card again. Alternatively, in another configuration, the system may prompt the student to simply swipe their campus card; once the card reader reads the valid card information, passage is immediately granted, without requiring a mandatory facial comparison. This single-factor verification method greatly shortens the verification time for a single person. Conversely, if another student's credit rating is low, the system will match them with the second verification strength. This strength means that a combination of multiple verification factors must be completed. At this time, the screen in front of the gate will clearly prompt the student to "swipe their card and face the camera to complete facial recognition." The system must simultaneously receive a valid card swipe signal and a high-match facial recognition result; only when both conditions are met is the verification considered successful.

[0039] The system verifies the identity of individuals according to the matched verification process strength. For students with high credit ratings, the system only needs to complete facial recognition. Once the verification is successful, the edge computing gateway immediately sends a high-level pulse signal to the control interface of the gate's mainboard. This signal drives the motor inside the gate, opening the barrier and allowing passage. The entire process may take less than a second. For students with low credit ratings, the system waits for both card swiping and facial recognition to be completed and successfully matched before sending the passage command. If either step fails, such as a facial match failure after card swiping, the system determines that the verification has failed. In this case, the gate will remain closed, and the system will generate a manual verification request. This request will appear as a pop-up on the computer screen of a nearby security personnel's workstation, indicating that a student's verification has failed, and attaching a snapshot of the student and card information for manual intervention by security personnel.

[0040] Throughout the operation of the differentiated verification mode, the system does not stop monitoring congestion parameters. The edge computing gateway continuously calculates the inflow rate of people. As the peak period passes and the flow of people gradually thins out, the time interval between consecutive passersby will lengthen again, and the flow change rate will drop to a lower level. When this parameter remains below a preset recovery condition for a period of time (e.g., for one minute), which is usually a threshold lower than the trigger condition to avoid frequent switching of the system in a critical state, the system will determine that the congestion has been relieved. At this time, the edge computing gateway will automatically shut down the differentiated verification mode, restoring the system to the regular verification mode that uses a uniform standard procedure for everyone. The prompts on the entrance screen will also be restored accordingly.

[0041] Through the above process, the technical solution of this application can intelligently provide a fast track for the vast majority of compliant and creditworthy teachers and students during peak hours, concentrating valuable verification resources and time on more rigorous screening of a small number of potentially risky individuals. This not only greatly improves overall traffic efficiency and effectively alleviates congestion, but also makes security management more targeted and efficient, achieving a dynamic balance between ensuring safety and improving the user experience.

[0042] Furthermore, to make the credit rating system more fair and accurate, the steps for determining the credit rating data of campus personnel can be refined as follows: Obtain historical access records of campus personnel within a preset time window; Identify violations in historical access records, including tailgating, impersonation, and access outside of permitted hours; Credit scores are calculated based on the frequency of violations, and personnel credit evaluation data is updated based on these scores.

[0043] In its implementation, the data analysis module running on the management server is configured with a task schedule, such as automatically retrieving and analyzing all passage logs from the past 180 days at 2:00 AM every day. This module has built-in algorithms for identifying different types of violations. For tailgating, the system analyzes sensor data within the gate channel. For example, a 3D depth camera installed at the top of the channel can construct a 3D point cloud model of the person passing through in real time. After a legitimate verification and opening command is issued, the system tracks the number of independent physical entities passing through the channel. If, within an opening cycle, the system identifies two or more spatially separated human silhouettes with slightly different trajectories passing through sequentially, it records this as a tailgating event and binds the event to the identity ID that triggered the opening. For impersonation, the system compares the captured facial image with the ID photo or registration photo pre-stored in the database associated with the card swipe or QR code scan during each verification. This comparison process can use a deep learning-based feature vector comparison algorithm to calculate a similarity score. If a score is below a strict threshold, such as 95%, but above a threshold that might be mistaken for a completely unrelated person, such as 50%, the system will flag it as a suspected impersonation incident. For access outside of permitted hours, the system associates one or more permission policy groups with each identity ID. These policy groups define the areas and time periods allowed for access. For example, an undergraduate student's access might be limited to entering the library from 8:00 AM to 10:00 PM on weekdays. When an undergraduate student attempts to swipe their card to enter in the early morning, the system will reject the request and automatically record an unauthorized access attempt outside of permitted hours.

[0044] After identifying these violations, the system updates the credit score according to a quantified scoring rule. For example, the system sets an initial credit score (e.g., 100 points) for each person passing through. Fifteen points are deducted for each confirmed tailgating incident; thirty points are deducted for each instance of identity theft; and five points are deducted for each attempt to pass through outside of permitted hours. To encourage good behavior, the system can also set bonus rules; for example, if a person has no violations for ninety consecutive days, the system will automatically add ten points, but the total score cannot exceed one hundred points. Through this dynamic, reward-and-penalty scoring mechanism, the credit score can more accurately reflect a person's recent passing behavior, thus providing a more reliable basis for differentiated verification.

[0045] Furthermore, in order to detect congestion earlier and more accurately, the step of acquiring congestion status parameters of traffic lanes in real time can be refined as follows: Real-time acquisition of the time interval between consecutive campus personnel entering the monitoring area; The rate of change in the influx of people into the campus is calculated based on time intervals and used as a congestion status parameter. The steps to determine whether the congestion status parameters meet the preset trigger conditions include: Determine if the congestion status parameter is greater than the preset acceleration threshold. If it is, then the trigger condition is met.

[0046] In its implementation, this process heavily relies on front-end sensing devices and edge computing capabilities. A monitoring zone approximately one meter wide is marked on the ground five to eight meters from the turnstile. When someone enters this zone, a camera mounted directly above uses its built-in video analytics, or is processed by a connected edge computing box, to identify key points on the person and track their movement. The system precisely records the moment each person's centroid crosses the boundary of the monitoring zone's entrance. By continuously recording these moments, the system obtains a time series, T1, T2, T3, ..., Tn. The differences between any two adjacent moments, Δt1 = T2 - T1, Δt2 = T3 - T2, constitute the time interval sequence of consecutive people entering the monitoring zone.

[0047] The core improvement lies in the fact that the system does not directly use the time interval Δt as the basis for judging congestion, because a single Δt value fluctuates greatly and is prone to misjudgment. The system focuses on the changing trend of this interval, that is, the acceleration of the pedestrian flow. The edge computing gateway maintains a sliding window of size N (e.g., N=5) to store the latest five time interval values. When each new time interval Δt_new is generated, the system calculates the average value Avg_t_current of the time intervals within the current window and compares it with the average value Avg_t_previous calculated at the previous time point. The rate of change in traffic flow can be defined as (Avg_t_previous - Avg_t_current) / Avg_t_previous. When pedestrian flow surges in, Δt continuously decreases, causing Avg_t_current to be less than Avg_t_previous, and this rate of change will be a positive value. If the pedestrian flow speed increases, this positive value will increase. The system compares this calculated rate of change in traffic flow with a preset acceleration threshold. For example, if the threshold is set at 30%, it means that when the average passage interval decreases by more than 30% in a short period of time, the system considers that the flow of people is converging at an unsustainable rate, and congestion is about to occur. This judgment method based on the rate of change has higher sensitivity and predictability than reacting only when the waiting queue length exceeds a certain fixed value. It allows the system to intervene in advance and initiate differentiated management before congestion actually forms and causes serious impact.

[0048] Furthermore, to ensure that the differentiation in verification process strength has a clear and actionable implementation path, it can be refined as follows: The strength of the verification process includes the number of categories of verification factors; The steps involved in matching the strength of the verification process to the credit rating include: When the credit rating is high, the first verification strength is matched, and the verification factor corresponding to the first verification strength is any one of biometric features or identity medium features. When the credit rating is low, a second verification strength is matched. The verification factors corresponding to the second verification strength include biometric features and identity medium features.

[0049] In practical implementation, system administrators can combine different verification factors in the backend configuration interface to define different verification strength levels. Verification factors are mainly divided into two categories. The first category is biometric features, which include facial images captured by cameras, fingerprint information obtained by fingerprint scanners, and iris patterns obtained by iris scanners. These are uniquely linked to each individual and are difficult to forge. The second category is identity medium features, which include encrypted ID numbers stored in the campus card chip and identity information contained in dynamic QR codes generated by mobile apps. These require a physical medium.

[0050] When the system identifies a person's credit rating as high, it automatically assigns them the highest level of verification strength. The core of this strength is single-factor authentication, meaning only one of the two main categories of verification factors needs to pass. Several parallel implementation methods are possible. The first is face-first, where the gate prioritizes face recognition by default. If successful, passage is granted immediately, even if the student has already taken out their campus card to swipe; the system doesn't need to wait for the swipe. The second is medium-first, where the system waits for the student to swipe their card or scan a QR code. Once valid information is read, passage is granted immediately, with face recognition serving as an auxiliary record and its comparison result not affecting the passage. The third method involves simultaneously activating face recognition and the card reader. The two verification channels operate in parallel; either channel completes and passes verification first, immediately triggering the passage command. This single-factor authentication design greatly simplifies the passage process for high-credit individuals and is key to rapid flow control during peak periods.

[0051] Conversely, when the system identifies a person as having a low credit rating, it automatically assigns a second level of verification. The core of this level is multi-factor authentication, which requires joint verification of both biometric features and identity medium characteristics. Specifically, the system requires students to first swipe their card or scan a QR code. After reading the identity medium information, the system uses this as an index to retrieve the person's biometric feature template. Then, the student must complete facial recognition or fingerprinting at a designated location. The system compares the biometric features collected on-site with the retrieved template, while simultaneously verifying the legitimacy of the identity medium. Only when both verification factors pass will the system issue a release order. This double-insurance verification process, although time-consuming, significantly improves verification security and ensures strict control over high-risk individuals.

[0052] Furthermore, to ensure that the entire verification and control process forms a complete closed loop, it can be further refined as follows: The steps for verifying the identity of personnel passing through the campus according to the matched verification process intensity, and controlling the release status of access facilities based on the verification results, include: Obtain verification data provided by personnel waiting to pass through the campus; The verification data is compared using a verification algorithm that matches the intensity of the verification process to obtain the verification results; When the verification result is successful, a release instruction is sent to the passage facility to switch the passage facility to the open state. If the verification result is that the verification fails, keep the passage facility closed and generate a manual verification request.

[0053] In the actual implementation, this process is led by the edge computing gateway. When a student arrives at the gate, the verification data they provide may be in various forms. The edge computing gateway determines which data needs to be processed and how to process it based on the strength of the verification process currently matched for that student.

[0054] If the matching is for the first verification strength (single factor), the gateway may only process video stream data. It extracts the best frame of the face image from the video stream, uses a loaded lightweight face recognition model (such as MobileFaceNet) to extract a face feature vector, and then calculates the cosine similarity between this vector and the student's registration feature vector, either locally cached or quickly retrieved from the cloud. If the calculated similarity score is higher than a preset passing threshold (e.g., 0.98), the verification result is considered passed.

[0055] If the matching is for the second verification strength (two-factor), the gateway needs to process both the video stream and the data from the card reader. It first reads the card serial number and uses it as an index to confirm the identity of the person being verified. Then, it performs the same face recognition and comparison process as described above. Finally, it performs a logical check: whether the card serial number is valid and whether the face similarity score is higher than a threshold. Only when both conditions are met is the verification result considered successful.

[0056] When the verification result is successful, the edge computing gateway's processor sends a control signal to the connected gate driver board via its general purpose input / output (GPIO) interface. For example, it pulls a pin's level high and maintains it for 200 milliseconds. Upon receiving this signal, the driver board controls the connected motor or electromagnet to unlock and open the gate's wing door or swing arm, switching the access facility to the open state.

[0057] Conversely, when the verification result is a failure, such as a failed face comparison or an invalid card, the edge computing gateway will not send any access command. Therefore, the gate driver board will not receive a signal, and the gate will naturally remain closed. Simultaneously, the gateway will encapsulate an event packet containing detailed information about the failure, such as the student's ID (if read), the captured photo, the reason for the failure (e.g., low facial similarity or unregistered card), and the time of occurrence. This event packet is then sent to the security monitoring center's host via the local area network using the TCP protocol. Upon receiving the event packet, the application on the monitoring host will parse its content and display a prominent alarm window on the monitoring interface. This is the generated manual verification request, prompting security personnel to immediately address this anomaly.

[0058] Building upon the above implementation methods, to further enhance the system's fault tolerance and user experience, especially when handling verification failures caused by non-subjective factors, a compensatory verification mechanism can be introduced. In practical applications, even legitimate users may experience initial biometric recognition failures due to wearing masks or hats, fingerprint wear, or wet fingers. Simply attributing all failures to verification failures and triggering manual intervention would not only reduce efficiency but also inconvenience users.

[0059] In response, after comparing the verification data using a verification algorithm that matches the strength of the verification process to obtain the verification result, the method further includes: When the verification result is not passed, the multi-dimensional auxiliary features of the personnel waiting to pass through the campus are obtained in real time within the preset monitoring area before entering the passage. The multi-dimensional auxiliary features include at least one of the following: physical characteristics, clothing characteristics, and characteristics of the items carried. The multidimensional auxiliary features are compared with the auxiliary reference features associated with the personnel to be admitted to the campus in the registration database; If the comparison result is greater than the preset compensation verification threshold, the verification result is corrected to pass verification; if the comparison result is less than the preset compensation verification threshold, the verification result is maintained as fail verification. Based on the final verification results, a credit compensation record is generated. According to the trigger frequency of the credit compensation record, the credit evaluation data of the personnel to be allowed to pass through the campus is dynamically downgraded.

[0060] In one specific implementation, when a student fails facial recognition at the turnstile, the system does not immediately reject the application. Assuming the student has already swiped their campus card, the system already knows their identity ID. At this point, the edge computing gateway immediately initiates a compensatory verification process. It retrieves video footage recorded over the past five seconds by a wide-angle camera mounted on the ceiling of the entrance passage. This video records the student's entire process from entering the monitoring area to walking towards the turnstile. The system uses a pre-trained multi-task deep learning model (e.g., a combination of YOLO and DeepSORT) to process this video and extract the student's multi-dimensional auxiliary features.

[0061] Postural features can be derived from the human skeletal structure extracted using skeletal keypoint detection algorithms, and parameters such as gait cycle, stride length, and shoulder width to height ratio can be calculated. Clothing features can be derived from information such as the color of the top (e.g., a red T-shirt) and the type of pants (e.g., blue jeans) extracted through image segmentation and color analysis. Carried object features are identified by object detection algorithms, specifically the items carried (e.g., carrying a black backpack).

[0062] Based on the aforementioned multi-dimensional auxiliary features, this multi-task deep learning model can employ a shared backbone network (such as ResNet or EfficientNet) and branch out into multiple task-specific head networks. For example, one head might be used for object detection (YOLO), another for pose estimation, and yet another for clothing attribute classification and object identification. The Re-ID part of DeepSORT can be a separate CNN network used to extract discriminative features of people. The backbone network is typically pre-trained on large image classification or object detection datasets. The head networks for each task are trained on their respective labeled datasets. Finally, the entire multi-task model is fine-tuned end-to-end on a comprehensive campus scene dataset containing all labeled information.

[0063] Since it is a multi-task model, the total loss function of the model is a weighted sum of the loss functions of each task (L_total=λ1×L_pose+λ2×L_cloth+λ3×L_object, where λ1, λ2, and λ3 are preset weighting coefficients, L_pose is the pose estimation loss, L_cloth is the clothing attribute classification loss, and L_object is the object detection loss).

[0064] Object detection loss (L_object) can be obtained using classification and regression losses; pose estimation loss (L_pose) can use keypoint regression loss (such as L1 loss); clothing attribute classification loss (L_cloth) can use cross-entropy loss. Re-ID models use triplet loss or cross-entropy loss. Furthermore, this model can employ Adam or SGD optimizers and perform learning rate scheduling. Data augmentation is particularly important for multi-task models, requiring the design of corresponding augmentation strategies for different tasks. Multi-task learning typically requires careful balancing of the loss weights for different tasks to avoid the optimization of one task dominating the entire training process, in order to extract more accurate auxiliary features.

[0065] These on-site extracted auxiliary features are compared with the auxiliary reference features associated with the student in the registration database. These auxiliary reference features are not registered once, but rather automatically and seamlessly learned and updated by the system as the user successfully passes through multiple times in the past. For example, the system stores the student's average gait feature vector from their ten most recent successful passes, color histograms of their most frequently worn clothing colors, etc. The comparison process can employ a Siamese Network, trained to determine whether two sets of input features (one set extracted on-site, the other set of reference features) belong to the same person. The network outputs a similarity score between 0 and 1.

[0066] If the score is greater than a preset compensation verification threshold (e.g., 0.85), it means that although facial recognition failed, the person's posture, clothing, and other characteristics highly match the historical records, and the system has a high degree of confidence that this is the person. In this case, the system will correct the verification result to pass and send a release command.

[0067] Meanwhile, to prevent this compensation mechanism from being abused or becoming a potential security vulnerability, the system generates a credit compensation record every time compensation verification is successfully triggered. The management server periodically tracks the trigger frequency of each person's credit compensation records. If a student triggers compensation verification more than three times within a month, the system considers their biometrics to have low validity or stability, or suspects deliberate avoidance of positive identification. As a risk management measure, the system dynamically downgrades their credit score. For example, in the next credit score calculation, the penalty for their violation is multiplied by a coefficient of 1.2, or the natural growth rate of their credit score is slowed down. This design provides convenience while retaining sensitivity and constraints against abnormal patterns.

[0068] In addition, in order to address the limitations of traditional infrared beam sensors in anti-tailgating functions (such as the inability to distinguish between people and luggage and the ease with which they can be deceived), this application introduces a more refined physical space perception technology.

[0069] The method also includes: The physical envelope information of campus personnel entering the access passage is acquired in real time. The physical envelope information is used to characterize the volume boundary of campus personnel in physical space. Logically bind the verification results to the corresponding physical envelope information; Real-time monitoring of the edge spacing between two adjacent physical envelopes.

[0070] In a practical implementation, a 3D Time-of-Flight (ToF) camera or a binocular stereo vision camera is installed directly above the gate. These cameras can output depth images or point cloud data in real time. The edge computing gateway continuously processes this data, using point cloud segmentation algorithms (such as Euclidean distance-based clustering) to divide the point cloud within the gate into one or more independent clusters. Each cluster represents an independent entity occupying a certain volume in physical space, i.e., a physical envelope. This information can be represented as a 3D bounding box, containing the entity's position, size, and orientation.

[0071] When a student completes identity verification at the turnstile and receives a pass, the system immediately logically binds this verification success event to the physical envelope currently within the turnstile's verification area. This binding is based on spatiotemporal correlation; that is, at the point of successful verification, the physical envelope closest to the verification device is considered the legitimate user.

[0072] Once the binding is complete, the system continuously tracks the movement trajectory of the authorized physical envelope within the channel. Simultaneously, the system monitors for the appearance of new, unauthorized physical envelopes within the channel. If a second physical envelope appears (e.g., a tailer), the system calculates the edge spacing between these two adjacent physical envelopes in real time—the distance between the two closest faces of their respective 3D bounding boxes in the forward direction. This precise measurement of the spacing provides crucial input data for subsequent proactive anti-tailgating strategies.

[0073] Based on precise monitoring of the physical envelope spacing, the system can implement proactive physical interventions to effectively prevent tailing.

[0074] The steps for controlling the release status of passage facilities based on the verification results include: If the edge spacing is less than the preset critical spacing value, the passage facility will be controlled to perform a blocking action to increase the distance between two adjacent physical envelope information. Once the previous physical envelope information has completely left the sensing area of ​​the access facility, the verification permission for the next physical envelope information is released.

[0075] In a specific scenario, student A successfully swipes their card, the turnstile wing opens, and their corresponding physical envelope begins to pass through the passage. At this moment, student B closely follows behind student A, attempting to rush in before the wing closes. The system detects student B's second physical envelope using a 3D camera and calculates that the edge distance between student B's and student A's physical envelope is only 20 centimeters. This value is less than the system's preset critical distance value (e.g., 50 centimeters).

[0076] Once the edge computing gateway determines that the distance is too small, it immediately sends a blocking action command to the turnstile driver board. This action is not a simple closing of the door, but a more intelligent intervention. For example, the wing door, which was originally fully open, will quickly close partially, leaving the passage width at only about 30 centimeters. This width is insufficient for an adult to pass through easily, but it will not completely trap the person in front. This sudden narrowing of the passage will force student B behind to subconsciously slow down or stop, thus physically increasing the distance between him and student A. At the same time, the turnstile may emit a rapid beeping sound and flashing red light as a warning.

[0077] While executing the blocking action, the system strictly adheres to the principle of one person, one authorization. Even if student B tries to swipe their campus card at the gate, the card reader will not react, and the facial recognition camera will not activate. This is because the system will lock the verification permission until it detects that student A's physical envelope has completely left the end-sensing area of ​​the gate channel, preventing it from being released to the next person. Only after the sensor confirms that the previous person has completely passed through will the gate return to standby mode and release the verification permission, allowing the next person (i.e., the blocked student B) to perform normal identity verification. This design fundamentally eliminates the security vulnerability of multiple people trying to pass through.

[0078] However, the aforementioned anti-tailgating strategy based on physical envelope spacing may misjudge a common scenario: when a person is carrying large luggage (such as a suitcase), the system may mistakenly identify the luggage as a tailgating person, thus incorrectly triggering a blocking action. To solve this problem, a deeper analysis of the inherent properties of the physical envelope is needed.

[0079] The steps for obtaining real-time physical envelope information of campus personnel entering access channels include: Obtain the motion vector field distribution information within the physical envelope information; Identify independent motion units in the physical envelope information based on the motion vector field distribution information; If the difference in motion vectors between independent motion units in two adjacent physical envelope information is less than a preset consistency deviation threshold, it is determined that there is a load coupling relationship between the two adjacent physical envelope information. The steps for controlling the release status of passage facilities based on the verification results include: When a coupling relationship with the load is determined, the preset critical distance value is dynamically reduced and compensated according to the geometric dimensions of the coupling relationship with the load, and the access facility is kept open until the information of two adjacent physical envelopes completely leaves the sensing area of ​​the access facility.

[0080] In its implementation, when the system acquires point cloud data within a channel using a 3D camera and segments it into two adjacent physical envelopes, it further analyzes the motion characteristics within each envelope. By comparing point cloud data from two consecutive frames (e.g., with a 30-millisecond interval), the system can use optical flow or point cloud registration algorithms to calculate the motion vectors of each point within each point cloud cluster, thus forming a motion vector field.

[0081] Within the physical envelope of a person, the motion vectors of their torso and limbs, while not entirely identical, exhibit an overall distribution of motion around the center of mass with inherent coordination. Conversely, within the physical envelope of a rigid suitcase, the motion vectors of all points are almost identical. By analyzing the distribution patterns of the vector field, the system can identify whether each physical envelope represents a single rigid or non-rigid motion unit.

[0082] When the system identifies two adjacent physical envelopes (one identified as a human figure and the other as a rigid object), it calculates the difference between the average motion vectors of these two envelopes. If a student is pulling a suitcase, the motion vectors of his body and the suitcase will be highly consistent in direction and velocity. Their vector difference will be very small, much smaller than a preset consistency deviation threshold. In this case, the system determines that there is a load coupling relationship between these two physical envelopes, that is, they are a combination of "person + luggage" rather than two independent people.

[0083] Once such a coupling relationship is detected, the system temporarily adjusts its anti-tailgating strategy. It measures the geometric dimensions of the physical envelope identified as luggage (e.g., its length in the direction of travel) and then dynamically reduces and compensates for a preset critical distance value (e.g., 50 cm), for example, by directly subtracting the length of the luggage. This means the system allows the combination to pass as a whole, even if the distance between them is zero. At this time, the system suppresses the triggering of blocking actions and keeps the gate wing open until it detects that both the person and the luggage's physical envelopes have completely left the sensing area of ​​the passage, at which point the wing closes. This intelligent recognition capability allows the system to ensure anti-tailgating security while also taking into account the normal passage needs of people carrying large items, significantly improving the system's intelligence level and user experience.

[0084] Secondly, in order to implement the aforementioned optimized management method for campus personnel access procedures, refer to Figure 2This application also proposes a campus personnel access control system, which includes: The data acquisition module 210 is used to determine the credit evaluation data of campus personnel based on their historical access behavior records. The parameter monitoring module 220 is used to acquire the congestion status parameters of the passage in real time and determine whether the congestion status parameters meet the preset trigger conditions. If the triggering conditions are met, the mode activation module 230 will activate the differentiated verification mode for the access channel; under the differentiated verification mode, the corresponding credit rating will be identified based on the current credit evaluation data of campus personnel. The identity verification module 240 is used to match the corresponding verification process intensity according to the credit level, verify the identity of the personnel to be passed through the campus according to the matched verification process intensity, and control the release status of the access facilities based on the verification results. The status recovery module 250 is used to continuously monitor congestion status parameters. When the congestion status parameters do not meet the preset recovery conditions, the differentiated verification mode is turned off.

[0085] The data acquisition module 210 can be a background service program deployed on a cloud management server. It is responsible for periodically accessing and analyzing the access logs stored in the database, and calculating and updating the credit score and level of each campus personnel according to the preset rules for identifying violations.

[0086] The parameter monitoring module 220 mainly consists of an edge computing gateway deployed at the entrance and exit site and sensors such as cameras connected to it. It is responsible for collecting and processing front-end data in real time, calculating the rate of change of pedestrian flow, and comparing it with preset thresholds to determine the congestion status.

[0087] The mode activation module 230 can also be implemented in the edge computing gateway or cloud server. After receiving the trigger signal from the parameter monitoring module 220, it is responsible for issuing instructions to switch the overall working status of the system from the normal mode to the differentiated verification mode.

[0088] The identity verification module 240 is the core execution unit of the entire system. It is integrated in the edge computing gateway and is responsible for dynamically selecting the verification process according to the instructions of the mode start module 230 and the credit data provided by the data acquisition module 210. It also interacts with hardware such as card readers, cameras, and gate driver boards to complete the entire process from data collection and comparison to control and release.

[0089] The state recovery module 250 works closely with the parameter monitoring module 220. It continuously monitors changes in congestion status parameters, and when the parameters fall back to a safe level, it triggers the system to exit the differentiated verification mode and resume normal management. These modules work together to form a complete closed-loop management system capable of automatic perception, decision-making, and execution.

[0090] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for optimizing campus personnel access management, characterized in that, The steps of this method include: Based on the historical access behavior records of campus personnel, determine the credit evaluation data of campus personnel; The congestion status parameters of the passageway are acquired in real time, and it is determined whether the congestion status parameters meet the preset triggering conditions. If the triggering condition is met, a differentiated verification mode is activated for the access channel; under the differentiated verification mode, the corresponding credit rating is identified based on the current campus personnel's credit evaluation data. Based on the credit rating, a corresponding verification process intensity is matched. The identity of the person to be admitted to the campus is verified according to the matched verification process intensity, and the release status of the access facilities is controlled according to the verification results. The congestion status parameters are continuously monitored, and the differentiated verification mode is turned off when the congestion status parameters meet the preset recovery conditions.

2. The method for optimizing campus personnel access management according to claim 1, characterized in that, The steps for determining the credit rating data of campus personnel based on their historical access behavior records include: Obtain historical access records of campus personnel within a preset time window; Identify violations in the historical access behavior records, including tailgating, impersonating another person, and accessing outside of permitted hours; A credit score is calculated based on the frequency of the violation, and the credit evaluation data of the person is updated based on the credit score.

3. The method for optimizing campus personnel access management according to claim 1, characterized in that, The steps for obtaining the congestion status parameters of the passageway in real time include: Real-time acquisition of the time interval between consecutive campus personnel entering the monitoring area; The rate of change in the influx of people into the campus is calculated based on the time interval and used as a congestion status parameter. The step of determining whether the congestion status parameter meets the preset triggering conditions includes: Determine whether the congestion status parameter is greater than a preset acceleration threshold. If it is, then the triggering condition is satisfied.

4. The method for optimizing campus personnel access management according to claim 1, characterized in that, The strength of the verification process includes the number of categories of verification factors. The step of matching the verification process strength corresponding to the credit rating includes: When the credit rating is a high credit rating, a first verification strength is matched, and the verification factor corresponding to the first verification strength is any one of biometric features or identity medium features. When the credit rating is low, a second verification strength is matched, and the verification factors corresponding to the second verification strength include biometric features and identity medium features.

5. The method for optimizing campus personnel access management according to claim 1, characterized in that, The steps of verifying the identity of personnel to be admitted to the campus according to the matched verification process intensity, and controlling the release status of access facilities based on the verification results, include: Obtain verification data provided by personnel waiting to pass through the campus; The verification data is compared using a verification algorithm that matches the intensity of the verification process to obtain the verification result; When the verification result is that the verification is passed, a release command is sent to the passage facility to switch the passage facility to the open state; If the verification result is that the verification fails, the access facility remains closed, and a manual verification request is generated.

6. The method for optimizing campus personnel access management according to claim 5, characterized in that, After the step of comparing the verification data using a verification algorithm that matches the strength of the verification process to obtain the verification result, the method further includes: When the verification result is not passed, the multi-dimensional auxiliary features of the person waiting to pass through the campus are acquired in real time within a preset monitoring area before entering the passage. The multi-dimensional auxiliary features include at least one of body shape features, clothing features, and carried items features. The multidimensional auxiliary features are compared with the auxiliary reference features associated with the personnel to be admitted to the campus in the registration database for similarity. If the comparison result is greater than the preset compensation verification threshold, the verification result is corrected to pass verification; if the comparison result is less than the preset compensation verification threshold, the verification result is maintained as fail verification. Based on the final verification results, a credit compensation record is generated. According to the triggering frequency of the credit compensation record, the credit evaluation data of the personnel waiting to pass through the campus is dynamically downgraded.

7. The method for optimizing campus personnel access management according to claim 1, characterized in that, The method also includes: The physical envelope information of campus personnel entering the passageway is acquired in real time, and the physical envelope information is used to characterize the volume boundary of campus personnel in physical space. Logically bind the verification result with the corresponding physical envelope information; Real-time monitoring of the edge spacing between two adjacent physical envelope information.

8. The method for optimizing campus personnel access management according to claim 7, characterized in that, The steps for controlling the release status of the passage facility based on the verification results include: If the edge spacing is less than a preset critical spacing value, the passage facility is controlled to perform a blocking action to increase the distance between two adjacent physical envelope information. After detecting that the previous physical envelope information has completely left the sensing area of ​​the access facility, the verification permission for the next physical envelope information is released.

9. The method for optimizing campus personnel access management according to claim 7, characterized in that, The step of acquiring the physical envelope information of campus personnel entering the access channel in real time includes: Obtain the motion vector field distribution information within the physical envelope information; Identify independent motion units in the physical envelope information based on the motion vector field distribution information; If the difference in motion vector between the independent motion units in two adjacent physical envelope information is less than a preset consistency deviation threshold, it is determined that there is a load coupling relationship between the two adjacent physical envelope information. The steps for controlling the release status of the passage facility based on the verification results include: When the existence of the personal load coupling relationship is determined, the preset critical distance value is dynamically reduced and compensated according to the geometric dimensions of the personal load coupling relationship, and the access facility is kept in the open state until the two adjacent physical envelope information completely leave the sensing area of ​​the access facility.

10. A campus personnel access management system for optimized operation, characterized in that, include: The data acquisition module is used to determine the credit evaluation data of campus personnel based on their historical access behavior records. The parameter monitoring module is used to acquire the congestion status parameters of the passage in real time and determine whether the congestion status parameters meet the preset trigger conditions. If the triggering condition is met, the mode activation module will activate the differentiated verification mode for the access channel; under the differentiated verification mode, the corresponding credit rating of the current campus personnel will be identified based on their credit evaluation data. The identity verification module is used to match the corresponding verification process intensity according to the credit level, verify the identity of the personnel to be passed through the campus according to the matched verification process intensity, and control the release status of the access facilities according to the verification results. The status recovery module is used to continuously monitor the congestion status parameters. When the congestion status parameters do not meet the preset recovery conditions, the differentiated verification mode is turned off.

Images