Campus safety control system integrating security monitoring and visitor management

Abstract

The invention discloses a campus safety control system integrating security monitoring and visitor management, and relates to the technical field of campus safety control, identity information, school entering time and a predicted staying area of visitors are collected when the visitors enter the school, and visitor labels are constructed; matching a visitor behavior reference model based on the historical behavior data, and generating a path track set and a behavior feature vector; calling a monitoring image in real time to extract a current behavior feature, performing similarity matching on the current behavior feature and a reference feature, and calculating a behavior deviation degree; carrying out spatial analysis in combination with the current position of the visitor and the reference trajectory offset section, and triggering an abnormal behavior alarm instruction; carrying out area early warning, abnormal track uploading and high-risk label marking; meanwhile, the feedback data is used for dynamic updating and incremental learning of the behavior model; according to the method, the whole-process dynamic monitoring and intelligent judgment of visitor behaviors are realized, and the real-time performance and expandability of a campus security and protection system are improved.

Description

Technical Field

[0001] This invention relates to the field of campus security control technology, specifically to a campus security control system that integrates security monitoring and visitor management. Background Technology

[0002] Currently, access control on campuses primarily relies on gatekeeper registration or visitor system authentication, with some schools deploying camera security systems for basic video surveillance. However, traditional security and visitor management systems operate independently, lacking data linkage and failing to provide real-time alerts and coordinated control for abnormal visitor behavior.

[0003] More seriously, in lower elementary schools or special education institutions, there is a significant age difference between visitors and students, making identification difficult. Some unauthorized visitors enter the school through ambiguous identities or brief interactions, increasing the risk of security incidents. Traditional systems struggle to determine a visitor's true purpose based on a single visitor identity, and cannot proactively intercept them by combining historical behavioral patterns.

[0004] Furthermore, the current system is almost unable to track the activity paths of visitors after they enter the campus, especially in campus corners with many blind spots in monitoring or insufficient network signal coverage, such as old teaching buildings and outdoor activity areas, which are very likely to become potential risk blind spots and lack a systematic and proactive risk identification mechanism. Summary of the Invention

[0005] The purpose of this invention is to provide a campus security control system that integrates security monitoring and visitor management to address the shortcomings of the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a campus security control system integrating security monitoring and visitor management, comprising: Information collection module: Collect visitor identity information V1 at the campus entrance, record their entry time T1 and expected stay area L1, and construct visitor tag set V; Behavior matching module: Based on visitor identity information V1 and its historical school visit records, match the corresponding visitor behavior reference model M1 from the visitor behavior model library; Behavioral modeling module: Based on model M1, obtain the typical activity path trajectory set P1 of the corresponding visitor in the campus, and the behavioral feature vector set F1 of each path node; Behavior recognition module: retrieves real-time security monitoring image streams on campus, extracts the actual behavior feature vector F2(t) of visitors at each time period t, and performs similarity matching between F2(t) and F1 to obtain the deviation index D(t); Anomaly detection module: If D(t) exceeds the preset threshold, it automatically generates an abnormal behavior alarm command by combining the visitor's current actual location information L2(t) with the offset segment of the path trajectory set P1. Linked Early Warning Module: Based on abnormal behavior alarm commands, the module issues early warnings for the area where the visitor is located, uploads the abnormal trajectory to the campus security terminal, and simultaneously marks the visitor as a high-risk individual. ; Model Update Module: Updates the feedback records of the corresponding model M1 in the visitor behavior model library.

[0007] Preferably, the step of matching the corresponding visitor behavior reference model M1 from the visitor behavior model library includes: The collected visitor identity information V1 is structured and encoded to generate feature vectors for model retrieval. The feature vector This includes visitor identity category, visit frequency, and average stay duration; Based on feature vectors Perform similarity matching retrieval in the visitor behavior model library, call the cosine similarity algorithm to calculate its similarity score with the existing behavior model index, and select the model with the highest score as the reference model M1.

[0008] Preferably, the step of obtaining the typical activity path trajectory set P1 of the corresponding visitor on campus based on model M1, and the behavioral feature vector set F1 of each path node, includes: Historical trajectory template data is extracted from the reference model M1, and geographic coordinate mapping is performed on the commonly used routes corresponding to visitor identity categories based on the campus electronic map to generate a preliminary set of routes. ; right Clustering of continuous coordinate points in the data, identifying key stopping points and turning points, and constructing a simplified path trajectory set P1; Behavioral features are extracted from the monitoring image segments at each node, and a four-dimensional feature vector containing walking speed, dwell time, direction change rate and interaction frequency is calculated to form the node behavioral feature set F1. The generated path trajectory set P1 is bound to the feature vector set F1 over time to construct a behavior reference template that can be dynamically updated with visitors.

[0009] Preferably, the step of retrieving the real-time campus security monitoring image stream, extracting the visitor's actual behavior feature vector F2(t) at each time period t, and performing similarity matching between F2(t) and F1 to obtain the deviation index D(t) includes: Real-time acquisition of image frame sequences and execution of human detection and target tracking to obtain the visitor's movement trajectory and keyframe images within time period t; Image features are extracted from the image sequence, and walking speed, dwell time, rate of change of direction and frequency of interaction are calculated to form the actual behavior feature vector F2(t); The feature vectors at the corresponding time points in F1 of the behavioral reference template are matched one-to-one according to their dimensions, and the similarity score S(t) between the two vectors is calculated by Euclidean distance. The deviation index D(t) is calculated in reverse based on the similarity score.

[0010] Preferably, the step of calculating the deviation index D(t) in reverse based on the similarity score includes: The calculated similarity score S(t) is used as input to execute the inverse transformation function. The initial deviation value is obtained; The deviation threshold Dmax is dynamically set by combining the visitor's identity category and the type of region they are in; By applying a weighted average to D(t) over a continuous time period using a sliding window method, a smoothed deviation sequence is obtained. ; If N consecutive time points If all exceed Dmax, it is identified as a high-risk event of abnormal behavior.

[0011] Preferably, the step of automatically generating an abnormal behavior alarm instruction by combining the visitor's current actual location information L2(t) with the offset segment of the path trajectory set P1 includes: Compare the visitor's location L2(t) at the current time point t with the behavioral reference path trajectory set P1 using geographic coordinates, calculate the shortest spatial distance d(t), and determine whether it falls within the offset segment of P1. If d(t) is greater than the set spatial tolerance threshold r, and the deviation point does not belong to its expected dwell area L1, it is determined to be a trajectory deviation event; By combining the current deviation index D(t) with the trajectory offset status, the risk level triggering rules are matched to determine whether the conditions for generating an alarm are met. If the triggering conditions are met, an abnormal behavior alarm command containing the visitor identifier, abnormal type, location of occurrence, and timestamp will be generated.

[0012] Preferably, the simultaneous labeling of visitors as high-risk tags... ,include: Receive the generated abnormal behavior alarm command and parse the visitor identifier, abnormal type, location L2(t) and timestamp t information contained therein; Issue early warning information to the teaching building or floor area corresponding to the visitor's current location to alert on-site personnel to abnormal behavior; The trajectory data during the period of abnormal behavior is cut from the path trajectory set P1 and marked as abnormal segment P1M, and uploaded to the campus security terminal platform with a risk level label attached. Update the visitor's identity tag from the original visitor tag V to a high-risk tag. And store it in the visitor management database.

[0013] Preferably, the update of the feedback record corresponding to model M1 in the visitor behavior model library includes: Receiving high-risk labels And the corresponding abnormal trajectory segment P1M, as the feedback input sample for the visitor behavior reference model M1; Perform trajectory difference analysis between P1M and the original behavior reference path trajectory set P1, extract the difference feature vector ΔF, and accumulate the behavior offset weight according to its occurrence frequency. Adjust the weights of the corresponding feature dimensions in the node behavior feature vector set F1 in the visitor behavior reference model M1; When the cumulative feedback samples reach a preset threshold, the local retraining process of the visitor behavior reference model M1 is triggered. The incremental learning algorithm is used to update the trajectory pattern and feature weight distribution, and the results are saved to the visitor behavior model library.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention constructs a campus security control method that integrates security monitoring and visitor behavior modeling, achieving closed-loop management of the entire process of visitors, from registration upon entry, behavior trajectory tracking, behavior feature identification to anomaly warning and model self-updating. Compared to existing security systems that rely solely on fixed paths and access control, this invention can dynamically identify the actual behavior of visitors during their time on campus and perform personalized comparisons and risk assessments based on historical behavior models.

[0015] 2. The behavioral deviation calculation, trajectory offset analysis, and high-risk labeling mechanism proposed in this invention not only provides a quantifiable basis for risk assessment but also effectively solves the problems of static models and difficulty in adapting to diverse visitor behaviors in existing technologies through feedback-driven continuous model learning and optimization. Through multi-module linkage and closed-loop data updates, this invention significantly enhances visitor security management capabilities in campus scenarios and is applicable to visitor management in different types of campuses. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart of the system modules of the present invention.

[0018] Figure 2 This is a flowchart of the method of the present invention.

[0019] Figure 3 The flowchart for updating the visitor behavior reference model of the present invention. Detailed Implementation

[0020] 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 embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] For examples, please refer to Figure 1 , 2 As shown in Figure 3, the campus security control system integrating security monitoring and visitor management described in this embodiment includes: Information collection module: Collect visitor identity information V1 at the campus entrance, record their entry time T1 and expected stay area L1, and construct visitor tag set V; Behavior matching module: Based on visitor identity information V1 and its historical school visit records, match the corresponding visitor behavior reference model M1 from the visitor behavior model library; Behavioral modeling module: Based on model M1, obtain the typical activity path trajectory set P1 of the corresponding visitor in the campus, and the behavioral feature vector set F1 of each path node; Behavior recognition module: retrieves real-time security monitoring image streams on campus, extracts the actual behavior feature vector F2(t) of visitors at each time period t, and performs similarity matching between F2(t) and F1 to obtain the deviation index D(t); Anomaly detection module: If D(t) exceeds the preset threshold, it automatically generates an abnormal behavior alarm command by combining the visitor's current actual location information L2(t) with the offset segment of the path trajectory set P1. Linked Early Warning Module: Based on abnormal behavior alarm commands, the module issues early warnings for the area where the visitor is located, uploads the abnormal trajectory to the campus security terminal, and simultaneously marks the visitor as a high-risk individual. ; Model Update Module: Updates the feedback records of the corresponding model M1 in the visitor behavior model library.

[0022] In this embodiment of the invention, the information collection module is used to collect and mark the identity information and initial behavioral state of visitors entering the campus, as the basis for subsequent behavior modeling and analysis.

[0023] Specifically, the information collection module includes an identity recognition unit, a time recording unit, and a region preset unit: The identity recognition unit is used to collect the visitor's identity information V1 when the visitor passes through the campus entrance gate, access control system or visitor registration terminal. The identity information includes, but is not limited to, visitor's name, ID number, contact information, visitor's name, and reason for visit. It can be automatically obtained through ID card scanning, facial recognition or visitor registration QR code. The time recording unit is used to synchronously record the visitor's entry time T1, which serves as the starting point for behavioral analysis. This time record can be obtained by timestamping the trigger signal from the access control device or facial recognition camera. The area preset unit is used to determine the expected activity area L1 of a visitor based on visitor registration information, the location of the person being visited, or system preset rules. L1 can be a teaching building, an office, or multiple designated functional areas, serving as the geographical restriction parameter for the visitor tag.

[0024] After completing the recording of V1, T1, and L1, the information collection module encapsulates the three into a visitor tag set V={V1,T1, L1} and stores the tag set in the visitor behavior analysis database for subsequent behavior model matching and monitoring data comparison.

[0025] In a preferred embodiment of the present invention, the behavior matching module is used to dynamically match the most suitable reference model M1 for the current visitor from the visitor behavior model library based on the visitor's historical behavior characteristics after the initial identity collection is completed. Specifically, the behavior matching module includes: First, the collected visitor identity information V1 is structured and encoded to generate feature vectors for matching behavioral models. The aforementioned It is a multidimensional numerical vector, which includes, but is not limited to, the following three dimensions: Visitor Identity Category: Based on the identity fields in V1 (such as "Parent", "Supplier", "Repairman", etc.), a preset identity category coding rule is established, for example: Parent = 1, Supplier = 2, Repairman = 3; Visit Frequency: The number of times the visitor has visited in the past 30 days, f, is counted as the frequency dimension value; Average Stay Duration: The average stay duration t_avg (in minutes) of each visit is calculated and rounded to one decimal place. The final feature vector is obtained. The format is [C, f, t_avg], where C is the identity category code, f is the visit frequency, and t_avg is the average dwell time. The visitor behavior model library is a multi-dimensional feature index database. Each model M(i) consists of the following: model number M(i), corresponding to a unique visitor behavior sample; historical feature vector. The corresponding path trajectory set P(i); and the behavioral feature vector set F(i) (used for subsequent similarity comparison). This library is built based on historical visitor records, automatically updated daily, and merges similar visitor behaviors through clustering methods (such as K-means) to improve model compactness and coverage.

[0026] In acquiring Next, similarity retrieval is performed in the behavior model library. Specifically: the cosine similarity algorithm is called to perform similarity retrieval on the current feature vector. Compared with all historical model vectors in the model library Calculate similarity; the cosine similarity calculation method is as follows: and Treating each element as a three-dimensional vector, calculate the cosine of their included angle, with a value ranging from 0 to 1; a larger value indicates higher similarity. Select the model M(i) with the highest similarity score (i.e., the largest cosine of the included angle) and use it as the reference behavior model M1 for this visitor. For example, when the current visitor's... Given the value [2, 5, 45.5], compare it with all vectors in the model library one by one, calculate each cosine value, and finally select the one with the highest score as M1.

[0027] After matching is completed, the selected model M1 is associated and bound with the current visitor's tag set V to form a data pair (V, M1), and pushed to the behavior recognition module for subsequent generation of the visitor's path trajectory set P1 and behavior feature vector set F1 on campus.

[0028] In a preferred embodiment of the present invention, after the behavior modeling module matches the behavior reference model M1 of the corresponding visitor, it further performs path modeling and behavior feature extraction operations to construct a typical path trajectory set P1 suitable for this visitor type and a set of behavior feature vectors F1 for each node, for subsequent real-time behavior deviation judgment and anomaly detection. Specifically, this includes: First, historical trajectory template data for visitors of the corresponding identity categories is extracted from the selected reference model M1. This trajectory template consists of multiple location records of historical visitors within the campus, with each record containing a timestamp, location information, and corresponding identity tag.

[0029] The campus electronic map module (GIS engine) is invoked to map the aforementioned location data into a corresponding sequence of two-dimensional geographic coordinates. For example, if the visitor is categorized as a "parent," the mapped path will prioritize covering Teaching Building A, Administration Building B, and the school gate area. After merging all similar trajectories, a preliminary path set is obtained. Each path is an ordered set of multiple geographic coordinates, representing the common routes used by this type of visitor.

[0030] Due to the initial path set This includes redundant coordinate points and abnormal offset trajectories. To improve the efficiency of behavior modeling, this step... Perform cluster analysis.

[0031] Density-based spatial clustering algorithms (such as DBSCAN) are used to group points with a spatial distance of less than a preset radius r (e.g., 2 meters) and a high dwell density among consecutive geographic coordinate points into the same cluster. Based on the clustering results, the following two types of key nodes are identified: Points of interest: such as classroom entrances, waiting areas, etc.; Turning points: such as stairwells, crossroads, etc.

[0032] These key nodes are organized sequentially to form a simplified path trajectory set P1, which is used to describe typical movement patterns of visitors on campus.

[0033] For each key node in the path trajectory set P1, the corresponding area's surveillance image segment is retrieved to extract visitor behavior features. Specifically, the following four metrics are calculated as dimensions of the node behavior feature vector: Walking speed: calculated by the visitor displacement and time difference between image frames, in meters per second; Dwell time: The time a visitor spends at this node, in seconds; Directional change rate: Analyzes the change in the angle between the paths entering and leaving the node, and quantifies the degree of directional deviation; Interaction frequency: This measures the frequency with which visitors interact or communicate with others at a node, which can be achieved through human detection and posture recognition algorithms.

[0034] Combining the above four eigenvalues ​​forms a four-dimensional behavioral feature vector. Where v is walking speed, t is dwell time, θ is the rate of change of direction, and f is the frequency of interaction. The feature vectors of all nodes are aggregated to form a complete set of behavioral feature vectors, F1.

[0035] Finally, each key node in the path trajectory set P1 is bound to its corresponding behavior feature vector Fi, and a time series relationship is constructed according to the order in which the nodes appear. This forms the final dynamic behavior reference template structure for behavior recognition. The template format is: M1={ (P1_1, F1_1), (P1_2, F1_2), ..., (P1_n, F1_n)}, where P1_n represents the nth path node and F1_n is the corresponding feature vector.

[0036] In a preferred embodiment of the present invention, the behavior recognition module is used to compare the visitor’s actual behavior on campus with the reference behavior features generated by the behavior modeling module, and identify whether the visitor has an abnormal behavior trend through similarity calculation and deviation judgment.

[0037] First, by using network security cameras deployed in various areas of the campus, the monitoring image stream corresponding to the visitor's current location is retrieved in real time, and a continuous image frame sequence within a time period t is extracted.

[0038] We employ deep learning-based human detection algorithms, such as YOLOv5 or HRNet, to perform identity matching and target tracking on individuals appearing in image frames. This process acquires the motion trajectory points and key behavioral frames of the target visitor within the time period t, providing data input for subsequent feature extraction.

[0039] Behavioral analysis was performed on the above image frames to extract the following four dynamic features, which were used to construct the actual behavioral feature vector F2(t) for that time period: Walking speed v(t): calculated by the displacement of the target trajectory point and the time interval, in meters per second; Dwell time t_s(t): The time spent at two adjacent critical nodes, in seconds; Directional change rate θ(t): Based on the forward direction vector, the mean value of the change in directional angle between each frame is calculated, reflecting the trajectory stability; Interaction frequency f(t): The number of close-range interaction behaviors is counted by analyzing the average distance and facial orientation between the visitor and others in the image.

[0040] The final constructed F2(t) is a four-dimensional vector: [v(t), t_s(t), θ(t), f(t)], which is used to compare with the reference behavioral feature F1.

[0041] The similarity score is calculated by comparing F2(t) with the feature vector F1(t) at the corresponding time point in the behavioral reference feature vector F1 generated by the reference model M1 in the behavioral modeling module. The similarity is calculated using the Euclidean distance algorithm. Specifically: Let F2(t) = [x1, x2, x3, x4], F1(t) = [y1, y2, y3, y4], then the similarity score is calculated. The similarity score S(t) ranges from 0 to 1. The closer the value is to 1, the closer the current behavior is to the expected behavior. This similarity score will serve as the basic input for calculating the deviation.

[0042] A reverse mapping method is used to convert S(t) into a deviation index D(t) to reflect the degree of abnormality of the current behavior. The conversion function is: The value of D(t) ranges from 0 to 1, and the larger the value of D(t), the more serious the deviation in behavior.

[0043] The deviation threshold Dmax is set according to the visitor's identity category (e.g., parent, off-campus technician) and the activity area type (e.g., teaching area, administrative area). The setting of Dmax is dynamically adjusted based on the historical deviation of the mean μ and standard deviation σ of the visitor's behavior in similar scenarios. The recommended setting is: , where μ and σ can be obtained statistically during the model training phase.

[0044] Considering that short-term behavioral disturbances or equipment jitter may cause temporary fluctuations in D(t), this invention introduces a sliding time window smoothing algorithm, specifically including: setting the sliding window length to T seconds, with multiple time points D(t) within the window; and performing a weighted average of all D(t) values ​​within the window to obtain the smoothing deviation. The weights can be distributed in a decreasing order over time, with values ​​closer to the current time point having higher weights, for example, by using an exponentially decaying weight distribution.

[0045] When determining the time points at consecutive N time points (e.g., N=3) When all exceed the currently set threshold Dmax, a high-risk event flag for abnormal behavior is triggered.

[0046] In a preferred embodiment of the present invention, the anomaly detection module is used to determine whether an abnormal behavior alarm condition has been met when a visitor's behavior deviates from the reference trajectory, by comprehensively comparing the visitor's current actual location information with the path trajectory set, and automatically generating a corresponding alarm command. This module relies on spatial location analysis and rule matching strategies to achieve accurate judgment of behavioral risks.

[0047] First, the visitor's actual location information L2(t) at the current time point t is compared with the reference path trajectory set P1 generated by the behavior modeling module at the geographic coordinate level.

[0048] Specifically, let P1 be a set of coordinate points of multiple nodes, and L2(t) be the two-dimensional coordinate value (x,y) of the current location information. Then, the system calculates the Euclidean distance between L2(t) and all nodes in P1 to obtain the minimum distance value d(t), which represents the shortest spatial offset between the current visitor and the desired path.

[0049] If d(t) is less than or equal to the preset spatial tolerance threshold r (e.g., set to 3 meters), the visitor is considered to still be within the expected path range; if d(t) is greater than r, the offset status judgment stage begins. The tolerance threshold r can be dynamically adjusted based on the campus environment structure, monitoring coverage radius, and historical path deviation distribution data.

[0050] Based on the determination that d(t) exceeds the threshold r, the system further determines whether the offset point falls within the visitor's expected stay area L1. The expected stay area L1 is usually generated when the visitor registers upon entering the school, representing their normal access range (e.g., administration building, laboratory, etc.).

[0051] If L2(t) deviates from P1 but is within the range of L1, it is considered a reasonable path change and the trajectory offset flag is not triggered. If L2(t) is neither in the path buffer of P1 nor in the dwell area defined by L1, it is determined to be a trajectory offset event, and the offset time point, position and minimum distance d(t) are recorded for subsequent rule matching.

[0052] Once a trajectory deviation event is marked, the system will combine the current behavior deviation index D(t) as input parameters to enter the risk rule matching engine for alarm condition judgment.

[0053] This rule engine uses a configurable rule set based on the IF-THEN structure. Example rules include: If D(t)≥0.6 AND d(t)≥5 meters AND the current area is a restricted area, then the risk level is high. IFD(t) ≥ 0.4 AND deviation duration ≥ 60 seconds THEN Risk level = Medium; Where D(t) is the behavior deviation calculated in the behavior recognition module, and d(t) is the position offset. The duration of deviation can be calculated cumulatively over a continuous time period.

[0054] If the matching result meets the preset high or medium risk level, the current visitor behavior is deemed to meet the conditions for triggering an alarm.

[0055] After the rule matching determines that the alarm triggering conditions are met, a structured abnormal behavior alarm instruction is immediately generated, which includes the following fields: Visitor identification: such as visitor ID, name, registration information; Anomaly types: such as "high deviation behavior", "path deviates to restricted area", etc.; Location of occurrence: Geographic coordinates or floor number of the current location L2(t); Timestamp: The exact time t when the abnormal behavior was triggered; Risk level: The level label (e.g., medium, high) output by the rules engine.

[0056] In a preferred embodiment of the present invention, after receiving an abnormal behavior alarm command, the linkage early warning module not only performs regional early warning and trajectory uploading operations, but also further assesses the risk level of the visitor who triggered the abnormal behavior and updates their identity tag to a high-risk tag. This enables continuous and dynamic monitoring of suspicious individuals. The process includes the following four steps: First, it receives an abnormal behavior alarm command generated by the anomaly detection module. The alarm command is a structured data object containing the following key fields: Visitor identification information: such as a unique visitor ID, registration number, or facial recognition code; Anomaly types: such as "path deviation + high behavior deviation"; Current location L2(t): represented by latitude and longitude coordinates or building number + floor number; Timestamp t: The precise time point at which the abnormal event occurred, in milliseconds.

[0057] After parsing the instruction, the visitor ID is bound to the current location data to locate their physical location and identity attributes for subsequent operations.

[0058] To enable on-site intervention and risk notification, the system pushes early warning information to the physical area corresponding to the visitor's current location by calling the campus broadcast control module or information publishing interface. Specifically, if the location is inside a teaching building, an "Abnormal Behavior Reminder" is displayed on the corresponding floor's electronic screen; if the location is outdoors, the system can issue voice or image alerts via a voice broadcast device or LED display. The broadcast content can be configured according to the level of abnormality, such as "Medium Risk Reminder" or "High Risk Warning," and the playback duration and frequency can be configured.

[0059] Further, from the visitor's behavioral reference path trajectory set P1, trajectory segments within the time period of the abnormal behavior are extracted and denoted as abnormal segment P1M. The extraction method is as follows: based on the alarm time point t and the set pre- and post-alarm compensation window T (e.g., 10 seconds before and after), the corresponding path segment is cut out; the resulting trajectory segment P1M includes a continuous sequence of location points, changes in behavioral characteristics, dwell and turning data, etc.; P1M, along with the visitor ID and risk level label, is packaged into a trajectory anomaly report and uploaded to the campus security terminal platform or security control console. This platform is used for subsequent dispatching of security personnel, locking camera perspectives, or retrieving video playback.

[0060] Call the visitor management database interface to update the current visitor's identity tag V to a high-risk tag. And complete persistent storage. The tag structure is updated as follows: The original tag V includes: {Visitor ID, Identity Category, Entry Time T1, Expected Region L1}; High-risk tag The extended fields are: {Original V Information, Risk Level, Abnormal Trajectory ID, Marking Time T, Processing Status Flag}. Where: the risk level is based on the rule engine matching results (e.g., medium, high); the abnormal trajectory ID is associated with the P1M number uploaded to the security platform; and the processing status flag indicates whether a response or intervention has been initiated.

[0061] In a preferred embodiment of the present invention, the model update module is used to feed back high-risk behaviors generated by visitors during actual behavior into their corresponding behavioral reference model M1, thereby achieving self-learning and dynamic optimization of the behavioral model. This module improves the accuracy of model M1 in recognizing similar future behaviors through a series of processing steps, including abnormal trajectory difference analysis, feature weight correction, and incremental learning updates.

[0062] When the alert module marks a visitor as high-risk... Subsequently, the system synchronously extracts the trajectory segment P1M corresponding to its abnormal behavior and uses both as the feedback update input for model M1.

[0063] P1M is a continuous trajectory segment extracted within the time window that triggers the abnormal behavior alarm, which includes spatial coordinate sequence and behavioral feature changes; The visitor identifier field carried in the sample is used to accurately locate its corresponding behavioral reference model M1; the input sample is structured into the following form: The data is temporarily stored in the model update cache, awaiting further processing. To accurately identify the offset characteristics between anomalous behavior and the reference model, a difference analysis is performed on the original path trajectory set P1 in P1M and M1. The specific operations are as follows: Compare the coordinates of trajectory nodes in corresponding time windows or the same spatial regions in P1 and P1M, and calculate key indicators such as spatial offset Δd, directional offset angle Δθ, and dwell time difference Δt_s. The above difference values ​​are transformed into a unified four-dimensional difference feature vector ΔF = [Δv, Δt_s, Δθ, Δf], where each dimension corresponds to a behavioral feature in F1; The frequency of occurrence of each type of ΔF in historical feedback samples is statistically analyzed to form a behavior offset statistical matrix, which is used to evaluate the magnitude of weight correction for feature dimensions.

[0064] The offset weights can be calculated using frequency normalization, i.e., the weight adjustment value for each feature dimension is calculated as the frequency of occurrence divided by the total number of feedbacks.

[0065] Based on the ΔF and cumulative offset statistics extracted from the feedback samples, the feature weights of each dimension of the behavioral feature vector set F1 in model M1 are dynamically adjusted. The adjustment method is as follows: Let fi be the feature of the i-th dimension in the original feature vector F1, and wi be its corresponding weight; if the offset of this dimension occurs frequently in ΔF, the system will update wi according to the following formula: New weight , where α is the learning rate constant (e.g., 0.1), and fi_impact is the frequency-normalized weight increment factor; the weight-adjusted feature vector is retained in model M1 as a reference vector for the next behavior comparison.

[0066] The system continuously counts the number of feedback samples from all visitor behavior models. When the cumulative number of feedback samples for a particular model M1 reaches a set threshold Q (e.g., Q = 10 times), a local retraining mechanism for that model is triggered. Retraining employs an incremental learning algorithm, with the main steps as follows: The training set is composed of a mixture of historical normal trajectory data and cumulative abnormal trajectory P1M samples; typical path trajectories are regenerated using a time series clustering-based learning method (such as DTW dynamic time warping + K-means variant). With the updated behavioral feature vector set Replace P1 and F1 in the original model M1 with the retraining results, and then store the updated model back into the visitor behavior model library.

[0067] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A campus security control system integrating security monitoring and visitor management, characterized in that: include: Information collection module: Collect visitor identity information V1 at the campus entrance, record their entry time T1 and expected stay area L1, and construct visitor tag set V; Behavior matching module: Based on visitor identity information V1 and its historical school visit records, match the corresponding visitor behavior reference model M1 from the visitor behavior model library; Behavior modeling module: Based on model M1, obtain the typical activity path trajectory set P1 of the corresponding visitor in the campus, and the behavioral feature vector set F1 of each path node; Behavior recognition module: retrieves real-time security monitoring image streams on campus, extracts the actual behavior feature vector F2(t) of visitors at each time period t, and performs similarity matching between F2(t) and F1 to obtain the deviation index D(t); Anomaly detection module: If D(t) exceeds the preset threshold, it automatically generates an abnormal behavior alarm command by combining the visitor's current actual location information L2(t) with the offset segment of the path trajectory set P1. Linked Early Warning Module: Based on abnormal behavior alarm commands, the module issues early warnings for the area where the visitor is located, uploads the abnormal trajectory to the campus security terminal, and simultaneously marks the visitor as a high-risk individual. ; Model Update Module: Updates the feedback records of the corresponding model M1 in the visitor behavior model library.

2. The campus security control system integrating security monitoring and visitor management according to claim 1, characterized in that: The step of matching the corresponding visitor behavior reference model M1 from the visitor behavior model library includes: The collected visitor identity information V1 is structured and encoded to generate feature vectors for model retrieval. The feature vector This includes visitor identity category, visit frequency, and average stay duration; Based on feature vectors Perform similarity matching retrieval in the visitor behavior model library, call the cosine similarity algorithm to calculate the similarity score between it and the existing behavior model index, and select the model with the highest score as the reference model M1.

3. The campus security control system integrating security monitoring and visitor management according to claim 1, characterized in that: The process of obtaining the typical activity path trajectory set P1 of the corresponding visitor on campus based on model M1, and the behavioral feature vector set F1 of each path node, includes: Historical trajectory template data is extracted from the reference model M1, and geographic coordinate mapping is performed on the commonly used routes corresponding to visitor identity categories based on the campus electronic map to generate a preliminary set of routes. ; right Clustering of continuous coordinate points in the data, identifying key stopping points and turning points, and constructing a simplified path trajectory set P1; Behavioral features are extracted from the monitoring image segments at each node, and a four-dimensional feature vector containing walking speed, dwell time, direction change rate and interaction frequency is calculated to form the node behavioral feature set F1. The generated path trajectory set P1 is bound to the feature vector set F1 over time to construct a behavior reference template that can be dynamically updated with visitors.

4. A campus security control system integrating security monitoring and visitor management according to claim 1, characterized in that: The process involves retrieving real-time campus security monitoring image streams, extracting the visitor's actual behavioral feature vector F2(t) for each time period t, and performing similarity matching between F2(t) and F1 to obtain the deviation index D(t), including: Real-time acquisition of image frame sequences and execution of human detection and target tracking to obtain the visitor's movement trajectory and keyframe images within time period t; Image features are extracted from the image sequence, and walking speed, dwell time, rate of change of direction and frequency of interaction are calculated to form the actual behavior feature vector F2(t); The feature vectors at the corresponding time points in F1 of the behavioral reference template are matched one-to-one according to their dimensions, and the similarity score S(t) between the two vectors is calculated by Euclidean distance. The deviation index D(t) is calculated in reverse based on the similarity score.

5. A campus security control system integrating security monitoring and visitor management according to claim 4, characterized in that: in, The reverse calculation of the deviation index D(t) based on the similarity score includes: The calculated similarity score S(t) is used as input to execute the inverse transformation function. The initial deviation value is obtained; The deviation threshold Dmax is dynamically set by combining the visitor's identity category and the type of region they are in; By applying a weighted average to D(t) over a continuous time period using a sliding window method, a smoothed deviation sequence is obtained. ; If N consecutive time points If all exceed Dmax, it is identified as a high-risk event of abnormal behavior.

6. A campus security control system integrating security monitoring and visitor management according to claim 1, characterized in that: The method of automatically generating abnormal behavior alarm instructions by combining the visitor's current actual location information L2(t) with the offset segment of the path trajectory set P1 includes: Compare the visitor's location L2(t) at the current time point t with the behavioral reference path trajectory set P1 using geographic coordinates, calculate the shortest spatial distance d(t), and determine whether it falls within the offset segment of P1. If d(t) is greater than the set spatial tolerance threshold r, and the deviation point does not belong to its expected dwell area L1, it is determined to be a trajectory deviation event; By combining the current deviation index D(t) with the trajectory offset status, the risk level triggering rules are matched to determine whether the conditions for generating an alarm are met. If the triggering conditions are met, an abnormal behavior alarm command containing the visitor identifier, abnormal type, location of occurrence, and timestamp will be generated.

7. A campus security control system integrating security monitoring and visitor management according to claim 6, characterized in that: in, The visitor was simultaneously labeled as high-risk. ,include: Receive the generated abnormal behavior alarm command and parse the visitor identifier, abnormal type, location L2(t) and timestamp t information contained therein; Issue early warning information to the teaching building or floor area corresponding to the visitor's current location to alert on-site personnel to abnormal behavior; The trajectory data during the period of abnormal behavior is cut from the path trajectory set P1 and marked as abnormal segment P1M, and uploaded to the campus security terminal platform with a risk level label attached. Update the visitor's identity tag from the original visitor tag V to a high-risk tag. And store it in the visitor management database.

8. A campus security control system integrating security monitoring and visitor management according to claim 7, characterized in that: The updated feedback records for the corresponding model M1 in the visitor behavior model library include: Receiving high-risk labels And the corresponding abnormal trajectory segment P1M, as the feedback input sample for the visitor behavior reference model M1; Perform trajectory difference analysis between P1M and the original behavior reference path trajectory set P1, extract the difference feature vector ΔF, and accumulate the behavior offset weight according to its occurrence frequency. Adjust the weights of the corresponding feature dimensions in the node behavior feature vector set F1 in the visitor behavior reference model M1; When the cumulative feedback samples reach a preset threshold, the local retraining process of the visitor behavior reference model M1 is triggered. The incremental learning algorithm is used to update the trajectory pattern and feature weight distribution, and the results are saved to the visitor behavior model library.

Images