Multi-identity multi-park campus access control recognition system
By building a traffic prediction model and a fast matching library, the campus access control system identification strategy is dynamically adjusted to solve the identification delay problem during high-traffic periods and achieve efficient and accurate access control management.
Patent Information
- Application Number
- CN202510938693.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-08
- Publication Date
- 2025-09-23
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing campus access control system is prone to recognition delays during high-traffic periods, resulting in low traffic efficiency and a lack of traffic prediction and dynamic adjustment capabilities.
Through the collection and monitoring module, a traffic prediction model is built, a quick matching library is established, personnel matching is prioritized in the quick matching library, high-frequency traffic areas are screened based on the geographic coordinate system and personnel affiliation information, quick and regular matching libraries are built, and identification strategies are dynamically switched.
Quickly switch to efficient recognition mode during peak traffic hours to reduce matching range and time, improve recognition speed and accuracy, and optimize access control management processes.
Smart Images
Figure CN120689960A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of intelligent access control recognition, and in particular relates to a multi-identity and multi-campus campus access control recognition system. Background Art
[0002] As campuses continue to expand, the flow of people on campus is becoming increasingly frequent. This is especially true in multi-campus environments. Efficient and secure access and management of personnel have become significant challenges. Specifically, existing access control systems face the following core issues when dealing with high traffic volumes: Traditional access control systems are prone to recognition delays during high-traffic periods, causing people to queue and affect traffic efficiency. Due to the lack of real-time monitoring and prediction capabilities of personnel flow, the recognition mode cannot be dynamically adjusted according to traffic changes, and it is impossible to quickly switch to a more efficient recognition strategy during traffic peaks. To this end, we propose a multi-identity and multi-campus campus access control recognition system. Summary of the Invention
[0003] The purpose of the present invention is to provide a multi-identity and multi-campus campus access control identification system to solve the problems raised in the above background technology.
[0004] To achieve the above-mentioned object, the present invention provides the following technical solutions: a multi-identity and multi-campus campus access control recognition system, comprising: a collection and monitoring module, a core area construction module, and a fast matching library construction module; The data collection and monitoring module uses the park entrance and exit as the monitoring area, divides the monitoring period and collection time into data collection, builds a traffic prediction model, analyzes the corresponding traffic core vector, time period core vector, and historical traffic trend vector within the sliding window, substitutes them into the traffic prediction model, outputs the traffic prediction value, and determines whether to trigger the rapid identification mode based on the traffic prediction value; Fast matching library construction module: Build the campus geographic coordinate system, screen and analyze the collection of high-frequency access control points, associate the campus building coordinates to build the core access area, combine the personnel's belonging coordinates with the cross-campus access status, screen the fast matching personnel and build the fast matching library, and generate the regular matching library at the same time; Access control matching module: collects biometric information of passers-by. If the quick recognition mode is triggered, it will be matched first over the quick matching library. If the quick matching library fails to match, the person's biometric information will be matched with the regular matching library.
[0005] Preferably, the data acquisition process of the acquisition monitoring module is: For each park on campus, the entrance and exit areas of the park are marked as monitoring areas; the park opening hours are divided into several monitoring periods, and several collection times are set within the monitoring period; For the current monitoring period, monitor the number of people entering the park, the number of people leaving the park, and the average movement speed of people at each collection time; Count the number of human targets at each collection time in the monitoring area to obtain the area of the monitoring area. Divide the number of human targets at each collection time by the area of the monitoring area to obtain the human density at each collection time. Obtain historical data from the past three months and classify it by date type. For each date type, extract the total number of people entering and leaving the park, the density of people in the monitoring area, and the average movement speed of people corresponding to each monitoring period, organize them, and construct a cluster vector. The cluster vector set under each date type is analyzed using density clustering algorithm; the monitoring period is divided into different clusters; For each cluster, the mean value of each parameter component in the cluster vector corresponding to the cluster is calculated during the monitoring period, and different weight coefficients are assigned to different parameter component means. Then, a weighted calculation is performed to obtain the cluster flow value. The cluster flow values corresponding to each cluster are sorted out to construct a time period-flow value correspondence table; the current monitoring period and the corresponding date type are substituted into the time period-flow value correspondence table for matching, and the cluster flow value corresponding to the current monitoring period is output; Obtain from the campus activity reporting platform whether there is a special activity event E in each monitoring period. If there is a special activity event in the current monitoring period, E=1, otherwise E=0.
[0006] Preferably, the process of obtaining the traffic core vector in the sliding window is: A sliding time window is set within the monitoring period. The window length is n acquisition moments, and the sliding step is set to m acquisition moments, where m <n; The number of people entering the park at n collection moments within the window , Number of people leaving the park , average moving speed of personnel and population density , perform statistical calculations: , and obtain the traffic core vector F, where i is the number of the collection time, i=1, 2, ..., n.
[0007] Preferably, the analysis process of the time period core vector and the historical traffic trend vector is: The cluster flow value L corresponding to the monitoring period corresponding to the collection moment with the highest frequency in the window is integrated with the special event flag E to construct the core vector of the period. , if the monitoring period E corresponding to any collection moment is 1, then E is set to 1, otherwise it is 0; Get the historical rate of change of traffic entering the park within the current window , Historical rate of change of traffic leaving the park and the historical population density change rate , and organize and construct historical traffic trend vectors .
[0008] Preferably, the specific process of obtaining the traffic prediction value and determining whether to trigger the rapid identification mode according to the traffic prediction value is as follows: Count the total number of people entering and exiting the park gate within the sliding time window, and record it as the personnel flow value; collect the flow core vector, time period core vector, historical flow trend vector and personnel flow value within each historical sliding time window; The flow core vector, time period core vector, and historical flow trend vector corresponding to each historical sliding time window are used as independent variables, and the personnel flow value in the next window corresponding to each sliding time window is used as the dependent variable. The independent and dependent variable data are fitted and trained using the least squares method to establish a linear relationship equation between the independent and dependent variables, thereby obtaining a personnel flow value prediction model. The traffic core vector, time period core vector and historical traffic trend vector in the current sliding time window are input into the personnel flow value prediction model, and the personnel flow prediction value in the current next sliding window is output, which is recorded as the traffic prediction value; A traffic prediction threshold is preset. If the traffic prediction value corresponding to the current sliding time window is greater than or equal to the corresponding threshold, the rapid recognition mode is triggered; When the traffic prediction value corresponding to the current sliding time window is less than the corresponding threshold, the fast recognition mode stops.
[0009] Preferably, the specific process of the fast matching library construction module to construct the park geographic coordinate system and screen and analyze the set of high-frequency access control points is as follows: Taking the upper left corner of the park as the origin, establish a plane rectangular coordinate system with the X axis pointing right and the Y axis pointing downward to obtain the park's geographic coordinate system. In the park's geographic coordinate system, mark each access control point in the park; Obtain the average personnel flow value corresponding to each access control point during the current monitoring period in the past month. If the average personnel flow value corresponding to an access control point is greater than the corresponding threshold, mark the access control point as a high-frequency access control point; organize all high-frequency access control points into a high-frequency access control point set.
[0010] Preferably, the specific process of constructing the core traffic area is: For each high-frequency access control point, calculate the Euclidean distance between the high-frequency access control point and the coordinates of the center points of each building in the park to obtain the associated Euclidean value; organize all building center points that meet the associated Euclidean value less than or equal to the preset associated Euclidean threshold to form an associated building point set; Merge the associated building point sets corresponding to all high-frequency access control points and remove duplicate points to obtain a global associated building point set; Merge the high-frequency access control point set with the global associated building point set to obtain the core access associated point set; The core pass-through point set is processed through the triangulation algorithm to generate a triangular mesh. The edges of all triangles in the triangulation network are traversed, the number of occurrences of each edge is counted, and the endpoints corresponding to the edges that appear only once are extracted and marked as core boundary points. In a clockwise direction, the core boundary points are connected in sequence through smooth curves to obtain the core traffic area; Obtain the global biometric database stored by the campus access control system. Each person in the global biometric database is assigned corresponding building coordinates. For each person in the global biometric database, if the coordinates of the building to which they belong are within the core access area, the person will be marked as a quick matching preliminary candidate.
[0011] Preferably, the specific process of screening quick matching personnel and building a quick matching library, while generating a regular matching library is as follows: For each quick match candidate, extract all parks that can be accessed except the current park in their permission configuration to form the cross-park access set of the candidate; For each park in the cross-park access set, retrieve the park personnel record table within the current sliding time window from the access control system; check whether the current person is in the table; If the current person exists in the park personnel record table of any park in the cross-park pass set, this person will be marked as an excluded person; Mark all quick match primaries who are not marked as excluded persons as quick match persons; The biometric information of all quick matching persons is organized and constructed into a quick matching library; after removing the persons in the quick matching library and their biometric information from the global person biometric library, the biometric information of the remaining persons is organized and constructed into a regular matching library.
[0012] Preferably, the specific working process of the access control matching module is: For each campus, when the rapid identification mode is triggered, a rapid matching library is built. When a person passes through the campus access control, the person's biometric information is collected. The biometric information includes: facial feature information, fingerprint information, iris information, and campus card information. Prioritize matching the identified person's biometric information with the quick matching database. If the match is successful, the access control will be opened. If the match fails, the person's biometric information will be matched with the regular matching library. If the match is successful, the access control will be opened. Otherwise, if the match is unsuccessful, the person will be prohibited from entering.
[0013] Compared with the prior art, the present invention has the following beneficial effects: (1) This multi-identity and multi-campus campus access control recognition system obtains the personnel flow data of the campus entrance and exit in real time through the collection and monitoring module, and conducts in-depth analysis in combination with historical data, and can automatically switch the recognition strategy according to the flow change; in the embodiment, the system establishes a personnel flow value prediction model through the flow core vector, time period core vector and historical flow trend vector in the sliding time window; during the period of stable flow, the conventional matching method is used for access control recognition; and during the period of peak flow, it quickly switches to the fast matching mode to ensure the efficiency and accuracy of the recognition process, so that the access control system can maintain the best performance in different flow scenarios.
[0014] (2) This multi-identity and multi-campus campus access control recognition system uses a fast matching library construction module to establish a campus geographic coordinate system and screen out a set of high-frequency access control points; for these high-frequency access control points, calculate their Euclidean distances to the center points of each building in the campus, build a set of associated building points, and then obtain a core access associated point set; use a triangulation algorithm to determine the core access area, and combine the building coordinates of the personnel in the global personnel biometric database to screen out the preliminary personnel for fast matching; then, based on the personnel's cross-campus access status, exclude personnel who are already in other campuses, and finally determine the fast matching personnel and build a fast matching library, and generate a regular matching library at the same time; this process accurately locates the high-frequency access area and personnel who may pass quickly. When matching access control, priority is given to matching in the fast matching library, which reduces the matching range and time, and improves the speed and accuracy of access control recognition; for example, for teachers and students who frequently move around near the core teaching area, the system can quickly identify and release them, optimizing the campus access control management process.
[0015] (3) This multi-identity and multi-campus campus access control recognition system, when executing in the quick recognition mode, when a person passes through the access control, it is first matched in the quick matching library, and if it fails, it is transferred to the regular matching library; the regular matching library only stores the biometric information of the remaining persons after removing the persons and their biometric information in the global person biometric library from the quick matching library; in this way, the operation of repeated recognition of the biometric information of persons that have been screened in the quick matching library is avoided, which effectively improves the access control recognition efficiency and reduces unnecessary computing resource consumption. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 It is a flowchart of the present invention. DETAILED DESCRIPTION
[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0018] Example 1: Please refer to Figure 1 The present invention provides a multi-identity and multi-campus campus access control recognition system, including: a collection and monitoring module, a fast matching library construction module and an access control matching module; The collection and monitoring module uses the park entrance and exit as the monitoring area, divides the monitoring period and collection time, builds a traffic prediction model, analyzes the corresponding traffic core vector, time period core vector and historical traffic trend vector in the sliding window, substitutes them into the traffic prediction model, outputs the traffic prediction value, and determines whether to trigger the rapid identification mode based on the traffic prediction value. The specific process is as follows: For each park on campus, the entrance and exit areas of the park are marked as monitoring areas; the park opening hours are divided into several monitoring periods, and several collection times are set within the monitoring period; During the current monitoring period, the millimeter-wave radar is used to monitor the number of people entering the park, the number of people leaving the park, and the average movement speed of people at each collection moment; The thermal imaging sensor is used to scan and identify people in the monitoring area, and the number of human targets corresponding to each collection time is counted to obtain the area of the monitoring area. The number of human targets corresponding to each collection time is divided by the area of the monitoring area to obtain the personnel density corresponding to each collection time.
[0019] Obtain historical data from the past three months and classify it by date type, including weekdays, weekends, and holidays. For each date type, extract the total number of people entering and leaving the park, the density of people in the monitoring area, and the average movement speed of people corresponding to each monitoring period, organize them, and construct a cluster vector. The cluster vector sets under each date type are analyzed using density clustering algorithms (such as DBSCAN); the monitoring period is divided into different clusters; For each cluster, the mean value of each parameter component in the cluster vector corresponding to the cluster is calculated during the monitoring period, and different weight coefficients are assigned to different parameter component means. Then, a weighted calculation is performed to obtain the cluster flow value. The cluster flow values corresponding to each cluster are sorted out to construct a time period-flow value correspondence table; the current monitoring period and the corresponding date type are substituted into the time period-flow value correspondence table for matching, and the cluster flow value corresponding to the current monitoring period is output.
[0020] Obtain information from the campus activity reporting platform regarding whether there are special activity events E in each monitoring period. If there are special activity events in the current monitoring period, E=1; otherwise, E=0. A sliding time window is set within the monitoring period. The window length is n acquisition moments, and the sliding step is set to m acquisition moments, where m <n; The number of people entering the park at n collection moments within the window , Number of people leaving the park , average moving speed of personnel and population density , perform statistical calculations: , we get the traffic core vector F; where i is the number of the collection time, i=1, 2, ..., n; this vector comprehensively reflects the total number of people entering and leaving the window, the average mobility activity, and the extreme value of the population density, and intuitively reflects the scale and distribution characteristics of the park traffic within the sliding time window.
[0021] The cluster flow value L corresponding to the monitoring period corresponding to the collection moment with the highest frequency in the window is integrated with the special event flag E to construct the core vector of the period. , if the monitoring period E corresponding to any collection moment is 1, then E is set to 1, otherwise it is 0; By using date types and special events to describe temporal patterns and sudden disturbances, and combining time-period clustered traffic values, we can achieve fine-grained calibration of the time dimension of traffic forecasts. Get the historical rate of change of traffic entering the park within the current window , Historical rate of change of traffic leaving the park and the historical population density change rate , and organize and construct historical traffic trend vectors ; Furthermore, the calculation formula for the historical rate of change of traffic into the park is: It is the average cumulative number of people entering the park during the same historical period (same date type, same window time range); similarly, the historical rate of change of traffic leaving the park and the historical rate of change of population density are calculated.
[0022] By analyzing the absolute change trend (increase / decrease) of historical traffic, we can identify whether the current traffic has deviated from the normal pattern, assist in determining the degree and direction of traffic anomalies, and improve the prediction model's sensitivity to trend changes. Count the total number of people entering and exiting the park gate within the sliding time window, and record it as the personnel flow value; collect the flow core vector, time period core vector, historical flow trend vector and personnel flow value within each historical sliding time window; The traffic core vector, time period core vector and historical traffic trend vector corresponding to each historical sliding time window are used as independent variables, and the personnel flow value in the next window corresponding to each sliding time window is used as the dependent variable. The independent variable and dependent variable data are fitted and trained by the least squares method, and the linear relationship equation between the independent variable and the dependent variable is established to obtain the personnel flow value prediction model.
[0023] The traffic core vector, time period core vector and historical traffic trend vector in the current sliding time window are input into the personnel flow value prediction model, and the personnel flow prediction value in the current next sliding window is output, which is recorded as the traffic prediction value; A traffic prediction threshold is preset. If the traffic prediction value corresponding to the current sliding time window is greater than or equal to the corresponding threshold, the rapid recognition mode is triggered; When the traffic prediction value corresponding to the current sliding time window is less than the corresponding threshold, the fast recognition mode stops.
[0024] The access control matching module collects the biometric information of passers-by. If the fast recognition mode is triggered, it will be matched first over the fast matching library. If the fast matching library fails, the biometric information of the person will be matched with the regular matching library. The specific process is as follows: For each park, when the quick identification mode is triggered, a quick matching library is built. When the park access control allows people to pass, the people's biometric information is collected. The people's biometric information includes: facial feature information, fingerprint information, iris information and campus card information; the identified people's biometric information is matched with the quick matching library first. If the match is successful, the access control is opened; if the match fails, the people's biometric information is matched with the regular matching library. If the match is successful, the access control is opened. Otherwise, if the match is unsuccessful, the people are prohibited from entering.
[0025] Furthermore, if the quick matching module is not triggered, normal mode matching is implemented, that is, the biometric information of the personnel is directly matched with the global biometric database of the park.
[0026] The quick matching library construction module constructs the park's geographic coordinate system, screens and analyzes the set of high-frequency access control points, associates the park's building coordinates to construct the core access area, combines the personnel's belonging coordinates with the cross-park access status, screens the quick matching personnel and builds the quick matching library, and generates the regular matching library at the same time. The specific process is as follows: Taking the upper left corner of the park as the origin, establish a plane rectangular coordinate system with the X axis pointing right and the Y axis pointing downward to obtain the park's geographic coordinate system. In the park's geographic coordinate system, mark the various access control points in the park, where the access control point is the coordinate of the center point where the access control is set up; obtain the average personnel flow corresponding to each access control point in the current monitoring period in the past month, preset the access control point flow threshold, and if the average personnel flow corresponding to the access control point is greater than the corresponding threshold, mark the access control point as a high-frequency access control point; organize all high-frequency access control points into a high-frequency access control point set For each high-frequency access control point, the Euclidean distance between the high-frequency access control point and the coordinates of the center points of each building in the park is calculated to obtain the associated Euclidean value; a preset associated Euclidean threshold is set, and all building center points whose associated Euclidean value is less than or equal to the preset associated Euclidean threshold are sorted into an associated building point set; Merge the associated building point sets corresponding to all high-frequency access control points and remove duplicate points to obtain a global associated building point set; Merge the high-frequency access control point set with the global associated building point set to obtain the core access associated point set; The core pass-through point set is processed through the Delaunay triangulation algorithm to generate a triangular mesh. The edges of all triangles in the triangulation network are traversed, the number of occurrences of each edge is counted, and the endpoints corresponding to the edges that appear only once are extracted and marked as core boundary points; the edges that appear only once are contour edges, and their endpoints are the external boundary vertices of the triangulation network.
[0027] In a clockwise direction, the core boundary points are connected in sequence through smooth curves to obtain the core traffic area; Obtain the global biometric database stored in the campus access control system. Each person in the preset global biometric database has corresponding building coordinates. The building coordinates include: teaching building 1, teaching building 2, dormitory building 1, first cafeteria, etc.
[0028] For each person in the global biometric database, if the building coordinates of the person are within the core access area, the person will be marked as a quick match candidate; For each quick match candidate, extract all parks that can be accessed except the current park in their permission configuration to form the cross-park access set of the candidate; For each park in the cross-park access set, retrieve the park personnel record table within the current sliding time window from the access control system; check whether the current person is in the table.
[0029] If the current person exists in the park personnel record table of any park in the cross-park pass set, this person will be marked as an excluded person; Mark all quick match primaries who are not marked as excluded persons as quick match persons; The biometric information of all quick matching persons is organized and constructed into a quick matching library; after removing the persons in the quick matching library and their biometric information from the global person biometric library, the biometric information of the remaining persons is organized and constructed into a regular matching library.
[0030] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.
Claims
1. A multi-identity, multi-campus campus access control recognition system, including: The acquisition and monitoring module, the fast matching library construction module, and the access control matching module are characterized by: The data collection and monitoring module uses the park entrance and exit as the monitoring area, divides the monitoring period and collection time into data collection, builds a traffic prediction model, analyzes the corresponding traffic core vector, time period core vector, and historical traffic trend vector within the sliding window, substitutes them into the traffic prediction model, outputs the traffic prediction value, and determines whether to trigger the rapid identification mode based on the traffic prediction value; Fast matching library construction module: Build the campus geographic coordinate system, screen and analyze the collection of high-frequency access control points, associate the campus building coordinates to build the core access area, combine the personnel's belonging coordinates with the cross-campus access status, screen the fast matching personnel and build the fast matching library, and generate the regular matching library at the same time; Access control matching module: collects biometric information of passers-by. If the quick recognition mode is triggered, it will be matched first over the quick matching library. If the quick matching library fails to match, the person's biometric information will be matched with the regular matching library.
2. The multi-identity and multi-campus campus access control recognition system according to claim 1 is characterized by: The process of data collection by the acquisition monitoring module is as follows: For each park on campus, the entrance and exit areas of the park are marked as monitoring areas; the park opening hours are divided into several monitoring periods, and several collection times are set within the monitoring period; For the current monitoring period, monitor the number of people entering the park, the number of people leaving the park, and the average movement speed of people at each collection time; Count the number of human targets at each collection time in the monitoring area to obtain the area of the monitoring area. Divide the number of human targets at each collection time by the area of the monitoring area to obtain the human density at each collection time. Obtain historical data from the past three months and classify it by date type. For each date type, extract the total number of people entering and leaving the park, the density of people in the monitoring area, and the average movement speed of people corresponding to each monitoring period, organize them, and construct a cluster vector. The cluster vector set under each date type is analyzed using density clustering algorithm; the monitoring period is divided into different clusters; For each cluster, the mean value of each parameter component in the cluster vector corresponding to the cluster is calculated during the monitoring period, and different weight coefficients are assigned to different parameter component means. Then, a weighted calculation is performed to obtain the cluster flow value. The cluster flow values corresponding to each cluster are sorted out to construct a time period-flow value correspondence table; the current monitoring period and the corresponding date type are substituted into the time period-flow value correspondence table for matching, and the cluster flow value corresponding to the current monitoring period is output; Obtain from the campus activity reporting platform whether there is a special activity event E in each monitoring period. If there is a special activity event in the current monitoring period, E=1, otherwise E=0.
3. The multi-identity, multi-campus campus access control recognition system according to claim 2, characterized in that: The process of obtaining the traffic core vector in the sliding window is: A sliding time window is set within the monitoring period. The window length is n acquisition moments, and the sliding step is set to m acquisition moments, where m <n; The number of people entering the park at n collection moments within the window , Number of people leaving the park , average moving speed of personnel and population density , perform statistical calculations: , and obtain the traffic core vector F, where i is the number of the collection time, i=1, 2, ..., n.
4. The multi-identity, multi-campus campus access control recognition system according to claim 3, characterized in that: The analysis process of the period core vector and the historical traffic trend vector is as follows: The cluster flow value L corresponding to the monitoring period corresponding to the collection moment with the highest frequency in the window is integrated with the special event flag E to construct the core vector of the period. , if the monitoring period E corresponding to any collection moment is 1, then E is set to 1, otherwise it is 0; Get the historical rate of change of traffic entering the park within the current window , Historical rate of change of traffic leaving the park and the historical population density change rate , and organize and construct historical traffic trend vectors .
5. The multi-identity and multi-campus campus access control recognition system according to claim 4 is characterized by: The specific process of obtaining the traffic prediction value and determining whether to trigger the fast identification mode based on the traffic prediction value is as follows: Count the total number of people entering and exiting the park gate within the sliding time window, and record it as the personnel flow value; collect the flow core vector, time period core vector, historical flow trend vector and personnel flow value within each historical sliding time window; The flow core vector, time period core vector, and historical flow trend vector corresponding to each historical sliding time window are used as independent variables, and the personnel flow value in the next window corresponding to each sliding time window is used as the dependent variable. The independent and dependent variable data are fitted and trained using the least squares method to establish a linear relationship equation between the independent and dependent variables, thereby obtaining a personnel flow value prediction model. The traffic core vector, time period core vector and historical traffic trend vector in the current sliding time window are input into the personnel flow value prediction model, and the personnel flow prediction value in the current next sliding window is output, which is recorded as the traffic prediction value; A traffic prediction threshold is preset. If the traffic prediction value corresponding to the current sliding time window is greater than or equal to the corresponding threshold, the rapid recognition mode is triggered; When the traffic prediction value corresponding to the current sliding time window is less than the corresponding threshold, the fast recognition mode stops.
6. The multi-identity, multi-campus campus access control recognition system according to claim 5, characterized in that: The specific process of the fast matching library construction module to build the park's geographic coordinate system and screen and analyze the set of high-frequency access control points is as follows: Taking the upper left corner of the park as the origin, establish a plane rectangular coordinate system with the X axis pointing right and the Y axis pointing downward to obtain the park's geographic coordinate system. In the park's geographic coordinate system, mark each access control point in the park; Obtain the average personnel flow value corresponding to each access control point during the current monitoring period in the past month. If the average personnel flow value corresponding to an access control point is greater than the corresponding threshold, mark the access control point as a high-frequency access control point; organize all high-frequency access control points into a high-frequency access control point set.
7. The multi-identity, multi-campus campus access control recognition system according to claim 6, characterized in that: The specific process of constructing the core traffic area is as follows: For each high-frequency access control point, calculate the Euclidean distance between the high-frequency access control point and the coordinates of the center points of each building in the park to obtain the associated Euclidean value; organize all building center points that meet the associated Euclidean value less than or equal to the preset associated Euclidean threshold to form an associated building point set; Merge the associated building point sets corresponding to all high-frequency access control points and remove duplicate points to obtain a global associated building point set; Merge the high-frequency access control point set with the global associated building point set to obtain the core access associated point set; The core pass-through point set is processed through the triangulation algorithm to generate a triangular mesh. The edges of all triangles in the triangulation network are traversed, the number of occurrences of each edge is counted, and the endpoints corresponding to the edges that appear only once are extracted and marked as core boundary points. In a clockwise direction, the core boundary points are connected in sequence through smooth curves to obtain the core traffic area; Obtain the global biometric database stored by the campus access control system. Each person in the global biometric database is assigned corresponding building coordinates. For each person in the global biometric database, if the coordinates of the building to which they belong are within the core access area, the person will be marked as a quick matching preliminary candidate.
8. The multi-identity, multi-campus campus access control recognition system according to claim 7, characterized in that: The specific process of screening quick matching personnel and building a quick matching library, while generating a regular matching library, is as follows: For each quick match candidate, extract all parks that can be accessed except the current park in their permission configuration to form the cross-park access set of the candidate; For each park in the cross-park access set, retrieve the park personnel record table within the current sliding time window from the access control system; check whether the current person is in the table; If the current person exists in the park personnel record table of any park in the cross-park pass set, this person will be marked as an excluded person; Mark all quick match primaries who are not marked as excluded persons as quick match persons; The biometric information of all quick matching persons is organized and constructed into a quick matching library; after removing the persons in the quick matching library and their biometric information from the global person biometric library, the biometric information of the remaining persons is organized and constructed into a regular matching library.
9. The multi-identity and multi-campus campus access control recognition system according to claim 1, characterized in that: The specific working process of the access control matching module is as follows: For each campus, when the rapid identification mode is triggered, a rapid matching library is built. When a person passes through the campus access control, the person's biometric information is collected. The biometric information includes: facial feature information, fingerprint information, iris information, and campus card information. Prioritize matching the identified person's biometric information with the quick matching database. If the match is successful, the access control will be opened. If the match fails, the person's biometric information will be matched with the regular matching library. If the match is successful, the access control will be opened. Otherwise, if the match is unsuccessful, the person will be prohibited from entering.