An airport video data real-time analysis system

Abstract

The application relates to the technical field of airport safety monitoring, and discloses an airport video data real-time analysis system. The system comprises a video stream space-time feature modeling module, a behavior trajectory graph building module, an abnormal area correlation analysis module, a risk level semantic judgment module and a situation structure visualization module. The system analyzes the multi-scale space-time features of the airport monitoring video stream, establishes a multi-dimensional behavior trajectory graph, analyzes the correlation of abnormal behavior areas, judges the risk level semantics, and finally generates an airport global risk situation heat distribution map. The system realizes the whole-process processing from video data collection to risk situation visualization, can accurately identify abnormal behavior areas, and clearly defines the risk level and category, thereby providing comprehensive and intuitive situation information for airport safety management and improving the intelligent level of airport safety management.

Description

Technical Field

[0001] This invention relates to the field of airport security monitoring technology, specifically to a real-time airport video data analysis system. Background Technology

[0002] With the rapid development of the air transport industry, airports, as densely populated and highly mobile transportation hubs, face unprecedented challenges in security management. Traditional airport video surveillance systems largely rely on manual inspections or single-dimensional behavior recognition technologies, making it difficult to cope with the needs of multi-target dynamic tracking and comprehensive risk warning in complex scenarios. Currently, most existing monitoring and analysis solutions can only achieve local behavior detection in a single scenario, such as identifying single events like people gathering or items left behind in a specific area. They lack correlation analysis between different monitoring areas, making it difficult to capture abnormal behaviors across areas in a timely manner.

[0003] In practical applications, airport monitoring networks typically consist of hundreds or even thousands of surveillance cameras. These cameras have different monitoring perspectives and their monitoring ranges are independent, creating multiple information silos. When abnormal behavior involves multiple monitoring areas, existing systems cannot effectively integrate cross-perspective behavioral trajectory data, making it difficult to construct a comprehensive behavioral map from macro to micro levels. Furthermore, traditional anomaly detection methods often rely on preset thresholds, such as setting fixed parameters like dwell time and movement speed to identify anomalies. This approach is poorly adaptable to complex and changing behavioral patterns, easily leading to missed or false detections.

[0004] Existing technologies lack semantic descriptive capabilities in risk level assessment, often only outputting simple alarm signals and failing to provide managers with detailed information such as risk categories, levels, and associated areas, resulting in low emergency response efficiency. Regarding situation visualization, existing systems mostly use single video feeds or simple charts, which are insufficient to intuitively present the risk distribution across the entire airport area, hindering managers from comprehensively understanding the security situation and making accurate decisions. Summary of the Invention

[0005] The purpose of this invention is to provide a real-time analysis system for airport video data to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a real-time analysis system for airport video data, the system comprising:

[0007] The video stream spatiotemporal feature modeling module is used to receive airport surveillance video stream data, perform multi-scale spatiotemporal feature analysis on continuous video frames, extract the trajectory segments of moving targets and the optical flow distribution of static background areas, identify the occurrence time sequence and spatial coordinates of key behavioral events, and generate a spatiotemporal behavioral feature sequence.

[0008] The behavior trajectory map construction module, based on the spatiotemporal behavior feature sequence, performs hierarchical behavior modeling of the target motion trajectory and regional optical flow changes, establishes a topological connection path from macro region to micro target, collects the spatial correlation strength and temporal continuity attributes between path nodes, and outputs a multi-dimensional behavior trajectory map structure.

[0009] The abnormal area association analysis module extracts the set of behavioral path nodes from different monitoring perspectives based on the multi-dimensional behavioral trajectory map structure, calculates the overlap of behavioral patterns of cross-regional nodes, filters the coordinates of frequently occurring endpoint areas, and forms an abnormal behavior area association set.

[0010] The risk level semantic determination module obtains the predefined risk semantic labels for the corresponding regions based on the coordinates of the endpoint regions in the abnormal behavior region association set, sorts them by priority according to the frequency of regional risk association, matches the risk category of the endpoint region of the behavior trajectory, and generates a multi-level risk semantic label group.

[0011] The situational structure visualization module maps the physical monitoring area nodes corresponding to each tag according to the multi-level risk semantic tag group, divides the behavioral trajectory data into area nodes according to risk level, constructs the spatiotemporal mapping relationship between area nodes and behavioral trajectories, and generates a heat map of the risk situation across the entire airport.

[0012] Preferably, the spatiotemporal behavioral feature sequence includes a set of moving target trajectory segments, a regional optical flow intensity distribution matrix, and a spatiotemporal coordinate index of behavioral events; the multi-dimensional behavioral trajectory map structure includes a hierarchical behavioral node mapping relationship, a spatiotemporal topological connection chain, and a set of node association strength parameters; the abnormal behavior region association set includes a set of cross-regional overlapping nodes, a frequency distribution of endpoint region coordinates, and association region filtering results; the multi-level risk semantic label group includes a risk label frequency sorting table, a endpoint region risk label mapping relationship, and risk level matching results; and the airport global risk situation heat map includes regional node spatial identifiers, spatiotemporal grouping results of behavioral trajectories, and a risk situation mapping structure.

[0013] Preferably, the video stream spatiotemporal feature modeling module includes:

[0014] The multi-scale feature parsing submodule acquires airport surveillance video stream data, uses a 3D convolutional network to decompose the spatiotemporal features of continuous video frames, separates the trajectory vector of moving targets from the background light flow field distribution matrix, records the starting frame number and spatial bounding box coordinates of each behavioral event on the time axis, calculates the ratio of the first trigger position of key behaviors in the spatiotemporal sequence to the number of continuous frames, and generates a spatiotemporal distribution index of behavioral events.

[0015] The dynamic segment reconstruction submodule extracts video frame sequences before and after key behavior triggers based on the spatiotemporal distribution index of the behavior events, performs spatiotemporal alignment operations on motion trajectory vectors and optical flow matrices according to behavior types, and integrates multiple event segments under the same behavior category to form a set of spatiotemporal feature segments.

[0016] The feature sequence generation submodule, based on the set of spatiotemporal feature segments, counts the trigger frequency of various behavioral events within a unit time window, rearranges the feature segment set according to the order of behavioral trigger time, splices multiple event segments of the same category into a continuous sequence according to the first trigger position, uses a dynamic time warping algorithm to calibrate the time offset of the behavioral feature sequence, and outputs a standardized spatiotemporal behavioral feature sequence.

[0017] Preferably, the behavior trajectory mapping module includes:

[0018] The behavior hierarchy mapping submodule loads predefined airport area spatial topology weights based on the standardized spatiotemporal behavior feature sequence, sorts the motion trajectory nodes and optical flow area nodes by hierarchical priority, and establishes a node rearrangement index from the top area of ​​the terminal building to the last level area of ​​the boarding gate.

[0019] The topology path generation submodule extracts the spatiotemporal correlation strength parameters between adjacent level nodes based on the node reordering index, records the bidirectional connection weight and temporal continuity index of each node, and uses a graph embedding algorithm to calculate the spatial density and temporal consistency coefficient of the node path to generate a spatiotemporal topology path with weight attributes.

[0020] The graph structure extraction submodule, based on the spatiotemporal topological path with weighted attributes, collects the spatial coordinate connection relationship and temporal continuity parameters of all nodes in the path, and constructs a multi-dimensional behavioral trajectory graph structure containing a node spatial location mapping table and a temporal correlation matrix.

[0021] Preferably, the abnormal region correlation analysis module includes:

[0022] The multi-view node acquisition submodule extracts the set of behavior path terminal nodes in the coverage area of ​​different cameras based on the multi-dimensional behavior trajectory map structure, and records the monitoring view number and path duration parameters corresponding to each terminal node.

[0023] The cross-regional overlap analysis submodule, based on the set of terminal nodes, calls the spatial coordinate sequence of nodes from any two monitoring perspectives, calculates the overlap degree of node distribution in the spatial grid, extracts the frequently occurring endpoint coordinate points in the overlapping area, counts the frequency difference of such coordinate points in the multi-view path, and uses a spatiotemporal density clustering algorithm to filter the abnormal area coordinate set whose spatial clustering exceeds a preset threshold.

[0024] The association set generation submodule, based on the abnormal area coordinate set, traces back the node number of the corresponding coordinate in the original behavior path, integrates the spatial mapping relationship between the monitoring view number and the area coordinate, and outputs an abnormal behavior area association set containing the area coordinate frequency distribution map.

[0025] Preferably, the risk level semantic determination module includes:

[0026] The regional semantic mapping submodule, based on the coordinates of the endpoint region in the abnormal behavior region association set, queries the predefined risk semantic label library in the airport electronic fence database to establish an index mapping table between regional coordinates and risk labels;

[0027] The risk frequency statistics submodule calculates the associated frequency value of each type of risk label in the abnormal area coordinate set according to the index mapping table, and generates a risk label priority sequence by sorting the frequency values ​​from high to low.

[0028] The risk level matching submodule, based on the risk tag priority sequence, matches the risk tag category with the highest associated frequency for each abnormal area coordinate, integrates the tag matching results of all area coordinates, and generates a multi-level risk semantic tag group containing risk level weight values.

[0029] Preferably, the situation structure visualization module includes:

[0030] The physical area mapping submodule, based on the multi-level risk semantic tag group, parses the airport physical area node number corresponding to each risk tag, and records the spatial coordinate range of the node in the electronic map and the number of existing risk trajectories.

[0031] The trajectory space partitioning submodule, based on the node number of the airport physical area, groups and maps the spatiotemporal data of behavioral trajectories to the corresponding geographical areas according to the risk level weight, and establishes a two-way spatial index structure between trajectory data and geographical areas.

[0032] The heatmap generation submodule, based on the bidirectional spatial index structure, calculates the risk trajectory density value within each geographical area, maps the density value to color gradient parameters, and overlays the airport electronic map to generate a visual heatmap layer containing the real-time risk intensity distribution.

[0033] Preferably, the system further includes a real-time early warning feedback module, which performs the following operations based on the airport's overall risk situation heat map:

[0034] Extract the set of coordinates of physical regions in the heat map whose risk intensity exceeds a preset threshold;

[0035] Based on the multi-level risk semantic tag group associated with the physical area coordinate set, the priority of risk types is determined;

[0036] Generate an early warning instruction data package containing the risk coordinates, risk type, and recommended handling measures;

[0037] The warning instruction data packet is pushed to the responsible terminal equipment in the corresponding area in real time.

[0038] Preferably, the real-time early warning feedback module further includes:

[0039] The early warning verification submodule receives the handling result data fed back by the responsible terminal;

[0040] Compare the original risk intensity value of the area to be verified with the real-time risk intensity value after treatment;

[0041] When the rate of decline in risk intensity fails to reach the preset target, the warning level will be automatically upgraded and a secondary response instruction will be triggered.

[0042] Preferably, the system further includes an adaptive feature update module for performing:

[0043] The rate of change of regional risk intensity in the heat map of the airport's overall risk situation is periodically collected.

[0044] When the rate of change of risk intensity in a specific physical area continuously exceeds the dynamic threshold, a video stream feature re-extraction instruction for the corresponding area is triggered.

[0045] Based on the re-extracted spatiotemporal behavioral feature sequence, the node association strength parameters in the multi-dimensional behavioral trajectory graph structure are reconstructed;

[0046] Based on the reconstructed node association strength parameters, update the cross-regional behavior pattern overlap calculation model in the abnormal region association analysis module;

[0047] The updated behavior pattern overlap calculation model is output to the risk level semantic determination module, and the generation logic of risk label priority sequence is optimized simultaneously.

[0048] Compared with the prior art, the beneficial effects of the present invention are:

[0049] By using the video stream spatiotemporal feature modeling module to perform multi-scale spatiotemporal feature analysis on continuous video frames, it is possible to accurately extract the trajectory segments of moving targets and the optical flow distribution of static background areas. This provides detailed and comprehensive basic data for subsequent behavior analysis, making the identification of key behavioral events more accurate and helping to capture subtle behavioral changes that are easily overlooked in traditional monitoring.

[0050] The behavior trajectory mapping module is based on spatiotemporal behavior feature sequences to perform hierarchical behavior modeling and establish topological connection paths from macro areas to micro targets. This hierarchical modeling method can associate the dispersed target motion trajectories with regional optical flow changes to form a complete behavior trajectory mapping structure, allowing managers to clearly understand the motion paths of different targets in the airport and the relationships between them, which facilitates a holistic grasp of the behavior dynamics within the airport.

[0051] The abnormal area correlation analysis module extracts the set of cross-regional behavior path nodes and calculates the overlap degree based on the multi-dimensional behavior trajectory map structure, and filters high-frequency endpoint areas. It can effectively discover abnormal behavior areas that are related from different monitoring perspectives, breaking the isolation of information in different areas in traditional monitoring, making cross-regional abnormal behavior tracking more efficient, and helping to discover potential security risks in a timely manner.

[0052] The risk level semantic determination module obtains predefined risk semantic tags based on the abnormal behavior area association set, prioritizes them, and generates multi-level tag groups by matching risk categories. This makes risk determination no longer a simple alarm, but can clearly define the category and level of risk, allowing managers to take corresponding countermeasures according to different risk levels and categories, thus enhancing the pertinence and effectiveness of risk handling.

[0053] The situational structure visualization module maps multi-level risk semantic tag groups to physical monitoring area nodes, constructs a spatiotemporal mapping relationship, and generates a global risk situational heat map. This visualization method can intuitively display complex risk information, allowing managers to quickly grasp the risk distribution across the entire airport, understand the risk levels of different areas and related behavioral trajectories, thereby allocating resources more efficiently and improving the overall effectiveness of airport safety management. Attached Figure Description

[0054] Figure 1 This is a schematic diagram illustrating the working principle of the airport video data real-time analysis system described in this invention.

[0055] Figure 2 Flowchart for the video stream spatiotemporal feature modeling module;

[0056] Figure 3 Flowchart for the module for constructing behavioral trajectory maps;

[0057] Figure 4 The flowchart for the risk level semantic determination module; Detailed Implementation

[0058] 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, and 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.

[0059] Please see Figure 1 This invention provides a real-time analysis system for airport video data, the system comprising:

[0060] The video stream spatiotemporal feature modeling module receives airport surveillance video stream data, performs multi-scale spatiotemporal feature analysis on continuous video frames, extracts the trajectory segments of moving targets and the optical flow distribution of static background areas, identifies the occurrence sequence and spatial coordinates of key behavioral events, and generates a spatiotemporal behavioral feature sequence.

[0061] Based on this sequence, the behavior trajectory map construction module performs hierarchical behavior modeling on the target's motion trajectory and regional optical flow changes, establishes a topological connection path from the macro region to the micro target, collects the spatial correlation strength and temporal continuity attributes between path nodes, and outputs a multi-dimensional behavior trajectory map structure.

[0062] Based on the map structure, the abnormal area association analysis module extracts the set of behavioral path nodes from different monitoring perspectives, calculates the overlap of behavioral patterns of cross-regional nodes, filters the coordinates of frequently occurring endpoint areas, and forms an abnormal behavior area association set.

[0063] The risk level semantic determination module obtains the predefined risk semantic labels for the corresponding regions based on the coordinates of the endpoint regions in the set, sorts them by priority according to the frequency of risk association in the regions, matches the risk category of the endpoint region of the behavior trajectory, and generates a multi-level risk semantic label group.

[0064] The situational structure visualization module maps the physical monitoring area nodes corresponding to each tag group, divides the behavioral trajectory data into area nodes according to risk level, constructs the spatiotemporal mapping relationship between area nodes and behavioral trajectories, and generates a heat map of the risk situation across the entire airport.

[0065] Example 1: See Figure 2 The video stream spatiotemporal feature modeling module of the airport video data real-time analysis system is responsible for performing multi-scale spatiotemporal feature analysis on the input surveillance video stream, generating standardized spatiotemporal behavioral feature sequences. This module consists of a multi-scale feature analysis submodule, a dynamic segment reconstruction submodule, and a feature sequence generation submodule. These submodules work together to complete the transformation from raw video data to structured feature sequences.

[0066] The multi-scale feature analysis submodule receives real-time video stream data transmitted from airport surveillance cameras and uses a 3D convolutional network to perform spatiotemporal feature decomposition on continuous video frames. This network analyzes inter-frame motion changes in the temporal dimension and extracts target contours and background optical flow distribution in the spatial dimension. The trajectory information of moving targets is quantized into vector form, recording the displacement direction and velocity changes of the target in continuous frames. The background optical flow field distribution is stored in matrix form, reflecting the macroscopic motion trend of static areas. Each detected behavioral event is labeled with its starting frame number and spatial bounding box coordinates; the bounding box is used to locate the physical area where the behavior occurred. The spatiotemporal distribution of key behaviors is quantified by calculating the ratio of the initial trigger position to the number of consecutive frames, generating a spatiotemporal distribution index of behavioral events. This index, based on the time axis, records the triggering sequence and spatial coverage of different behavior categories.

[0067] The dynamic segment reconstruction submodule extracts video frame sequences within a specific time window before and after the triggering of key behaviors based on the spatiotemporal distribution index of behavioral events. The extracted segments must contain the complete cycle of the behavior, preserving the entire process from initial triggering to termination. The trajectory vector of the moving target and the background optical flow matrix are aligned in the spatiotemporal dimension to eliminate spatial deviations caused by different camera perspectives. Multiple event segments of the same behavior category are integrated into a unified feature representation, forming a spatiotemporal feature segment set. This set is stored categorized by behavior type, with segments within each category having a consistent spatiotemporal feature dimension.

[0068] The feature sequence generation submodule performs statistical analysis on the spatiotemporal feature segment set, calculating the trigger frequency of various behavioral events within a unit time window. Behavioral events are rearranged according to their trigger time order to ensure continuity on the timeline. Multiple event segments of the same category are spliced ​​together based on their first trigger position to form a continuous behavioral feature sequence. A dynamic time warping algorithm is used to calibrate the temporal offset between different instances, eliminating time differences caused by detection delays or video stream jitter. The calibrated feature sequence has a unified time reference, ultimately outputting a standardized spatiotemporal behavioral feature sequence.

[0069] The spatiotemporal behavior feature sequence comprises three main components. The set of moving target trajectory segments records the motion paths of all detected targets, stored as a vector sequence, with each vector containing the target's position and motion state within a single frame. The regional optical flow intensity distribution matrix stores the macroscopic motion patterns of the background region; the matrix's rows and columns correspond to the spatial grid division of the video frames, and the matrix element values ​​represent the optical flow intensity of the corresponding grid. The behavior event spatiotemporal coordinate index is organized in a tree structure, with the root node recording the behavior category and child nodes storing the starting frame number and bounding box coordinates of all instances within that category.

[0070] The multi-dimensional behavioral trajectory mapping structure consists of hierarchical behavioral node mapping relationships, spatiotemporal topological connection chains, and a set of node association strength parameters. The hierarchical behavioral node mapping relationships are organized in a tree structure, with top-level nodes corresponding to major airport functional areas, middle-level nodes representing specific functional zones, and bottom-level nodes associated with specific monitoring points. The spatiotemporal topological connection chains record the transfer paths between nodes, with each chain storing the source node, target node, transfer time, and spatial distance. The node association strength parameter set includes spatial density coefficients and temporal continuity indicators, quantifying the degree of association between nodes.

[0071] The anomalous behavior region association set is generated through cross-regional node analysis. The cross-regional overlapping node set stores nodes with the same physical location detected from different camera perspectives, with each node labeled with its corresponding camera number and occurrence time. The endpoint region coordinate frequency distribution statistics analyze the termination positions of all behavior paths, recording the number of times each position appears in different time periods. The association region filtering results are generated using a spatial clustering algorithm, labeling the coordinates of frequently clustered anomalous regions and their associated behavior categories.

[0072] The multi-level risk semantic label group contains three types of structured data. The risk label frequency sorting table arranges all detected risk types in descending order of frequency of occurrence, with each type indicating the number of times it was triggered within the analysis period. The endpoint region risk label mapping relationship is stored using a hash table, where the key is the region coordinates and the values ​​are all risk labels associated with those coordinates and their frequencies. The risk level matching results record the final risk category determined for each abnormal region, including the risk level weight value and the basis for the determination.

[0073] The airport's overall risk situation heat map consists of three parts. The spatial identifiers for regional nodes use Geographic Information System (GIS) coordinates to accurately mark the physical boundaries of each monitored area. The spatiotemporal grouping results of behavioral trajectories are stored according to risk level, with each group of trajectory data labeled with its corresponding risk category and time period. The risk situation mapping structure is stored using a graph database, where nodes represent physical areas, edges represent risk propagation paths, and edge weights reflect the intensity of risk transmission.

[0074] The video stream spatiotemporal feature modeling module operates without manual intervention; all processing steps are executed automatically. The parameters of the 3D convolutional network are pre-trained using historical monitoring data, enabling it to adapt to environmental differences at various airports. The behavior event detection algorithm supports online updates, dynamically loading new detection models when new behavior categories are added. Spatiotemporal alignment is implemented using a feature point matching algorithm, exhibiting robustness to changes in camera viewpoint. The sliding window size of the dynamic time warping algorithm adaptively adjusts according to the video frame rate, ensuring temporal calibration accuracy at different sampling rates. The output format of the standardized feature sequence is compatible with subsequent modules and can be directly used for behavior trajectory map construction.

[0075] In the multi-scale feature parsing process, a lightweight neural network is used for moving target detection to meet real-time requirements while ensuring accuracy. A sparse optical flow algorithm is used for background optical flow calculation, focusing on monitoring light variation patterns in densely populated areas. The spatiotemporal distribution index of behavioral events uses time-encoding technology for compressed storage, reducing memory usage. The behavior classifier in the dynamic fragment reconstruction stage supports incremental learning, allowing for gradual improvement of the behavior category system. Frequency statistics during feature sequence generation employ a sliding time window, with the window size dynamically adjusted according to the airport's operational status.

[0076] The storage of spatiotemporal behavioral feature sequences employs a columnar database, optimizing the read and write efficiency of large-scale trajectory data. The regional optical flow matrix utilizes sparse matrix compression technology, storing only non-zero elements and their coordinates. A behavioral event index tree enables fast range queries, supporting event retrieval based on time and spatial conditions. The construction of the multi-dimensional behavioral trajectory map employs a distributed computing framework, supporting parallel processing of massive nodes and edges. The anomaly region detection algorithm introduces spatial topological constraints to avoid misclassifying normal transit areas as anomalies. Risk level matching fuses multi-source data, including historical event records and real-time sensor readings. Heatmap generation utilizes GPU-accelerated rendering to ensure real-time visualization of large-scale scenes.

[0077] The hardware implementation of this module is based on a heterogeneous computing architecture, with the CPU handling logic control and data scheduling, and the GPU accelerating deep learning inference and image processing. An in-memory database is used to cache intermediate results, reducing disk I / O latency. Network communication employs zero-copy technology to optimize the transmission efficiency of video stream data. Sub-modules within the module communicate asynchronously via message queues, achieving loose coupling and high concurrency processing. An error handling mechanism monitors the operational status of each stage and automatically initiates a recovery process in case of anomalies. A performance monitoring system collects processing latency and resource utilization in real time, providing data support for system optimization.

[0078] The design of the video stream spatiotemporal feature modeling module takes into account the specific needs of airport monitoring scenarios. Changes in lighting conditions are automatically corrected using a white balance algorithm to avoid the impact of day-night cycles on feature extraction. Target occlusion issues in densely populated scenes are mitigated through a multi-target tracking algorithm, maintaining trajectory continuity. Camera shake interference is eliminated using electronic image stabilization technology, ensuring the accuracy of motion detection. Compatibility with video sources of different resolutions is achieved through an adaptive scaling mechanism, uniformly processing high-definition and standard-definition video streams. The operational characteristics of multi-timezone airports are addressed through UTC time standardization, unifying the time base of all data.

[0079] Example 2: See Figure 3The airport video data real-time analysis system's behavior trajectory mapping module and abnormal area correlation analysis module work together to transform standardized spatiotemporal behavioral feature sequences into structured behavior trajectory maps, and identify abnormal area correlation sets from them. These two modules, through hierarchical modeling and cross-perspective analysis, achieve accurate characterization of personnel flow patterns and potential risk areas within the airport.

[0080] The behavior trajectory mapping module first processes the input spatiotemporal behavior feature sequence through the behavior hierarchy mapping submodule. This submodule loads predefined airport spatial topology weight data, which reflects the importance and connectivity of various functional areas within the terminal. The trajectory nodes of moving targets are categorized according to their physical location, while background optical flow area nodes are divided based on the monitoring field of view. The node hierarchy is constructed from macro to micro levels. Top-level nodes correspond to the overall terminal zoning, such as the departure hall, security checkpoint, and waiting area; mid-level nodes are refined to specific functional areas, such as check-in counters and baggage carousels; and bottom-level nodes are precise down to the microscopic area covered by a single camera. Each node is assigned a unique hierarchical index, forming a complete spatial topology structure.

[0081] The topology path generation submodule performs analysis based on hierarchical indexing, extracting the transition relationships between adjacent hierarchical nodes. For each pair of nodes with spatiotemporal association, its bidirectional connection weight is recorded, which integrates factors such as spatial distance, transition frequency, and time interval. The temporal continuity index is calculated by analyzing the temporal regularity of transitions between nodes in historical data, reflecting the stability of the path. The graph embedding algorithm projects nodes into a low-dimensional vector space, capturing temporal evolution patterns while preserving spatial proximity; the generated vectors are used to calculate the density score of the node paths. The spatiotemporal topology path is ultimately stored in the form of a weighted directed graph, with edge weights containing both spatial association strength and temporal continuity coefficients.

[0082] The graph structure extraction submodule integrates the generated topological paths to construct a multi-dimensional behavioral trajectory graph. A node spatial location mapping table records the precise coordinates of all nodes in the airport electronic map and their respective levels. The temporal correlation matrix uses a sparse matrix format to store the temporal relationships between nodes, with matrix element values ​​representing the statistical characteristics of the transfer time intervals. The complete graph structure supports multi-dimensional queries, enabling analysis of inter-regional pedestrian flow trends at a macro level and tracking of the movement paths of specific targets at a micro level. The graph update mechanism employs incremental processing, with newly arrived spatiotemporal behavioral features being integrated into the existing structure in real time.

[0083] The anomaly area correlation analysis module operates based on a multi-dimensional behavioral trajectory map. The multi-view node acquisition submodule first extracts the behavioral path terminal nodes from different camera perspectives; these nodes represent the target's final location within the monitored field of view. Each terminal node is associated with its corresponding camera number, arrival timestamp, and duration of stay. The node data undergoes normalization processing to eliminate the impact of time synchronization errors between different cameras. The cross-regional overlap analysis submodule performs spatial matching on the multi-view terminal nodes, transforming the coordinate system of each camera to the airport's global coordinate system.

[0084] The spatial grid partitioning strategy divides the airport plane into uniform cells and counts the frequency of node occurrences from different viewpoints within each cell. A distribution overlap algorithm calculates the similarity of node distributions between any two viewpoints within the same grid, identifying high-frequency regions of common interest to multiple viewpoints. A spatiotemporal density clustering algorithm performs secondary analysis on these regions, considering the temporal distribution characteristics of node occurrences and selecting coordinate points that exhibit clustering patterns in both time and space. Continuous verification is introduced during the generation of the anomaly region coordinate set; only regions that repeatedly appear within consecutive time windows are retained.

[0085] The association set generation submodule performs backtracking analysis on the selected abnormal regions, finding their corresponding nodes in the original behavioral trajectory map. Each abnormal region is associated with complete information about its source path, including the sequence of nodes traversed, the transfer time series, and a list of associated cameras. The frequency distribution map of regional coordinates is visualized using a heatmap format, intuitively showing the degree of abnormality in different regions. The output of the abnormal behavior region association set adopts a hierarchical storage structure: the top layer records the spatial distribution characteristics of the regions, the middle layer stores the path backtracking information, and the bottom layer retains the original node data.

[0086] The behavioral trajectory graph construction module is implemented using a distributed graph computing framework, supporting parallel processing of large-scale nodes and edges. Spatial indexing is introduced to accelerate the hierarchical index construction process, and an R-tree structure is used to optimize regional query efficiency. Negative sampling is employed during graph embedding training to reduce computational complexity without sacrificing accuracy. The weight calculation for spatiotemporal topological paths uses a sliding time window, with the window size dynamically adjusted based on airport operating hours. The incremental update mechanism is implemented through a difference detection algorithm, recalculating only the local graph affected by new data.

[0087] The anomaly region correlation analysis module employs a time alignment algorithm for multi-view data processing to compensate for time deviations caused by network transmission delays. The granularity of the spatial grid is automatically adjusted based on airport passenger flow, with finer grids used during peak hours to improve positioning accuracy. The parameters of the density clustering algorithm are learned from historical data to adapt to the spatial distribution characteristics of different areas. The anomaly region verification phase introduces behavioral pattern comparison, performing similarity analysis between the current clustering pattern and historical normal patterns. The storage of correlation sets uses a columnar compression format to optimize the storage and retrieval efficiency of high-frequency coordinate points.

[0088] The error handling and recovery mechanisms are comprehensive. Input data validation uses pattern matching, and illegal formats are immediately isolated and processed. Critical states during processing are periodically persisted, and recovery resumes from the most recent consistent state after a failure. Automatic degradation is triggered when resources exceed limits, prioritizing the operation of core functions. Network partition detection is implemented through a heartbeat mechanism, switching to local caching mode in case of anomalies. Security audit logs record all data access traces, supporting post-event traceability and analysis.

[0089] The module's extensibility design supports future changes in requirements. The behavior graph model reserves attribute extension interfaces, allowing the addition of new dimensions without affecting existing logic. The analysis algorithm's plug-in design supports dynamically loading different implementations. The storage backend is abstracted into a unified interface, adaptable to different database technologies. The computing resource pool supports horizontal scaling, adding processing nodes as needed. Protocol version compatibility is guaranteed through an adaptation layer, allowing new and old systems to coexist.

[0090] Example 3: See Figure 4 The risk level semantic judgment module and the situation structure visualization module of the airport video data real-time analysis system work together to transform the set of abnormal behavior area associations into multi-level risk semantic label groups, and finally generate a heat map of the risk situation distribution across the entire airport. These two modules, through semantic mapping and spatial visualization technologies, achieve a quantitative assessment and intuitive presentation of airport security risks.

[0091] The initial processing of the risk level semantic determination module is completed by the regional semantic mapping submodule. This submodule connects to the airport's electronic fence database, which stores predefined risk attributes for all functional areas of the terminal. Each physical area is recorded in the database with a unique geocode, associated with three types of basic information: area function type (e.g., security checkpoint, baggage claim area), historical risk event statistics, and preset risk semantic tags. The tagging system adopts a hierarchical classification method, with the top level consisting of three risk levels: major, medium, and general. The lower level is further subdivided into 12 specific risk categories, including crowd gathering, abnormal detention, and reverse flow. The mapping process uses a bidirectional index structure; the forward index quickly finds tags through geocodes, while the reverse index supports searching related areas by tag.

[0092] The risk frequency statistics submodule performs multi-dimensional analysis on the coordinate set of abnormal areas. Each coordinate point carries timestamp information, and a time decay factor is introduced during statistics, with recent events having a higher weight than historical events. Frequency statistics adopt a sliding time window mechanism, with the window width dynamically adjusted according to the airport's operational status. The window is set to 30 minutes for normal periods and automatically shortened to 15 minutes for special periods (such as large-scale flight delays). During the generation of the risk label priority sequence, conflicts are resolved for coordinate points simultaneously associated with multiple labels, prioritizing labels with higher relevance to the area's function. The frequency statistics results are normalized using the following formula:

[0093]

[0094] in, Indicates risk label The normalized frequency probability, For the region The spatiotemporal weighting coefficients, It is an indicator function (when) and The value is 1 when associated, otherwise 0. This represents the total number of abnormal regions. Total number of risk label types. Spatiotemporal weighting coefficient. It is derived by weighting three factors: area, population density, and event duration.

[0095] The risk level matching submodule performs the final risk assessment. For each abnormal area coordinate, the top three most frequently associated risk tags are matched to form a candidate tag set. These candidate tags are filtered by the rule engine, eliminating options that clearly contradict the area's function (e.g., the "Check-in queue overload" tag will not appear in the baggage claim area). A confidence score is appended to the matching result, which combines frequency probability, area function fit, and historical verification accuracy. The output of the multi-level risk semantic tag group is organized in a tree structure, with the root node representing the risk level, child nodes representing specific risk types, and leaf nodes storing the list of associated area coordinates and their matching parameters.

[0096] The physical area mapping submodule of the situational structure visualization module parses multi-level risk semantic label groups, converting abstract labels into specific geographic elements. The airport electronic map adopts a layered vector data model, including basic geographic information such as building outline layers, facility location layers, and passageway layers. Each risk label is associated with a specific map element type; for example, the "personnel gathering" label maps to an open area polygon, and the "equipment anomaly" label is associated with a facility location. The regional node numbering system is consistent with the airport asset management system, ensuring the accuracy of cross-system data association. The number of existing risk trajectories is obtained through spatial connection queries, and the historical trajectory data volume within the range of each geographic element is statistically analyzed.

[0097] The trajectory spatial partitioning submodule establishes a bidirectional index between behavioral trajectories and geographical regions. Risk level weights are used as the partitioning basis, dividing trajectory data into three processing priority queues. The spatial index adopts an improved R* tree structure, adding a risk level dimension to the traditional spatial partitioning, forming a three-dimensional index space. The construction process of the bidirectional index uses a batch loading strategy, first coarsely partitioning by spatial range, and then finely partitioning within each block by risk level. The index update mechanism supports incremental maintenance; newly arrived trajectory data triggers local index reconstruction, avoiding the overhead of global reconstruction.

[0098] The core of the heatmap generation submodule is the calculation of the risk density field. The geographical region is discretized into 50cm × 50cm grid cells, and the risk density value of each cell is synthesized from three components: real-time trajectory density. Historical baseline density Risk level weight The composite calculation employs a non-linear overlay model to highlight the visual contrast of high-risk areas. Color mapping uses the HSL color space, with hue representing risk type (red for personnel-related risks, blue for equipment-related risks), saturation reflecting absolute density, and brightness adjustment automatically adapting to ambient lighting conditions. The overlay of the heatmap layer and electronic map uses alpha blending technology to preserve the legibility of the underlying map. The visualization system supports multi-level zooming, dynamically adjusting the rendering granularity and data aggregation level of the heatmap at different view levels.

[0099] The hardware deployment of the risk level semantic determination module adopts a highly available architecture. The database cluster uses a master-slave replication mode, with the electronic fence database deployed on in-memory computing nodes to achieve millisecond-level query response. The rule engine runs on a dedicated inference server, and the loaded risk determination rules support hot updates. The caching system adopts a layered design, with frequently accessed tag mapping tables residing in memory, and all data stored on an SSD array. Service quality control is implemented for computing resource allocation to ensure that the determination latency during peak periods does not exceed a set threshold.

[0100] The situational visualization module's rendering pipeline is optimized for large-scale geographic data. Vector map data utilizes GPU instantiation rendering technology, capable of rendering tens of thousands of geographic features in a single batch. Thermal field calculations are processed in parallel using computational shaders, with each grid cell allocated an independent thread. A dynamic load balancing mechanism monitors the rendering frame rate and automatically reduces the rendering precision of non-critical areas when performance degrades. A multi-view synchronization system maintains consistency in content displayed across different terminals, supporting collaborative viewing on command center screens and mobile devices.

[0101] Data exchange between modules adopts a standardized protocol. Risk label group transmission uses the Apache Avro binary format, containing complete pattern definitions. The geographic coordinate system adopts UTM projection on the WGS84 ellipsoid to ensure spatial analysis accuracy. Time series data is appended with NTP timestamps, and synchronization accuracy is controlled at the millisecond level. The message queue adopts a partitioned mode, dividing communication channels according to airport functional areas to reduce unnecessary network transmission.

[0102] Example 4: The real-time early warning feedback module of the airport video data real-time analysis system generates early warning instructions for areas exceeding risk thresholds based on the airport's overall risk situation heat map, and achieves closed-loop verification of the early warning effect. This module forms a complete process from risk detection to response verification through a multi-level early warning mechanism and a response feedback system.

[0103] Taking a typical international airport Terminal 2 as an example, when the system detects an abnormal risk cluster in the departure hall's Gate 3 area, the real-time early warning feedback module initiates the processing flow. The heat map data displays a real-time risk intensity value of 87 for this area (threshold set to 75), and the associated multi-level risk semantic tag group is marked as "Abnormal Crowd Cluster - Level 2 Risk". The system automatically extracts the physical coordinate information of this area and, combined with the spatial coding of the airport's electronic map, generates an early warning instruction data package containing the following elements:

[0104]

[0105] The early warning instruction data packet is pushed to three terminals via the airport's dedicated communication network: the mobile terminal of the duty supervisor in the ground services department, the large screen system in the terminal's central control room, and the emergency broadcasting equipment near Gate 3. The data packet is encapsulated in JSON format, containing a standardized field structure and metadata description to ensure correct parsing by different systems. Upon receiving the packet, the mobile terminal immediately triggers a vibration alarm, displays the warning details interface, and automatically retrieves real-time monitoring footage of the area. The large screen system in the central control room highlights the risk area on an electronic map, with the locations of security resources within a 500-meter radius simultaneously marked. The emergency broadcasting equipment plays pre-recorded evacuation guidance messages, with the volume automatically adjusted according to the ambient noise level.

[0106] The early warning verification submodule initiates the monitoring process after the instruction is issued. The system continuously collects feedback data from three dimensions: the start time and type of the handling action recorded in the responsible terminal's operation log; the video analysis subsystem monitors changes in personnel density in the target area; and the mobile terminal's GPS location confirms the arrival of personnel on site. If the risk intensity value does not decrease to the preset 15% target within 5 minutes of the early warning being issued, the system automatically upgrades the early warning level. Based on the original Level 2 risk, a supplementary instruction of "equipment inspection requirements" is added, notifying the electromechanical engineering department to check the operating status of the turnstiles in the area. The upgraded early warning data package adds related elements: historical data comparison results, equipment failure probability analysis, and cross-departmental collaborative handling guidelines.

[0107] The early warning feedback mechanism includes a multi-level confirmation process. Frontline staff report the initial handling situation via mobile devices, including written descriptions of the scene, photographic evidence, and preliminary judgments. Upon arrival, professional security personnel submit a secondary verification report, detailing the reasons for the abnormal gathering (such as queues caused by temporary check-in system malfunctions). The central control room operator synthesizes feedback from all parties and marks the event handling status in the system. All feedback information is stored in a structured manner, forming a complete early warning event handling file.

[0108] An anomaly handling mechanism ensures system robustness. When network interruption causes command transmission failure, the system automatically switches to a backup communication channel, prioritizing the delivery of warnings to high-risk areas. When mobile terminals are offline, warning commands are temporarily stored at the regional edge computing node and resent after the connection is restored. For false alarms, staff can report the cause of the false alarm via their terminals; the system records this feedback to optimize the risk assessment model. Areas with multiple false alarms will automatically have their sensitivity reduced, simultaneously triggering a recalibration of video analysis parameters.

[0109] The dynamic adjustment mechanism for early warning priorities optimizes resource allocation based on real-time conditions. During widespread flight delays, the system temporarily increases the risk monitoring frequency in check-in areas. During special events, the early warning response level for VIP channels is automatically upgraded. During low-passenger-flow periods at night, early warning thresholds for non-critical areas are appropriately relaxed. The dynamic adjustment parameters are derived from the airport's operations database and are synchronized in real-time with key information such as flight status and special event schedules.

[0110] Historical early warning data analysis supports continuous system optimization. Daily automatically generated early warning effectiveness reports include multiple indicators: average early warning response time, risk mitigation success rate, and multi-departmental collaboration efficiency. This data is presented in a dashboard format to help managers identify process bottlenecks. The system performs monthly backtesting tests on early warning scenarios, simulating typical risk events to verify the completeness of the handling procedures. Test results are used to update the emergency response knowledge base, supplementing it with new handling methods and contact information.

[0111] Integration with other airport security systems extends early warning functionality. When an alert involves suspicious items, historical images from security X-ray machines are automatically retrieved for comparison. Personnel identification alerts trigger facial recognition systems for focused monitoring. Fire risk alerts link with the fire control system, opening emergency exits in advance. The integration interface utilizes the airport's unified middleware platform, ensuring secure and reliable data exchange between systems.

[0112] The hardware deployment of the early warning feedback module takes into account actual operation and maintenance needs. Mobile terminals feature an industrial-grade drop-proof design, with battery life sufficient for 12 hours of continuous operation. The central control room server is equipped with redundant power supplies and network interfaces to ensure high availability. Edge computing nodes are deployed in the low-voltage electrical rooms throughout the terminal building to achieve localized data processing. All hardware devices are integrated into the airport asset management system and undergo regular preventative maintenance.

[0113] The user interface is designed according to ergonomic principles. The mobile terminal warning interface uses red, yellow, and blue color coding to distinguish risk levels, with key information displayed in larger font. The control room's large screen system supports gesture zooming, and important alarms automatically pop up in the foreground. Voice announcements have been acoustically optimized, remaining clearly readable even in noisy environments. The interface language supports switching between Chinese and English, catering to the needs of international airport staff.

[0114] An access control system ensures the secure operation of the early warning system. Different levels of staff are granted differentiated operating permissions: frontline staff can only confirm warnings and report handling status; supervisors can adjust warning levels; and system administrators can modify judgment parameters. All operation records are audited in detail and stored in conjunction with employee ID numbers. Sensitive operations require secondary authentication, and modifications to key parameters require review by multiple people.

[0115] The real-time early warning feedback module demonstrated stable performance during actual operation. The average latency from typical early warning generation to delivery to the terminal was controlled within 3 seconds, meeting real-time response requirements. The system supports processing over 200 early warning events per day, with concurrent processing during peak hours buffered via message queues. Network bandwidth usage was optimized, and the average size of early warning data packets was compressed to below 50KB. Mobile terminal signal coverage reached 99.7% across the entire terminal area, ensuring reliable delivery of early warning commands.

[0116] Example 5: The adaptive feature update module of the airport video data real-time analysis system continuously monitors changes in risk situation and dynamically adjusts feature extraction and behavior analysis strategies to ensure the system adapts to the dynamic changes in the airport operating environment. This module periodically assesses the trend of risk intensity changes, triggering feature re-extraction and model update processes to maintain the accuracy and timeliness of the analysis system.

[0117] After the module starts, a risk intensity baseline is established first. The system loads historical data from the past 30 days of airport-wide risk situation heat map, and statistically analyzes the distribution characteristics of daily risk intensity values ​​by region. For each functional area of ​​the terminal, the average and standard deviation of its risk intensity are calculated to form the initial setting of dynamic thresholds. Baseline data is stored in a time-series database, supporting efficient range queries and statistical analysis. The monitoring of the risk intensity change rate uses a 15-minute basic time window, collecting the latest heat values ​​of each area in real time within the window period and comparing them with the baseline data.

[0118] When the rate of change of risk intensity in a specific area continuously exceeds the dynamic threshold, the system initiates a feature re-extraction process. The rate of change assessment employs a sliding window mechanism; a valid trigger is only determined when the threshold is exceeded for three consecutive monitoring windows, avoiding misjudgments caused by instantaneous fluctuations. The trigger command includes a detailed description of the affected area: geographic boundary coordinates, a list of associated cameras, and an abnormal time period marker. Upon receiving the command, the video stream spatiotemporal feature modeling module prioritizes retrieving the original video data for the abnormal time period from the cache; if the cache has expired, it retrieves it again from the storage system. The feature extraction process uses an incremental update mode, parsing only the video stream within the abnormal area while retaining existing analysis results for other areas.

[0119] The reconstruction process of spatiotemporal behavioral feature sequences incorporates a change detection algorithm. The system compares the newly extracted feature sequences with historical normal patterns to identify deviations in trajectory features and optical flow distribution. For moving target trajectories, it analyzes dimensions such as velocity change patterns, path deviation, and dwell time distribution; for background optical flow fields, it detects indicators such as intensity distribution changes, directional consistency, and regional correlation. Deviation features are marked as key change points for subsequent map structure updates.

[0120] The node association strength parameters of the multi-dimensional behavioral trajectory map are updated using a gradual adjustment strategy. The system maintains two sets of association parameters: long-term stable parameters reflecting the inherent connectivity characteristics between regions, and short-term dynamic parameters capturing changes under the current operational status. The update process first adjusts the short-term dynamic parameters, and only after the changes persist for more than a preset period are they gradually integrated into the long-term parameters. When recalculating the spatiotemporal association strength between nodes, factors such as the frequency and duration of path occurrences and node transfer speeds during abnormal periods are considered. Graph embedding is used for synchronous updates, fine-tuning point positions in a low-dimensional vector space to reflect the latest behavioral pattern changes.

[0121] The cross-regional behavior pattern overlap calculation model of the anomaly region association analysis module is updated accordingly. The system redefines the spatial correlation metric between regions based on the reconstructed node association strength. A time decay factor is introduced into the overlap calculation, meaning recent behavior patterns have a greater impact on the results than historical data. The model update employs an online learning mechanism, continuously adjusting the classification boundaries based on newly arriving anomaly region data to adapt to the evolution of risk patterns. The calculation process distinguishes between inherent regional characteristics and temporary changes to avoid over-adapting to short-term anomalies.

[0122] The risk level semantic determination module receives updated behavioral pattern information and optimizes its label priority generation logic. The system reassesses the matching degree between each risk label and regional features, and adjusts the weight allocation of label frequency statistics. The semantic mapping rules are dynamically expanded, and when new abnormal behaviors are detected, temporary risk labels are automatically created for manual review. The parameters of the determination engine are adaptively calibrated, and differentiated determination strategies are implemented for regions with significant changes while maintaining the stability of the overall determination framework.

[0123] The module's resource management employs an intelligent scheduling strategy. Computing resources are prioritized for update tasks in areas of increasing risk, ensuring real-time performance in critical areas. Memory cache dynamically adjusts the resident data based on the frequency of area changes, with feature sequences from frequently updated areas remaining permanently in memory. Network bandwidth allocation implements quality of service control to ensure the transmission stability of video stream re-extraction tasks. The storage system adopts a hot and cold data tiering mechanism, storing data from recently active areas in a high-speed storage layer.

[0124] All update operations are logged in detail, including metadata such as triggering conditions, processing time, and resource consumption.

[0125] The unique scenarios of airport operations are addressed specifically in the module design. During seasonal flight changes, the sensitivity of change detection is automatically relaxed to adapt to changes in passenger flow patterns brought about by new timetables. During major events, the monitoring frequency of VIP channels is temporarily increased to ensure rapid adaptation in VIP areas. Under extreme weather conditions, the judgment threshold for outdoor areas is adjusted to consider the impact of weather on pedestrian behavior. Update tasks in non-operational areas are shut down during nighttime maintenance periods to save computing resources.

[0126] Module maintainability is improved through standardization. Configuration parameters are managed declaratively, and change history is tracked using version control tools. Monitoring metrics cover all stages of the update process, including trigger accuracy, feature extraction latency, and model convergence status. Diagnostic tools are integrated into the operations and maintenance platform, supporting root cause analysis in case of update anomalies. Automated testing verifies core update logic, ensuring that code modifications do not affect basic functionality.

[0127] The adaptive feature update module endows the analysis system with the ability to evolve dynamically, freeing it from the limitations of initial training data. By continuously sensing environmental changes and adjusting itself, the system can adapt to diverse scenarios such as seasonal fluctuations in airport passenger flow patterns, adjustments to operational strategies, and unexpected events. This self-updating characteristic significantly extends the system's effective service life, reduces manual maintenance costs, and provides sustained and reliable technical support for airport safety management. The module's operation forms a closed loop from data to knowledge to optimization, driving the entire analysis system to continuously evolve towards higher accuracy and greater adaptability.

[0128] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0129] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A real-time analysis system for airport video data, characterized in that, include: The video stream spatiotemporal feature modeling module is used to receive airport surveillance video stream data, perform multi-scale spatiotemporal feature analysis on continuous video frames, extract the trajectory segments of moving targets and the optical flow distribution of static background areas, identify the occurrence time sequence and spatial coordinates of key behavioral events, and generate a spatiotemporal behavioral feature sequence. The behavior trajectory map construction module, based on the spatiotemporal behavior feature sequence, performs hierarchical behavior modeling of the target motion trajectory and regional optical flow changes, establishes a topological connection path from macro region to micro target, collects the spatial correlation strength and temporal continuity attributes between path nodes, and outputs a multi-dimensional behavior trajectory map structure. The abnormal area association analysis module extracts the set of behavioral path nodes from different monitoring perspectives based on the multi-dimensional behavioral trajectory map structure, calculates the overlap of behavioral patterns of cross-regional nodes, filters the coordinates of frequently occurring endpoint areas, and forms an abnormal behavior area association set. The risk level semantic determination module obtains the predefined risk semantic labels for the corresponding regions based on the coordinates of the endpoint regions in the abnormal behavior region association set, sorts them by priority according to the frequency of regional risk association, matches the risk category of the endpoint region of the behavior trajectory, and generates a multi-level risk semantic label group. The situational structure visualization module maps the physical monitoring area nodes corresponding to each tag according to the multi-level risk semantic tag group, divides the behavioral trajectory data into area nodes according to risk level, constructs the spatiotemporal mapping relationship between area nodes and behavioral trajectories, and generates a heat map of the risk situation of the entire airport. It also includes an adaptive feature update module for performing: The rate of change of regional risk intensity in the heat map of the airport's overall risk situation is periodically collected. When the rate of change of risk intensity in a specific physical area continuously exceeds the dynamic threshold, a video stream feature re-extraction instruction for the corresponding area is triggered. Based on the re-extracted spatiotemporal behavioral feature sequence, the node association strength parameters in the multi-dimensional behavioral trajectory graph structure are reconstructed; Based on the reconstructed node association strength parameters, update the cross-regional behavior pattern overlap calculation model in the abnormal region association analysis module; The updated behavior pattern overlap calculation model is output to the risk level semantic determination module, and the generation logic of risk label priority sequence is optimized simultaneously.

2. The airport video data real-time analysis system according to claim 1, characterized in that, The spatiotemporal behavioral feature sequence includes a set of moving target trajectory segments, a regional optical flow intensity distribution matrix, and a spatiotemporal coordinate index of behavioral events. The multi-dimensional behavioral trajectory map structure includes a hierarchical behavioral node mapping relationship, a spatiotemporal topological connection chain, and a set of node association strength parameters. The abnormal behavior region association set includes a set of cross-regional overlapping nodes, a frequency distribution of endpoint region coordinates, and the filtering results of associated regions. The multi-level risk semantic label group includes a risk label frequency sorting table, a endpoint region risk label mapping relationship, and risk level matching results. The airport's overall risk situation heat map includes regional node spatial identifiers, spatiotemporal grouping results of behavioral trajectories, and a risk situation mapping structure.

3. The airport video data real-time analysis system according to claim 1, characterized in that, The video stream spatiotemporal feature modeling module includes: The multi-scale feature parsing submodule acquires airport surveillance video stream data, uses a 3D convolutional network to decompose the spatiotemporal features of continuous video frames, separates the trajectory vector of moving targets from the background light flow field distribution matrix, records the starting frame number and spatial bounding box coordinates of each behavioral event on the time axis, calculates the ratio of the first trigger position of key behaviors in the spatiotemporal sequence to the number of continuous frames, and generates a spatiotemporal distribution index of behavioral events. The dynamic segment reconstruction submodule extracts video frame sequences before and after key behavior triggers based on the spatiotemporal distribution index of the behavior events, performs spatiotemporal alignment operations on motion trajectory vectors and optical flow matrices according to behavior types, and integrates multiple event segments under the same behavior category to form a set of spatiotemporal feature segments. The feature sequence generation submodule, based on the set of spatiotemporal feature segments, counts the trigger frequency of various behavioral events within a unit time window, rearranges the feature segment set according to the order of behavioral trigger time, splices multiple event segments of the same category into a continuous sequence according to the first trigger position, uses a dynamic time warping algorithm to calibrate the time offset of the behavioral feature sequence, and outputs a standardized spatiotemporal behavioral feature sequence.

4. The airport video data real-time analysis system according to claim 3, characterized in that, The behavior trajectory mapping construction module includes: The behavior hierarchy mapping submodule loads predefined airport area spatial topology weights based on the standardized spatiotemporal behavior feature sequence, sorts the motion trajectory nodes and optical flow area nodes by hierarchical priority, and establishes a node rearrangement index from the top area of ​​the terminal building to the last level area of ​​the boarding gate. The topology path generation submodule extracts the spatiotemporal correlation strength parameters between adjacent level nodes based on the node rearrangement index, records the bidirectional connection weight and temporal continuity index of each node pair, and uses a graph embedding algorithm to calculate the spatial density and temporal consistency coefficient of the node path to generate a spatiotemporal topology path with weight attributes. The graph structure extraction submodule, based on the spatiotemporal topological path with weighted attributes, collects the spatial coordinate connection relationship and temporal continuity parameters of all nodes in the path, and constructs a multi-dimensional behavioral trajectory graph structure containing a node spatial location mapping table and a temporal correlation matrix.

5. The airport video data real-time analysis system according to claim 1, characterized in that, The abnormal region correlation analysis module includes: The multi-view node acquisition submodule extracts the set of behavior path terminal nodes in the coverage area of ​​different cameras based on the multi-dimensional behavior trajectory map structure, and records the monitoring view number and path duration parameters corresponding to each terminal node. The cross-regional overlap analysis submodule, based on the set of terminal nodes, calls the spatial coordinate sequence of nodes from any two monitoring perspectives, calculates the overlap degree of node distribution in the spatial grid, extracts the frequently occurring endpoint coordinate points in the overlapping area, counts the frequency difference of such coordinate points in the multi-view path, and uses a spatiotemporal density clustering algorithm to filter the abnormal area coordinate set whose spatial clustering exceeds a preset threshold. The association set generation submodule, based on the abnormal area coordinate set, traces back the node number of the corresponding coordinate in the original behavior path, integrates the spatial mapping relationship between the monitoring view number and the area coordinate, and outputs an abnormal behavior area association set containing the area coordinate frequency distribution map.

6. The airport video data real-time analysis system according to claim 1, characterized in that, The risk level semantic determination module includes: The regional semantic mapping submodule, based on the coordinates of the endpoint region in the abnormal behavior region association set, queries the predefined risk semantic label library in the airport electronic fence database to establish an index mapping table between regional coordinates and risk labels; The risk frequency statistics submodule calculates the associated frequency value of each type of risk label in the abnormal area coordinate set according to the index mapping table, and generates a risk label priority sequence by sorting the frequency values ​​from high to low. The risk level matching submodule, based on the risk tag priority sequence, matches the risk tag category with the highest associated frequency for each abnormal area coordinate, integrates the tag matching results of all area coordinates, and generates a multi-level risk semantic tag group containing risk level weight values.

7. The airport video data real-time analysis system according to claim 1, characterized in that, The situation structure visualization module includes: The physical area mapping submodule, based on the multi-level risk semantic tag group, parses the airport physical area node number corresponding to each risk tag, and records the spatial coordinate range of the node in the electronic map and the number of existing risk trajectories. The trajectory space partitioning submodule, based on the node number of the airport physical area, groups and maps the spatiotemporal data of behavioral trajectories to the corresponding geographical areas according to the risk level weight, and establishes a two-way spatial index structure between trajectory data and geographical areas. The heatmap generation submodule, based on the bidirectional spatial index structure, calculates the risk trajectory density value within each geographical area, maps the density value to color gradient parameters, and overlays the airport electronic map to generate a visual heatmap layer containing the real-time risk intensity distribution.

8. The airport video data real-time analysis system according to claim 1, characterized in that, It also includes a real-time early warning feedback module, which performs the following operations based on the airport's overall risk situation heat map: Extract the set of coordinates of physical regions in the heat map whose risk intensity exceeds a preset threshold; Based on the multi-level risk semantic tag group associated with the physical area coordinate set, the priority of risk types is determined; Generate an early warning instruction data package containing the risk coordinates, risk type, and recommended handling measures; The warning instruction data packet is pushed to the responsible terminal equipment in the corresponding area in real time.

9. The airport video data real-time analysis system according to claim 8, characterized in that, The real-time early warning feedback module also includes: The early warning verification submodule receives the handling result data fed back by the responsible terminal; Compare the original risk intensity value of the area to be verified with the real-time risk intensity value after treatment; When the rate of decline in risk intensity fails to reach the preset target, the warning level will be automatically upgraded and a secondary response instruction will be triggered.

Images