Split-screen display control method and system based on dynamic layout adjustment

Abstract

The invention relates to the field of split-screen display control, in particular to a split-screen display control method and system based on dynamic layout adjustment. The invention discloses a split-screen display control system based on dynamic layout adjustment. The split-screen display control system comprises a risk division module, an association graph construction module, a dominant association dimension selection module and a layout strategy adjustment module. According to the method, the physical space constraint of a traditional static layout is broken through by constructing a semantic association network of a monitoring scene, and a dynamic semantic association strength map between pictures is established; based on the map, the system can autonomously identify a monitoring picture cluster with strong semantic coupling, and instantly drive screen resources to dynamically gather together to a core picture set with close event association logic; it is ensured that an operator can visually grasp a conduction chain and an evolution panorama of a cross-regional event in a single window without manual splicing.

Description

Technical Field

[0001] The present invention relates to the field of split-screen display control, and in particular to a split-screen display control method and system based on dynamic layout adjustment. Background Art

[0002] In the field of large-scale security monitoring, command centers usually need to use split-screen systems to simultaneously display dozens or even hundreds of surveillance images to ensure global situational awareness. However, the current mainstream split-screen display technology has significant flaws: its pre-set fixed layout mode (such as static grid arrangement) cannot adapt to the dynamic changes of sudden security incidents, resulting in delayed response to key information. When monitoring personnel deal with multi-event related scenarios such as the chain reaction of criminal behavior, the spread of fire or equipment group failure, they have to repeatedly switch the focus of the screen, resulting in a long average emergency response time.

[0003] Although traditional systems attempt to use simple visual markers to mark abnormal images, their risk classification mechanism is relatively crude; operators need to repeatedly compare image details to determine the nature of the event, which can easily cause visual fatigue in continuous monitoring scenarios; especially when multiple areas are abnormal concurrently, the existing layout cannot automatically establish logical associations such as the movement routes of criminal gangs, the diffusion paths of hazardous chemical leaks, or the transmission chains of equipment failures, forcing operators to rely on experience and cognition to manually assemble images.

[0004] The deeper problem lies in the misplaced expression of the correlation between monitoring areas; the fixed split-screen strategy is based only on the physical location of the equipment, ignoring the fact that spatial proximity in actual scenarios sometimes deviates from the correlation between events: for example, equipment areas that are physically isolated but process-connected, and the flow paths of personnel between non-adjacent buildings, and other correlation models cannot be directly mapped through coordinate distance.

[0005] Therefore, in response to the above problems, the present invention proposes a split-screen display control method and system based on dynamic layout adjustment, which essentially solves the problem of adaptability of monitoring scenarios. Summary of the Invention

[0006] The present invention breaks through the physical space constraints of traditional static layouts by constructing a semantic association network of monitoring scenes. The system analyzes the semantic features of events in video streams and preset cross-regional linkage rules in real time, and establishes a dynamic semantic association strength map between images based on this. Based on this map, the system can autonomously identify clusters of monitoring images with strong semantic coupling, and instantly drive screen resources to dynamically converge towards a set of core images with tight event association logic. This split-screen reorganization mechanism, which is dominated by the intrinsic logic of events rather than physical coordinates, realizes the essential transition from "spatial tiled view" to "event context view", ensuring that operators can intuitively grasp the transmission chain and evolution panorama of cross-regional events in a single window without manual piecing together.

[0007] A split-screen display control method based on dynamic layout adjustment, comprising: For any of the several monitoring areas on the monitoring display screen, at any monitoring time point, the abnormal event detection model is used to analyze the video of the monitoring area and output the abnormal event detection results of the monitoring area, specifically including the probability of occurrence of various abnormal events between 0 and 1. The highest probability of occurrence is selected as the risk value of the monitoring area; based on the risk value of the monitoring area and the preset multi-level risk threshold range, the monitoring area is divided into high, medium and low risks, and different types of warning borders are added to the monitoring area based on the division results; Analyze the spatial correlation and event correlation between each monitoring area and construct a monitoring area correlation map. The spatial correlation is generated based on the geographic coordinate information of the monitoring area, and the event correlation is generated based on the historical occurrence of abnormal events in the monitoring area. Based on the historical occurrence of abnormal events in all monitored areas, spatial correlation or event correlation is selected as the dominant correlation dimension for each event type; At the current monitoring time point, determine whether any monitoring area is high-risk. If not, execute the daily layout strategy based on the historical abnormal event occurrence of each monitoring area; if so, execute the driven layout strategy based on all high-risk monitoring areas and their abnormal event detection results on the basis of the daily layout strategy.

[0008] Preferably, the abnormal events include the following three categories: Criminal incidents refer to incidents with intentional human characteristics; Disaster events refer to events with physical diffusion characteristics; Fault events indicate abnormal device functions.

[0009] Preferably, the spatial correlation and event correlation between each monitoring area are analyzed, and the specific operations are as follows: Based on the geographic coordinate data of the corresponding locations in all monitoring areas, the coordinate distance values ​​of each pair of corresponding locations in the monitoring areas are obtained; all the obtained coordinate distance values ​​are sorted from small to large, and the sorting results are linearly mapped to the range of 0-1, and then the difference between the mapping result and 1 is calculated to obtain the spatial correlation of each pair of monitoring areas; For any pair of monitoring areas A and B, the number of abnormal events that occurred in monitoring areas A and monitoring areas B in the historical data is recorded respectively, recorded as the total number of A times and the total number of B times. At the same time, when an abnormal event occurs in monitoring area A or monitoring area B, the number of abnormal events that occur simultaneously in the other monitoring area within a preset time threshold is obtained, recorded as the AB co-occurrence number; the ratio of the AB co-occurrence number to the total number of A times is calculated to obtain the first co-occurrence probability of monitoring areas A and B; the ratio of the AB co-occurrence number to the total number of B times is calculated to obtain the second co-occurrence probability of monitoring areas A and B; the maximum value of the first co-occurrence probability and the second co-occurrence probability is taken as the event correlation degree of monitoring areas A and B.

[0010] Preferably, based on the historical occurrence of abnormal events in all monitoring areas, spatial correlation or event correlation is selected as the dominant correlation dimension for each event type. The specific operation is as follows: For each crime incident, the first monitoring area where an anomaly is discovered is used as the initial area. After an anomaly is discovered in the initial area, if anomalies occur in other monitoring areas within a preset time threshold, the other monitoring areas where anomalies occur are used as subsequent impact areas, and the current crime incident is used as a sample crime incident. For each sample crime event, based on the spatial correlation between the current initial area and all other monitoring areas, all other monitoring areas are sorted from large to small, and then the rankings of all current subsequent impact areas are queried, and the average ranking of all subsequent impact areas is calculated to obtain the spatial correlation ranking of the current sample crime event; at the same time, based on the event correlation between the current initial area and all other monitoring areas, all other monitoring areas are sorted from large to small, and then the rankings of all current subsequent impact areas are queried, and the average ranking of all subsequent impact areas is calculated to obtain the event correlation ranking of the current sample crime event; Calculate the average spatial correlation ranking of all sample crime events to obtain the final spatial correlation ranking of crime events; calculate the average event correlation ranking of all sample crime events to obtain the final event correlation ranking of crime events; If the final spatial correlation ranking of crime-related events is higher than the final event correlation ranking, the spatial correlation is selected as the dominant correlation dimension of crime-related events; If the final spatial correlation ranking of crime-related events is lower than the final event correlation ranking, the event correlation degree is selected as the dominant correlation dimension of crime-related events; For disaster-related events and failure-related events, the same analysis operations as those for crime-related events are performed.

[0011] Preferably, based on the historical occurrence of abnormal events in each monitoring area, a daily layout strategy is executed, specifically including: Step 1: For a display screen with N×M monitoring areas, arrange the monitoring areas in order from left to right and from top to bottom. First, select the monitoring area with the most abnormal events in the historical data as the area to be arranged, and place it in the first position on the display screen. Query the type of abnormal events that occur most frequently in the area to be arranged. If the event type uses a spatial correlation-dominated strategy, obtain the spatial correlation of all other monitoring areas. Then, arrange the monitoring areas with spatial correlations higher than the preset spatial correlation threshold in subsequent positions of the area to be arranged according to the size of the spatial correlation. If the event type adopts the event correlation dominant strategy, the event correlation of all other monitoring areas is obtained, and the monitoring areas with event correlation higher than the preset event correlation threshold are arranged in the subsequent positions of the area to be arranged according to the size of the event correlation; Step 2: Then, for the remaining unarranged monitoring areas, perform the same operation as step 1 in the remaining positions of the display screen; Step 3: Repeat step 2 until all monitoring areas are arranged.

[0012] Preferably, based on all high-risk monitoring areas and their abnormal event detection results, a driving layout strategy is executed, specifically including: Based on the daily layout strategy, all current high-risk monitoring areas are sorted from highest to lowest according to risk value, and the following layout adjustment operations are performed in sequence from the highest to the lowest high-risk monitoring areas: Query the abnormal event category of the current high-risk monitoring area. If the current abnormal event category adopts the spatial correlation dominant strategy, obtain the spatial correlation of all other monitoring areas except the current high-risk monitoring area, and swap the monitoring areas with spatial correlation higher than the spatial correlation threshold with the monitoring areas at subsequent positions of the current high-risk monitoring area in order of spatial correlation; if the current abnormal event category adopts the event correlation dominant strategy, obtain the event correlation of all other monitoring areas except the current high-risk monitoring area, and swap the monitoring areas with event correlation higher than the event correlation threshold with the monitoring areas at subsequent positions of the current high-risk monitoring area in order of event correlation; If a high-risk monitoring area ranked subsequently is actively repositioned during a layout adjustment operation performed on a high-risk monitoring area, the layout adjustment operation will no longer be performed on the high-risk monitoring area ranked subsequently.

[0013] Preferably, the abnormal event detection model is established based on a three-dimensional convolutional neural network.

[0014] A split-screen display control system based on dynamic layout adjustment, comprising: The risk classification module includes an abnormal event detection unit and a risk classification unit. The abnormal event detection unit is used to analyze the video of the monitoring area using the abnormal event detection model and output the abnormal event detection results of the monitoring area. The risk classification unit is used to classify the monitoring area into high, medium and low risks based on the risk value of the monitoring area and the preset multi-level risk threshold range, and add different types of warning borders to the monitoring area based on the classification results. The correlation map construction module is used to analyze the spatial correlation and event correlation between each monitoring area and construct the monitoring area correlation map; The dominant correlation dimension selection module is used to select spatial correlation or event correlation as the dominant correlation dimension for each event type based on the historical occurrence of abnormal events in all monitoring areas; The layout strategy adjustment module is used to determine whether any monitoring area is high-risk at the current monitoring time point. If not, the daily layout strategy is executed based on the historical abnormal event occurrence of each monitoring area; if so, the driving layout strategy is executed based on all high-risk monitoring areas and their abnormal event detection results on the basis of the daily layout strategy.

[0015] The present invention has the following advantages: 1. The present invention breaks through the physical space constraints of traditional static layouts by constructing a semantic association network of monitoring scenes. The system analyzes the semantic features of events in video streams and preset cross-regional linkage rules in real time, and establishes a dynamic semantic association strength map between images based on this. Based on this map, the system can autonomously identify clusters of monitoring images with strong semantic coupling and instantly drive screen resources to dynamically converge towards a set of core images with tight event association logic. This split-screen reorganization mechanism, which is dominated by the intrinsic logic of events rather than physical coordinates, realizes an essential transition from a "spatial tiled view" to an "event context view", ensuring that operators can intuitively grasp the transmission chain and evolution panorama of cross-regional events in a single window without manual piecing together.

[0016] 2. This invention uses statistical methods to dynamically assign dominant association strategies to different event types, completely abandoning subjective preset rules; daily layout is based on the clustering arrangement of historical high-frequency events, while event-driven layout realizes the priority reconstruction of high-risk areas through risk value sorting and position exchange strategy; this data closed-loop design not only enhances the system's generalized adaptability to different monitoring scenarios, but also gives it the ability to self-optimize in the long term evolution. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a structural diagram of a split-screen display control system based on dynamic layout adjustment adopted in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0019] Embodiment 1, a split-screen display control method based on dynamic layout adjustment, comprising: For any of the several monitoring areas on the surveillance display, at any monitoring time point, the abnormal event detection model is used to analyze the video of the monitoring area and output the abnormal event detection results for the monitoring area, specifically including the probability of occurrence of various abnormal events between 0 and 1 (such as a crime probability of 0.7 and a disaster probability of 0.1). The highest probability is selected as the risk value of the monitoring area. Based on the risk value of the monitoring area and the preset multi-level risk threshold range, the monitoring area is divided into high, medium, and low risks, and different types of warning borders are added to the monitoring area based on the classification results. For example, low-risk monitoring areas use static borders, medium-risk monitoring areas use periodic flashing borders, and high-risk monitoring areas use pulsed borders with warning icons. Through the synergy of AI quantification of risk and dynamic visual enhancement, the speed of discovering abnormal targets and the accuracy of handling priority judgment are significantly improved. At the same time, it provides a decision-making basis for subsequent layout reorganization, forming a closed-loop security control chain of "risk identification-visual warning-spatial optimization", fundamentally solving the problems of missed detection of key targets and response delays in scenarios with monitoring information overload. The multi-level risk threshold range is determined by technical personnel in relevant fields. The spatial and event correlations between monitoring areas are analyzed to construct a monitoring area correlation map. Spatial correlation is generated based on the geographic coordinates of the monitoring areas and measures the physical proximity between areas. Event correlation is generated based on the historical occurrence of abnormal events in the monitoring areas and reflects the historical correlation between areas in the occurrence of abnormal events. The core function of constructing the monitoring area correlation map is to provide a data foundation and relationship network basis for subsequent intelligent screen layout decisions (daily layout strategies and driven layout strategies). It quantifies the physical proximity between areas and the linkage of historical event patterns, enabling the system to understand and utilize these potential "influence propagation paths" or "attention linkage relationships." For example, when a fire (disaster) warning is issued at a gas station (area A), the system can not only quickly locate the adjacent oil depot (area B) based on spatial correlation, but also identify the attached parking lot (area C) that has historically frequently experienced cascading equipment failures with the station based on event correlation. This map transforms physical space information and historical event patterns into knowledge and structure, and is the key cornerstone for the system to achieve "dynamic, correlated, and risk-oriented" split-screen display intelligent control.

[0020] Based on the historical occurrence of abnormal events across all monitored areas, spatial or event correlation is selected as the dominant correlation dimension for each event type. The core of this step is to intelligently select the most appropriate impact assessment criteria for each abnormal event type (crime, disaster, or fault) based on their correlation characteristics. The system conducts a thorough analysis of historical abnormal event records across all monitored areas, focusing specifically on the occurrence patterns of each type of event. Based on this historical empirical data, the system "learns" whether, for a crime-related incident, areas physically close to the incident (high spatial correlation) are more likely to experience chain reactions, or whether areas with a history of coordinated crime (high event correlation) pose a higher risk. Similarly, separate judgments are made for "disaster" (such as a spreading fire) or "fault" (such as a cascading equipment failure) events. After computational verification, the most reliable and predictive dimension (spatial correlation or event correlation) for each type of event is ultimately selected as the "dominant correlation dimension" for that event type. This eliminates the need for rigid, universal rules in the system's perception and decision-making, allowing it to be adaptively optimized based on historical evidence and event characteristics, significantly improving the rationality of layout adjustments and the efficiency of risk warnings. At the current monitoring time point, it is determined whether any monitoring area is high-risk. If not, the daily layout strategy is executed based on the historical occurrence of abnormal events in each monitoring area. This strategy arranges the screen position according to the occurrence pattern of abnormal events in the past in all monitoring areas (which areas are more frequent and what types of events are frequent) to ensure that daily monitoring can give priority to historical high-risk points; if so, on the basis of the daily layout strategy, the driven layout strategy is executed based on all high-risk monitoring areas and their abnormal event detection results; this strategy will use the pre-established regional association map to intelligently display other areas that are most closely associated with high-risk areas in physical space or historical event chains near the high-risk areas, so as to accurately guide the operator's attention to the most dangerous outbreak core and the key areas most likely to be affected in an emergency, realize dynamic focus and linkage warning based on real-time risk situation, and significantly improve the response speed to emergencies and the ability to prevent and control potential chain risks.

[0021] The abnormal events include the following three categories: Criminal incidents refer to incidents with intentional human characteristics; Disaster events refer to events with physical diffusion characteristics; Fault events indicate abnormal device functions.

[0022] Analyze the spatial correlation and event correlation between each monitoring area. The specific operations are as follows: Based on the geographic coordinate data of the corresponding locations in all monitoring areas, the coordinate distance values ​​of each pair of corresponding locations in the monitoring areas are obtained; all the obtained coordinate distance values ​​are sorted from small to large, and the sorting results are linearly mapped to the range of 0-1, and then the difference between the mapping result and 1 is calculated to obtain the spatial correlation of each pair of monitoring areas; For any pair of monitoring areas A and B, the number of abnormal events that occurred in monitoring areas A and monitoring areas B in the historical data is recorded respectively, recorded as the total number of A times and the total number of B times. At the same time, when an abnormal event occurs in monitoring area A or monitoring area B, the number of abnormal events that occur simultaneously in the other monitoring area within a preset time threshold is obtained, recorded as the AB co-occurrence number; the ratio of the AB co-occurrence number to the total number of A times is calculated to obtain the first co-occurrence probability of monitoring areas A and B; the ratio of the AB co-occurrence number to the total number of B times is calculated to obtain the second co-occurrence probability of monitoring areas A and B; the maximum value of the first co-occurrence probability and the second co-occurrence probability is taken as the event correlation degree of monitoring areas A and B.

[0023] Based on the historical occurrence of abnormal events in all monitoring areas, spatial correlation or event correlation is selected as the dominant correlation dimension for each event type. The specific operations are as follows: For each crime incident, the first monitoring area where an anomaly is discovered is used as the initial area. After an anomaly is discovered in the initial area, if anomalies occur in other monitoring areas within a preset time threshold, the other monitoring areas where anomalies occur are used as subsequent impact areas, and the current crime incident is used as a sample crime incident. For each sample crime event, based on the spatial correlation between the current initial area and all other monitoring areas, all other monitoring areas are sorted from large to small, and then the rankings of all current subsequent impact areas are queried, and the average ranking of all subsequent impact areas is calculated to obtain the spatial correlation ranking of the current sample crime event; at the same time, based on the event correlation between the current initial area and all other monitoring areas, all other monitoring areas are sorted from large to small, and then the rankings of all current subsequent impact areas are queried, and the average ranking of all subsequent impact areas is calculated to obtain the event correlation ranking of the current sample crime event; Calculate the average spatial correlation ranking of all sample crime events to obtain the final spatial correlation ranking of crime events; calculate the average event correlation ranking of all sample crime events to obtain the final event correlation ranking of crime events; If the final spatial correlation ranking of crime-related events is higher than the final event correlation ranking, the spatial correlation is selected as the dominant correlation dimension of crime-related events; If the final spatial correlation ranking of crime-related events is lower than the final event correlation ranking, the event correlation degree is selected as the dominant correlation dimension of crime-related events; For disaster-related events and failure-related events, the same analysis operations as those for crime-related events are performed.

[0024] Based on the historical abnormal events in each monitoring area, daily layout strategies are implemented, including: Step 1: For a display screen with N×M monitoring areas, arrange the monitoring areas in order from left to right and from top to bottom. First, select the monitoring area with the most abnormal events in the historical data as the area to be arranged. Place the area to be arranged in the first position of the display screen, that is, the first row and first column. Query the type of abnormal events that occur most frequently in the area to be arranged. If the event type adopts the spatial correlation dominant strategy, obtain the spatial correlation of all other monitoring areas. Place the monitoring areas with spatial correlations higher than the preset spatial correlation threshold in the subsequent positions of the area to be arranged according to the size of the spatial correlation. If the event type adopts the event correlation dominant strategy, the event correlation of all other monitoring areas is obtained, and the monitoring areas with event correlation higher than the preset event correlation threshold are arranged in the subsequent positions of the area to be arranged according to the size of the event correlation; Step 2: Then, for the remaining unarranged monitoring areas, perform the same operation as step 1 in the remaining positions of the display screen; Step 3: Repeat step 2 until all monitoring areas are arranged.

[0025] Based on all high-risk monitoring areas and their abnormal event detection results, drive layout strategies are executed, including: Based on the daily layout strategy, all current high-risk monitoring areas are sorted from highest to lowest according to risk value, and the following layout adjustment operations are performed in sequence from the highest to the lowest high-risk monitoring areas: Query the abnormal event category of the current high-risk monitoring area. If the current abnormal event category adopts the spatial correlation dominant strategy, obtain the spatial correlation of all other monitoring areas except the current high-risk monitoring area, and swap the monitoring areas with spatial correlation higher than the spatial correlation threshold with the monitoring areas at subsequent positions of the current high-risk monitoring area in order of spatial correlation; if the current abnormal event category adopts the event correlation dominant strategy, obtain the event correlation of all other monitoring areas except the current high-risk monitoring area, and swap the monitoring areas with event correlation higher than the event correlation threshold with the monitoring areas at subsequent positions of the current high-risk monitoring area in order of event correlation; If a high-risk monitoring area ranked subsequently is actively repositioned during a layout adjustment operation performed on a high-risk monitoring area, the layout adjustment operation will no longer be performed on the high-risk monitoring area ranked subsequently.

[0026] The abnormal event detection model is built based on a three-dimensional convolutional neural network.

[0027] A split-screen display control system based on dynamic layout adjustment, such as Figure 1 Shown, including: The risk classification module includes an abnormal event detection unit and a risk classification unit; the abnormal event detection unit is used to analyze the video of any monitoring area among the several monitoring areas on the monitoring display screen at any monitoring time point using the abnormal event detection model, and output the abnormal event detection results of the monitoring area, specifically including the probability of occurrence of various abnormal events between 0 and 1, and select the highest probability of occurrence as the risk value of the monitoring area; the risk classification unit is used to classify the monitoring area into high, medium and low risks according to the risk value of the monitoring area and the preset multi-level risk threshold interval, and add different types of warning borders to the monitoring area based on the classification results; The correlation map construction module is used to analyze the spatial correlation and event correlation between each monitoring area and construct a monitoring area correlation map. The spatial correlation is generated based on the geographic coordinate information of the monitoring area, and the event correlation is generated based on the historical occurrence of abnormal events in the monitoring area. The dominant correlation dimension selection module is used to select spatial correlation or event correlation as the dominant correlation dimension for each event type based on the historical occurrence of abnormal events in all monitoring areas; The layout strategy adjustment module is used to determine whether any monitoring area is high-risk at the current monitoring time point. If not, the daily layout strategy is executed based on the historical abnormal event occurrence of each monitoring area; if so, the driving layout strategy is executed based on all high-risk monitoring areas and their abnormal event detection results on the basis of the daily layout strategy.

[0028] It should be understood that those skilled in the art may make improvements or modifications based on the above description, and all such improvements and modifications shall fall within the scope of protection of the appended claims. Any portion of this specification not described in detail is prior art known to those skilled in the art.

Claims

1. A split-screen display control method based on dynamic layout adjustment, characterized in that: include: For any of the several monitoring areas on the monitoring screen, at any monitoring time point, the abnormal event detection model is used to analyze the video of the monitoring area and output the abnormal event detection results of the monitoring area, including the probability of occurrence of various abnormal events between 0 and 1. The highest probability is selected as the risk value of the monitoring area; Based on the risk value of the monitoring area and the preset multi-level risk threshold range, the monitoring area is divided into high, medium and low risks, and different types of warning borders are added to the monitoring area based on the division results; Analyze the spatial correlation and event correlation between each monitoring area and construct a monitoring area correlation map. The spatial correlation is generated based on the geographic coordinate information of the monitoring area, and the event correlation is generated based on the historical occurrence of abnormal events in the monitoring area. Based on the historical occurrence of abnormal events in all monitored areas, spatial correlation or event correlation is selected as the dominant correlation dimension for each event type; At the current monitoring time point, determine whether any monitoring area is high-risk. If not, execute the daily layout strategy based on the historical abnormal event occurrence of each monitoring area; if so, execute the driven layout strategy based on all high-risk monitoring areas and their abnormal event detection results on the basis of the daily layout strategy.

2. The method for controlling split-screen display based on dynamic layout adjustment according to claim 1, characterized in that: The abnormal events include the following three categories: Criminal incidents refer to incidents with intentional human characteristics; Disaster events refer to events with physical diffusion characteristics; Fault events indicate abnormal device functions.

3. The method for controlling split-screen display based on dynamic layout adjustment according to claim 2, characterized in that: Analyze the spatial correlation and event correlation between each monitoring area. The specific operations are as follows: Based on the geographic coordinate data of the corresponding locations in all monitoring areas, the coordinate distance values ​​of each pair of corresponding locations in the monitoring areas are obtained; all the obtained coordinate distance values ​​are sorted from small to large, and the sorting results are linearly mapped to the range of 0-1, and then the difference between the mapping result and 1 is calculated to obtain the spatial correlation of each pair of monitoring areas; For any pair of monitoring areas A and B, the number of abnormal events that occurred in monitoring area A and monitoring area B in the historical data is recorded respectively, recorded as the total number of A times and the total number of B times. At the same time, the number of abnormal events that occurred in monitoring area A or monitoring area B within the preset time threshold when an abnormal event occurred in the other monitoring area is obtained, recorded as the AB co-occurrence number; Calculate the ratio of the number of co-occurrences of AB to the total number of A to obtain the first co-occurrence probability of monitoring areas A and B; Calculate the ratio of the number of co-occurrences of AB to the total number of B to obtain the second co-occurrence probability of monitoring areas A and B; The maximum value of the first co-occurrence probability and the second co-occurrence probability is used as the event correlation degree of monitoring areas A and B.

4. The method for controlling split-screen display based on dynamic layout adjustment according to claim 3, characterized in that: Based on the historical occurrence of abnormal events in all monitoring areas, spatial correlation or event correlation is selected as the dominant correlation dimension for each event type. The specific operations are as follows: For each crime incident, the first monitoring area where an anomaly is discovered is used as the initial area. After an anomaly is discovered in the initial area, if anomalies occur in other monitoring areas within a preset time threshold, the other monitoring areas where anomalies occur are used as subsequent impact areas, and the current crime incident is used as a sample crime incident. For each sample crime event, based on the spatial correlation between the current initial area and all other monitoring areas, all other monitoring areas are sorted from large to small, and then the rankings of all current subsequent impact areas are queried, and the average ranking of all subsequent impact areas is calculated to obtain the spatial correlation ranking of the current sample crime event; at the same time, based on the event correlation between the current initial area and all other monitoring areas, all other monitoring areas are sorted from large to small, and then the rankings of all current subsequent impact areas are queried, and the average ranking of all subsequent impact areas is calculated to obtain the event correlation ranking of the current sample crime event; Calculate the average of the spatial correlation rankings of all sample crime events to obtain the final spatial correlation ranking of crime events; Calculate the average of the event correlation rankings of all sample crime events to obtain the final event correlation ranking of crime events; If the final spatial correlation ranking of crime-related events is higher than the final event correlation ranking, the spatial correlation is selected as the dominant correlation dimension of crime-related events; If the final spatial correlation ranking of crime-related events is lower than the final event correlation ranking, the event correlation degree is selected as the dominant correlation dimension of crime-related events; For disaster-related events and failure-related events, the same analysis operations as those for crime-related events are performed.

5. The method for controlling split-screen display based on dynamic layout adjustment according to claim 4, characterized in that: Based on the historical abnormal events in each monitoring area, daily layout strategies are implemented, including: Step 1: For a display screen with N×M monitoring areas, arrange the monitoring areas in order from left to right and from top to bottom. First, select the monitoring area with the most abnormal events in the historical data as the area to be arranged, and place it in the first position on the display screen. Query the type of abnormal events that occur most frequently in the area to be arranged. If the event type uses a spatial correlation-dominated strategy, obtain the spatial correlation of all other monitoring areas. Then, arrange the monitoring areas with spatial correlations higher than the preset spatial correlation threshold in subsequent positions of the area to be arranged according to the size of the spatial correlation. If the event type adopts the event correlation dominant strategy, the event correlation of all other monitoring areas is obtained, and the monitoring areas with event correlation higher than the preset event correlation threshold are arranged in the subsequent positions of the area to be arranged according to the size of the event correlation; Step 2: Then, for the remaining unarranged monitoring areas, perform the same operation as step 1 in the remaining positions of the display screen; Step 3: Repeat step 2 until all monitoring areas are arranged.

6. The method for controlling split-screen display based on dynamic layout adjustment according to claim 5, characterized in that: Based on all high-risk monitoring areas and their abnormal event detection results, drive layout strategies are executed, including: Based on the daily layout strategy, all current high-risk monitoring areas are sorted from highest to lowest according to risk value, and the following layout adjustment operations are performed in sequence from the highest to the lowest high-risk monitoring areas: Query the abnormal event category of the current high-risk monitoring area. If the current abnormal event category adopts the spatial correlation dominant strategy, obtain the spatial correlation of all other monitoring areas except the current high-risk monitoring area, and swap the monitoring areas with spatial correlation higher than the spatial correlation threshold with the monitoring areas at subsequent positions of the current high-risk monitoring area in order of spatial correlation; if the current abnormal event category adopts the event correlation dominant strategy, obtain the event correlation of all other monitoring areas except the current high-risk monitoring area, and swap the monitoring areas with event correlation higher than the event correlation threshold with the monitoring areas at subsequent positions of the current high-risk monitoring area in order of event correlation; If a high-risk monitoring area ranked subsequently is actively repositioned during a layout adjustment operation performed on a high-risk monitoring area, the layout adjustment operation will no longer be performed on the high-risk monitoring area ranked subsequently.

7. The method for controlling split-screen display based on dynamic layout adjustment according to claim 6, characterized in that: The abnormal event detection model is built based on a three-dimensional convolutional neural network.

8. A split-screen display control system based on dynamic layout adjustment, characterized in that: The system is applied to a split-screen display control method based on dynamic layout adjustment as described in any one of claims 1 to 7 above, comprising: a risk division module, including an abnormal event detection unit and a risk division unit; the abnormal event detection unit is used to analyze the video of the monitoring area using an abnormal event detection model and output the abnormal event detection result of the monitoring area; the risk division unit is used to divide the monitoring area into high, medium and low risks according to the risk value of the monitoring area and the preset multi-level risk threshold interval, and add different types of warning borders to the monitoring area based on the division results; The correlation map construction module is used to analyze the spatial correlation and event correlation between each monitoring area and construct the monitoring area correlation map; The dominant correlation dimension selection module is used to select spatial correlation or event correlation as the dominant correlation dimension for each event type based on the historical occurrence of abnormal events in all monitoring areas; The layout strategy adjustment module is used to determine whether any monitoring area is high-risk at the current monitoring time point. If not, the daily layout strategy is executed based on the historical abnormal event occurrence of each monitoring area; if so, the driving layout strategy is executed based on all high-risk monitoring areas and their abnormal event detection results on the basis of the daily layout strategy.

Images