Banking scene panoramic verification method and device based on unmanned vehicle and unmanned aerial vehicle linkage

Abstract

The application provides a bank scene panoramic verification method and device based on unmanned vehicle and unmanned aerial vehicle linkage, relates to the technical field of intelligent monitoring, and comprises the following steps: analyzing the distribution characteristics of static monitoring equipment, combining the space-time risks of the bank scene, and establishing an abnormality identification network; performing space-time analysis on the real-time monitoring data of the static monitoring equipment through the abnormality identification network, identifying abnormal events; positioning the space-time observation data compensation demand based on the abnormal events, calling the unmanned vehicle and / or unmanned aerial vehicle to perform verification tasks according to the space-time observation data compensation demand; fusing the multi-source data in the verification process, establishing panoramic verification results, and triggering corresponding disposal processes and data archiving. The application solves the technical problem that most of the prior art relies on a single data source, can only provide a fixed perspective and monitoring range, results in data loss or poor monitoring effect, and further results in the decline of the identification ability of abnormal events, thereby affecting the safety prevention and control effect.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology, specifically to a method and equipment for conducting on-site panoramic verification of banks based on the linkage of unmanned vehicles and drones. Background Technology

[0002] With the continuous development of intelligent monitoring systems, traditional monitoring methods can no longer meet the increasingly complex security needs. Especially in high-security locations such as banks, monitoring equipment often has blind spots and monitoring dead zones, making it difficult to achieve comprehensive, real-time, and efficient security monitoring. Bank sites are usually composed of a large number of static monitoring devices, such as cameras and sensors. These devices have certain limitations in terms of monitoring range and spatiotemporal adaptability. In addition, with changes in environmental conditions, such as changes in lighting and weather, as well as the complexity of target behavior, static monitoring equipment often struggles to identify abnormal events in real time. In fact, when monitoring equipment malfunctions, the on-site monitoring capability will be greatly reduced, affecting the effectiveness of security control. Summary of the Invention

[0003] This application provides a method and equipment for panoramic on-site verification of banks based on the linkage of unmanned vehicles and drones. It aims to solve the technical problem that most existing technologies rely on a single data source, which can only provide a fixed perspective and monitoring range, resulting in data loss or poor monitoring effect, which in turn leads to a decrease in the ability to identify abnormal events and affects the effectiveness of security and prevention.

[0004] The first aspect disclosed in this application provides a method for panoramic on-site verification of banks based on the linkage of unmanned vehicles and drones. The method includes: analyzing the distribution characteristics of static monitoring equipment and establishing an anomaly identification network in combination with the spatiotemporal risks of the bank site; performing spatiotemporal analysis on the real-time monitoring data of the static monitoring equipment through the anomaly identification network to identify abnormal events; locating the spatiotemporal observation data compensation needs based on the abnormal events, and calling unmanned vehicles and / or drones to perform verification tasks according to the spatiotemporal observation data compensation needs; fusing multi-source data in the verification process to establish panoramic verification results, and triggering corresponding handling procedures and data archiving.

[0005] The second aspect disclosed in this application provides a computer device, the electronic device comprising: a processor; and a memory for storing processor-executable instructions; wherein the processor is used to execute the above-described method for on-site panoramic verification of banks based on unmanned vehicles and drones.

[0006] One or more technical solutions provided in this application have at least the following beneficial effects: By analyzing the layout of static monitoring equipment at bank sites and the spatiotemporal risks, and combining this with the distribution characteristics of existing monitoring facilities, an intelligent anomaly detection network is established. This network can accurately assess the monitoring capabilities of each device and dynamically adjust monitoring strategies based on risk factors in the scenario, thereby effectively improving the comprehensive coverage of the monitoring range and target area. The anomaly detection network receives and analyzes real-time data from static monitoring equipment, performing spatiotemporal analysis to identify abnormal events at the bank site. Through this process, it can react quickly based on known risk models and real-time data, promptly detecting potential security threats. In blind spots not covered by static monitoring equipment or due to environmental changes... In cases of decreased monitoring capabilities, to compensate for the spatiotemporal data loss caused by abnormal event location, unmanned vehicles and / or drones are deployed for dynamic supplementary monitoring. These unmanned devices can fill the blind spots of static equipment, acquire close-range behavioral characteristics and high-altitude perspective data of the target area, and ensure comprehensive observation of the target area. During the verification process, data from unmanned vehicles, drones, and static monitoring equipment are collected and fused in real time. Through comparison and integration of multi-source data, a panoramic verification result is generated. This panoramic verification result provides comprehensive information support for subsequent handling procedures and provides a complete record for data archiving, ensuring that data support can be provided for future safety incident tracing, analysis, and optimization.

[0007] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the process for a method of conducting on-site panoramic verification of a bank based on the linkage of unmanned vehicles and drones, provided in an embodiment of this application.

[0009] Figure 2 This is a schematic diagram of the structure of an exemplary computer device provided in an embodiment of this application.

[0010] Explanation of reference numerals in the attached drawings: Processor 21, Memory 22, Input device 23, Output device 24. Detailed Implementation

[0011] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0012] Example 1, as Figure 1 As shown in the embodiment of this application, a method for conducting on-site panoramic verification of banks based on the linkage of unmanned vehicles and drones is provided. The method includes: By analyzing the distribution characteristics of static monitoring equipment and combining this with the spatiotemporal risks at the bank site, an anomaly identification network is established.

[0013] In bank operations, the deployment of static monitoring equipment, such as cameras and sensors, is based on different monitoring needs and coverage areas. First, the geographical distribution, field of view, and imaging quality of these devices are analyzed to gain a comprehensive understanding of their observation capabilities and limitations. Then, considering the spatiotemporal risks at the bank site—which refer to events or anomalies that may occur at different times and locations, such as theft or vandalism—potential high-risk areas and time periods are identified through spatiotemporal environment analysis. Based on the equipment distribution characteristics and spatiotemporal risks, an anomaly detection network is established. This network uses monitoring data to identify whether anomalies occur on-site. A comprehensive analysis of equipment observation capabilities and spatiotemporal risks forms a network structure used to continuously analyze on-site data.

[0014] Based on the task requirements and the spatiotemporal characteristics of static monitoring equipment, the anomaly detection network adopts a convolutional neural network (CNN) structure. It extracts local spatial features through convolutional layers and reduces dimensionality through pooling layers to improve computational efficiency. CNNs can effectively process spatial features in image and video data for anomaly detection in surveillance videos, such as illegal intrusion and object movement. Using video surveillance equipment, the input data is video frame data, which is a multidimensional array. For example, a video frame is represented as an image array: [H, W, C], where HHH is the height, W is the width, and C is the number of color channels. The spatiotemporal analysis unit is represented as a matrix, with each spatiotemporal unit corresponding to a spatial location and time period. Each cell contains observation data and risk information for that spatiotemporal location. The output is a binary or multi-class label indicating whether an anomaly event has occurred at a certain spatiotemporal location. For multi-class tasks, there are multiple labels representing different types of anomalies, such as theft and equipment malfunction. The output also includes the specific spatiotemporal location and time of the anomaly.

[0015] The training data for the anomaly detection network includes labeled normal and anomalous events. In a bank setting, historical surveillance data, such as camera video or sensor data, is used, with annotations indicating the specific location and time of anomalous events, such as theft or equipment malfunction. When the amount of data is limited, data augmentation techniques are used to generate more training samples. Since the goal is to identify anomalous events, a supervised learning method is employed. Using labeled training data, the network weights are optimized through backpropagation algorithms and loss functions, such as cross-entropy loss or mean squared error. Ultimately, the anomaly detection network is able to accurately identify anomalous events.

[0016] Anomaly identification networks are used to perform spatiotemporal analysis on real-time monitoring data from static monitoring equipment to identify abnormal events.

[0017] Static monitoring equipment continuously captures and transmits data, such as video surveillance and sensor data. This data is sent to an anomaly detection network for spatiotemporal analysis. This analysis, based on the monitoring equipment's observation capabilities (such as field of view and time period), identifies potential abnormal events, such as unauthorized intrusions or equipment malfunctions. This analysis is based on observation data from the equipment within a specific time and spatial range, combined with a spatiotemporal risk model. Through data analysis, abnormal events are filtered from a large amount of real-time monitoring data, and the specific spatiotemporal location of the anomaly is marked. At this point, information such as the type of abnormal event, its occurrence time, and its specific location is extracted, preparing for subsequent verification and processing.

[0018] Based on the spatiotemporal observation data compensation requirements for the aforementioned abnormal events, unmanned vehicles and / or drones are invoked to perform verification tasks according to the spatiotemporal observation data compensation requirements.

[0019] After identifying anomalies, the spatiotemporal environment at the time of the event is analyzed to determine if there are any monitoring blind spots or insufficient field of view. Based on an assessment of existing static monitoring equipment data, it is determined what data needs to be supplemented. This supplementary data requirement comes from unmanned vehicles (UAVs) used for close-range behavior analysis or drones used to acquire high-level spatial distribution data. According to the compensation requirements of the anomaly, appropriate UAVs and / or drones are dispatched to perform verification tasks. These unmanned devices execute tasks according to the plan based on target location, monitoring requirements, and other information, acquiring missing monitoring data. The UAVs and drones will supplement spatiotemporal observation data according to actual needs, ensuring the comprehensiveness of the verification process.

[0020] The multi-source data during the verification process is integrated to establish a comprehensive verification result, and corresponding handling procedures and data archiving are triggered.

[0021] During the verification process, unmanned vehicles, drones, and static monitoring equipment all generate data. To obtain comprehensive verification results, data from different sources are fused. Specifically, close-range behavioral feature data from unmanned vehicles and high-altitude spatial distribution information from drones are fused. Then, these fused data are spatiotemporally aligned to ensure consistency and complementarity. By fusing multi-source data, a comprehensive panoramic verification result is generated. This result integrates data from various monitoring devices, comprehensively presenting the monitoring situation at the bank and providing detailed event analysis and behavior recognition. Based on the panoramic verification result, corresponding handling procedures are triggered. For example, if an anomaly is detected, alarms are automatically activated, manpower is dispatched, and equipment maintenance is performed. In addition, all data and verification results are archived for easy review and analysis later.

[0022] For example, suppose a gathering event occurs in a bank lobby, with customers concentrated in one area, potentially posing a security risk. The handling process is as follows: First, analyze the distribution of monitoring equipment in the lobby, including cameras, sensors, etc., and combine this with the spatiotemporal risks of the bank, such as peak business hours or specific areas, to establish an anomaly identification network. The anomaly identification network analyzes the real-time monitoring data of the lobby, identifies abnormal crowd gathering in the lobby during a certain period, and marks the area as a potential risk point. Based on the specific location of the gathering event and the risk exposure difference, determine whether additional drones / unmanned vehicles are needed for on-site verification to ensure monitoring coverage. The drones provide high-angle images, which, combined with ground sensor data, generate panoramic verification results, triggering alarms and emergency response procedures.

[0023] Test experiments show that the panoramic inspection method based on the linkage of unmanned vehicles and drones improves the accuracy of anomaly identification by 15%, especially in high-traffic areas. In addition, the response time is reduced by 40% compared with traditional methods. This is due to the real-time data transmission of drones and the rapid coverage capability of unmanned vehicles. Specifically, the traditional method takes an average of 15 minutes to identify abnormal events, while the new system, after using unmanned vehicles and drones, reduces the response time to 9 minutes, significantly improving safety and response efficiency.

[0024] Furthermore, by analyzing the distribution characteristics of static monitoring equipment and combining this with the spatiotemporal risks at the bank site, an anomaly identification network is established, including: The bank site is spatially and temporally discretized to construct multiple spatiotemporal analysis units. Functional attributes, behavioral triggering attributes, and historical anomaly attributes of each spatiotemporal analysis unit are extracted to determine the corresponding spatiotemporal risk value. The location parameters, field-of-view parameters, imaging parameters, and environmental adaptation parameters of each static monitoring device are analyzed to establish an effective observation matrix for each spatiotemporal analysis unit. Based on the spatiotemporal risk value and the effective observation matrix, a risk exposure difference matrix is ​​established, and an anomaly identification network is constructed based on this risk exposure difference matrix.

[0025] Dividing the bank's premises into several spatial units, each representing a specific geographical area (e.g., a floor, a corridor, a room), allows for detailed analysis of the risks and monitoring capabilities of each area. The discretized spatial units facilitate precise location of monitoring areas and potential anomalies. Dividing time into multiple time periods (e.g., hours, minutes, seconds) ensures detailed temporal analysis of bank activities. Spatiotemporal risk analysis is based not only on spatial location but also on the specific time of event occurrence. Temporal discretization helps analyze the dynamic changes in the bank's premises from a temporal perspective, such as monitoring needs and risk levels at different times. The discretized spatial and temporal results together constitute a spatiotemporal analysis unit, each consisting of a specific spatial location and a corresponding time period. This allows for independent risk analysis of each spatiotemporal analysis unit, and each unit can contain different monitoring data, event behaviors, and risk assessments.

[0026] Functional attributes refer to the basic functions possessed by the spatiotemporal analysis unit, such as the monitoring function of a certain area, the business peak during a certain period, etc. Based on the functions of different areas, it is possible to determine the contribution of that area to the bank's operational security and the potential risks it may face. Behavioral triggering attributes refer to the probability of certain events or behaviors occurring and the security incidents they may trigger. For example, a large flow of people during a certain period may increase the risk of certain security incidents. By analyzing these behavioral triggering attributes, it is possible to predict which behaviors may cause anomalies in a specific spatiotemporal analysis unit. Historical anomaly attributes are extracted based on anomaly events that occurred in historical monitoring data. This historical data on anomalies provides a reference for the risk assessment of the current spatiotemporal analysis unit.

[0027] By comprehensively analyzing the aforementioned functional attributes, behavioral triggering attributes, and historical anomaly attributes, a spatiotemporal risk value is calculated for each spatiotemporal analysis unit. This value reflects the probability and severity of anomalies occurring within that unit over a specific time period. Specifically, the spatiotemporal risk value is quantified through a comprehensive analysis of functional attributes, behavioral triggering attributes, and historical anomaly attributes. These attributes assess the risk of each unit from three perspectives: equipment function, behavioral patterns, and historical data. The final spatiotemporal risk value is obtained through weighted summation. Historical data, by calculating the frequency and impact of events, provides the basis for risk assessment.

[0028] The installation location of each static monitoring device is analyzed, as its location directly affects its monitoring range and coverage area. Field of view parameters include the device's viewing angle and monitoring distance; different devices have different viewing angles and distances. Analyzing these parameters clarifies the device's observation range. Imaging parameters refer to the device's imaging quality, such as resolution and image sharpness. These factors affect whether the monitoring device can clearly capture the required details, thus impacting its ability to identify abnormal events. The device is affected by environmental factors, such as changes in lighting and weather, leading to a decrease in monitoring effectiveness. Therefore, it is necessary to determine the device's environmental adaptability to ensure its reliability under different environmental conditions. Based on the above parameters, an effective observation matrix is ​​established, representing the observation capability of each monitoring device in different spatiotemporal analysis units. This matrix reflects the monitoring effect of the device in each spatiotemporal analysis unit, thus affecting subsequent risk identification and data analysis.

[0029] Based on spatiotemporal risk values ​​and the effective observation matrix, the risk exposure difference for each spatiotemporal analysis unit is calculated. This risk exposure difference refers to the difference between the potential risk of a spatiotemporal analysis unit and the current observation capability of the monitoring equipment. Specifically, it's the difference between the potential risk and the observation capability. If the potential risk is higher than the observation capability, the exposure difference is positive, indicating that there is an insufficiently monitored risk in that area; a lower exposure difference indicates good monitoring coverage. Combining the spatiotemporal risk values ​​with observation degradation characteristic curves, such as the observation capability decline curves of equipment under different environmental conditions, the risk exposure difference of each spatiotemporal analysis unit in its current state is calculated. This coupling, based on equipment performance degradation and environmental factors, further improves the accuracy of risk assessment. Based on the risk exposure difference matrix, an anomaly identification network is established. This network identifies which areas and time periods have larger risk exposure differences, thus constructing a network structure reflecting the high-risk, low-perception coupling relationship. Through this network, monitoring data can be prioritized for high-risk areas, improving the efficiency of anomaly event identification.

[0030] Furthermore, based on the aforementioned spatiotemporal risk values ​​and the effective observation matrix, a risk exposure difference matrix is ​​established, including: Based on the effective observation matrix, a time-series statistical analysis of the observation capabilities of each spatiotemporal analysis unit at different time periods is performed to construct the equipment observation stability distribution. Combined with historical anomaly identification results, the identification success rate, false alarm rate, and missed alarm rate of each spatiotemporal analysis unit are inversely mapped to generate an identification credibility distribution. Based on the observation stability distribution and the identification credibility distribution, observation degradation analysis is performed on the spatiotemporal analysis units to identify absolute blind zones, time-varying blind zones, and false identification blind zones. According to the absolute blind zones, time-varying blind zones, and false identification blind zones, a joint analysis of target imaging quality, illumination change trends, and dynamic occlusion patterns is performed to extract degradation characteristic curves for each spatiotemporal analysis unit. Based on the degradation characteristic curves and spatiotemporal risk values, the risk exposure difference of each spatiotemporal analysis unit is calculated, and the risk exposure difference matrix is ​​constructed.

[0031] To evaluate the long-term performance and stability of the equipment, a time-series statistical analysis was conducted on the observation capabilities of each spatiotemporal analysis unit based on the established effective observation matrix. Specifically, by performing time-series analysis on the monitoring data of each spatiotemporal analysis unit, the observation stability of the equipment in different time periods was calculated. For example, the monitoring capability of the equipment in certain time periods is affected by factors such as illumination and weather changes. The statistical indicators in the time-series statistical analysis include: mean, which measures the average observation capability of the equipment over a certain period, revealing its overall performance; variance, which measures the volatility of observation capability; a larger variance indicates less stable observation capability, affecting the identification of abnormal events; standard deviation, the square root of the variance, which directly reflects the amplitude of fluctuations in the equipment's observation capability; skewness, which measures the degree of skewness in the distribution of the equipment's observation capability; a skewness greater than 0 indicates high observation capability, and a skewness less than 0 indicates low observation capability; and kurtosis, which measures the sharpness of the distribution of the equipment's observation capability, reflecting extreme fluctuations in observation capability. Based on these statistical indicators, a stability distribution for equipment observation is constructed. The purpose of this stability distribution is to reflect the stability of the equipment across various spatiotemporal analysis units. Specifically, the equipment's volatility can be described by its standard deviation or variance over a given time period. The stability distribution is represented by a distribution curve, such as a normal distribution or a gamma distribution, with the specific form determined based on the statistical results. This distribution reflects the observational stability of the equipment over different time periods and can indicate which equipment or spatiotemporal analysis units exhibit significant fluctuations in their observational capabilities, thereby affecting the accuracy of anomaly event identification.

[0032] Historical anomaly identification results provide monitoring and analysis of past events. Based on these results, it analyzes which spatiotemporal analysis units experienced anomalous events in the past and whether the identification of these events was successful. Specifically, the success rate refers to the percentage of accurately identified anomalous events; a higher success rate means the equipment can accurately identify anomalous events. The false alarm rate refers to the proportion of normal events incorrectly marked as anomalous; a high false alarm rate means too many invalid alarms are triggered, resulting in wasted resources. The missed alarm rate refers to the proportion of genuine anomalous events not identified; a high missed alarm rate means potential anomalous events are not detected in time. By combining historical identification results with the spatiotemporal risk values ​​of each spatiotemporal analysis unit and the equipment's observation stability, a reverse mapping is performed. This process combines the equipment's identification performance, including success rate, false alarm rate, and missed alarm rate, with the observation quality of its spatiotemporal analysis unit to determine the identification accuracy of different equipment and spatiotemporal analysis units. Specifically, these performance indicators are combined with the equipment's observation capabilities, and a weighted function is used to synthesize the influence of each indicator to obtain a comprehensive reliability score. The weighting function implies that a higher confidence value indicates the device can accurately identify anomalous events within that spatiotemporal analysis unit when its success rate is high, its false alarm rate and false negative rate are low, and its observation capabilities are stable. Conversely, a lower confidence value indicates poor identification performance within that spatiotemporal analysis unit, potentially leading to missed anomalous events or excessive false alarms. Through inverse mapping, a confidence distribution is generated, reflecting the reliability of the device in identifying anomalous events across different time periods within each spatiotemporal analysis unit.

[0033] By combining the observation stability distribution and recognition reliability distribution of the equipment, the observation degradation of each spatiotemporal analysis unit is analyzed. Observation degradation refers to the decline in the observation capability of the equipment in certain spatiotemporal analysis units, resulting in its inability to effectively monitor target areas or events. Absolute blind zones refer to areas or time periods that the equipment cannot observe. These areas are completely unable to acquire monitoring data due to limitations in the equipment's installation location, field of view, or imaging quality. Absolute blind zones are areas that cannot be effectively monitored under any conditions. Time-varying blind zones refer to areas where the equipment loses its observation capability during certain time periods. For example, a camera may capture images clearly during the day, but due to changes in lighting, the image may be blurry or unable to form a picture at night. Time-varying blind zones are caused by environmental changes or equipment performance degradation and are time-dependent. Identifying false blind zones refers to areas that the system considers to be blind zones, but in reality, these areas do not require monitoring or can be supplemented by other methods of observation. False blind zones are caused by data analysis errors or model assumptions.

[0034] By identifying absolute blind zones, time-varying blind zones, and false blind zones, we determine which areas and time periods experience degradation in observation capabilities. Target imaging quality relates to whether monitoring equipment can clearly capture target images under specific conditions. For example, factors such as equipment resolution, imaging algorithms, and the distance and size of the target affect imaging quality. In blind zones or degraded areas, target imaging is affected, thus requiring analysis of the equipment's imaging performance in these areas. Illumination variations are a significant factor affecting imaging quality. Equipment may experience image quality degradation during certain time periods, such as the transition between day and night, or under different weather conditions. Dynamic occlusion refers to the obstruction of the target's line of sight by moving objects or other environmental factors in the monitoring area. This occlusion affects the equipment's observation performance, leading to the loss or misjudgment of target information. By comprehensively analyzing target imaging quality, illumination variation trends, and dynamic occlusion patterns, we extract degradation characteristic curves for each spatiotemporal analysis unit. These curves reflect the monitoring degradation of the equipment under different conditions within a specific spatiotemporal analysis unit, providing a dynamic representation of the equipment's performance over time. To better simulate the degradation performance of monitoring equipment under various external conditions, mathematical models are used to describe this degradation process. For example, an exponential decay model is employed to describe physical phenomena such as changes in lighting and equipment performance degradation, showing rapid degradation over time or space. When the equipment's monitoring capabilities are affected by environmental factors such as lighting or weather, the degradation trend often manifests as a rapid decline, followed by a gradual stabilization. The model expression is: D(t,x) represents the degree of degradation of the spatiotemporal analysis unit (t,x) at time t and spatial location x, i.e., the observation capability of the device. λ represents the initial observation capability of the equipment under ideal conditions, λ is the degradation coefficient controlling the rate of decay, and f(t,x) is a function of external factors affecting equipment performance, representing the influence of environmental factors such as illumination and shading, or environmental factors related to time t and spatial location x, such as light intensity and the movement of obstructions. For example, the observation capability of the equipment will decline rapidly during the transition from day to night, especially in low-light environments. This decay process is represented by the exponential decay model described above.

[0035] The degradation characteristic curve provides the monitoring capability decline trend of the equipment under different environmental conditions. The spatiotemporal risk value reflects the risk level of abnormal events within a specific spatiotemporal analysis unit. Combining these two parameters, the actual monitoring effect of the equipment in a specific area and time period is evaluated. Risk exposure difference refers to the difference between the potential risk of a certain spatiotemporal analysis unit and the observation capability of the monitoring equipment. If the risk value of a certain spatiotemporal analysis unit is high, but the monitoring capability of the area is low, then the risk exposure difference of that spatiotemporal analysis unit is large. The risk exposure difference of all spatiotemporal analysis units is calculated to construct a risk exposure difference matrix. This matrix reflects the relationship between the monitoring effect and risk of each spatiotemporal analysis unit, revealing which areas and time periods have high risk exposure differences and require priority attention and enhanced monitoring.

[0036] Furthermore, by performing spatiotemporal analysis on real-time monitoring data from static monitoring equipment through an anomaly identification network, abnormal events can be identified, including: Real-time monitoring data from static monitoring equipment is mapped to corresponding spatiotemporal analysis units according to the spatial location and temporal state of the target. Based on the risk exposure difference corresponding to the spatiotemporal analysis unit, the mapped monitoring data is subjected to risk weighting processing to generate risk correlation features. Based on the risk correlation features, spatiotemporal correlation analysis is performed on the change process of the target behavior in continuous time and multiple spatial units to extract behavioral evolution features. Based on the behavioral evolution features and the risk correlation features, the decidability of each spatiotemporal analysis unit is quantified to generate a decidability index. The decidability index is compared with an anomaly threshold to identify the abnormal event.

[0037] Static monitoring equipment continuously generates real-time monitoring data, including images, video streams, and sensor readings, reflecting the equipment's observations within the monitored area. To map this data to spatiotemporal analysis units, the spatial location of the target (e.g., personnel, objects, vehicles) is first defined. This location information is then used to associate the target with its corresponding spatiotemporal analysis unit. Besides spatial location, the target's temporal state must also be defined. For example, a target may appear in different spatiotemporal analysis units within a specific time period. This temporal state includes the time of appearance, duration of activity, and interactions with other targets. Mapping real-time monitoring data to corresponding spatiotemporal analysis units based on the target's spatial location and temporal state organizes monitoring data from different times and spaces into specific units, ensuring the spatiotemporal correlation of the data and facilitating subsequent analysis and anomaly identification.

[0038] Risk exposure difference represents the difference between the potential risk of a spatiotemporal analysis unit and the observation capability of the equipment. After obtaining the risk exposure difference of a spatiotemporal analysis unit, the monitoring data is weighted according to this difference. That is, the monitoring data corresponding to different spatiotemporal analysis units are assigned different weights based on the magnitude of their risk exposure differences, with higher weights given to areas with larger risk exposure differences because they are more likely to experience anomalies. Risk correlation feature refers to the coupling relationship between target behavior and the risk environment of its spatiotemporal analysis unit. In other words, this feature not only describes the target's behavior, such as whether a person lingers in a certain area, but also combines the environmental conditions in which this behavior occurs, such as the high risk of this area during a specific time period. By coupling risk exposure difference with target behavior, we can better understand the environmental factors behind the behavior and further enhance the ability to identify and judge abnormal events. Through this feature, we can capture behaviors occurring in different risk environments and improve the accuracy of abnormal behavior judgment. This feature not only focuses on the behavior itself, but also on the specific risk environment in which the behavior occurs.

[0039] Based on the generated risk correlation characteristics, the analysis examines the changes in target behavior across different time and spatial units. The goal of spatiotemporal correlation analysis is to extract behavioral evolution characteristics that describe the changes in target behavior over time and space. Specifically, these characteristics include: temporal evolution characteristics, i.e., changes in target behavior over different time periods, such as dwell time, frequency of behavior, and movement speed; and spatial evolution characteristics, i.e., migration patterns of the target between different spatial units, such as whether the target moves along areas with low observation capability or high risk areas, or whether it avoids monitoring blind spots. Through spatiotemporal correlation analysis of target behavior, a better understanding of the behavioral evolution process can be achieved, which helps in determining the anomalies of target behavior, as abnormal behavior often exhibits certain obvious temporal or spatial patterns.

[0040] Determinability refers to the reliability of anomaly determination of target behavior under current spatiotemporal conditions. Combining behavioral evolution characteristics and risk correlation characteristics, the determinability of each spatiotemporal analysis unit is quantified using the following formula: , It is the decidability index of the spatiotemporal analysis unit (t,x). It is the quantification result of the evolution characteristics of target behavior on the spatiotemporal analysis unit (t,x), reflecting the change pattern of target behavior. It is the risk correlation feature corresponding to the spatiotemporal analysis unit (t,x), reflecting the situation where the target behavior occurs in high-risk areas and time periods. and The weighting coefficient represents the relative importance of behavioral evolution characteristics and risk correlation characteristics to the decidability index. For example, if the target behavior in a spatiotemporal analysis unit exhibits a clear anomalous pattern and the risk exposure difference in that area is large, then the decidability of that unit is high. The decidability index is an indicator that measures whether a target behavior can be accurately identified as an anomalous event. This index is derived through quantitative analysis of behavioral evolution characteristics and risk correlation characteristics; a higher value indicates that the target behavior is more easily identified as anomalous.

[0041] The system compares the determinable index with a preset anomaly threshold. If the determinable index exceeds the threshold, the target behavior is considered to meet the characteristics of an anomalous event, and the target behavior is marked as an anomalous event. The threshold can be adjusted based on historical data and risk assessment to ensure that anomalous events can be identified within a reasonable time and space.

[0042] Furthermore, based on the aforementioned risk correlation characteristics, spatiotemporal correlation analysis is performed on the change process of the target behavior in continuous time and multiple spatial units to extract behavioral evolution characteristics, including: Based on the aforementioned risk correlation characteristics, the behavioral states of the same target over a continuous time period are analyzed to extract dwell time, frequency of behavioral changes, and trajectory change characteristics, thereby determining the target's temporal evolution characteristics. Correlation analysis is also performed on the target's migration paths across multiple spatiotemporal analysis units to identify whether it moves along areas with low observation capabilities or blind zone chains formed by pseudo-blind zones, thus determining the target's spatial avoidance characteristics. Finally, the temporal evolution characteristics and spatial avoidance characteristics are fused to generate the spatiotemporal evolution characteristics of the target's behavior.

[0043] Analyzing the target's behavioral state over consecutive time periods means tracking changes in the target's state across different time intervals. Dwell time refers to the length of time a target remains within a specific area. Excessive dwell time indicates abnormal behavior, especially when the target is in a high-risk area. Dwell time is extracted from monitoring data within each spatiotemporal analysis unit. Behavioral change frequency refers to the frequency with which the target's behavioral state changes. For example, does the target frequently change position, dwell time, or trajectory? Frequent and significant behavioral changes indicate abnormal behavior. Trajectory change characteristics describe the target's movement path and trajectory. Analyzing whether the trajectory is smooth and whether there are sudden changes, and abnormal trajectory changes such as irregular trajectories or bypassing certain areas, indicates abnormal behavior. By analyzing dwell time, behavioral change frequency, and trajectory changes, the target's temporal evolution characteristics are extracted, reflecting its behavioral patterns over time.

[0044] By tracking the spatial location of a target, its movement trajectory is plotted, and its migration between different spatial units is analyzed. The target's migration path reveals whether it is deliberately evading the coverage area of ​​surveillance equipment. For example, the target may move along areas with low surveillance capability or areas identified as false blind spots. These areas have weak surveillance capabilities, and the target may be consciously using these areas for evasion. When a target moves between low-observation-capability areas or false blind spots, a blind spot chain is formed. This behavioral pattern indicates that the target is deliberately avoiding highly monitored areas and using blind spots to hide. In this case, the spatial evasion characteristics of the target's behavior are very significant. By analyzing the target's migration path, it is possible to identify whether the target exhibits obvious evasion behavior, such as bypassing specific areas or only operating within specific areas. Spatial evasion characteristics can further indicate abnormal behavior of the target, especially those cases of deliberate evasion of surveillance.

[0045] By integrating temporal evolution characteristics with spatial avoidance characteristics, this integration process combines the target's behavioral patterns in both time and space dimensions to generate more comprehensive spatiotemporal evolution characteristics. These characteristics can fully describe the target's behavioral patterns, including its performance in both time and space dimensions.

[0046] Furthermore, before deploying unmanned vehicles and / or drones to perform verification tasks according to spatiotemporal observation data compensation requirements, the following steps are also included: The monitoring data corresponding to the abnormal event is evaluated to determine whether the current monitoring data meets the preset sufficient conditions. If the preset sufficient conditions are not met, the information compensation conditions are determined based on the risk exposure difference and behavioral characteristics corresponding to the abnormal event, and the collaborative interference factor is located. Based on the difference characteristics before and after the disturbance of the collaborative interference factor, the verification value index is evaluated in combination with the abnormal event. When the verification value index reaches the requirements of panoramic verification, unmanned vehicles and drones are matched based on the compensation needs of spatiotemporal observation data. Using the matching results, unmanned vehicles and / or drones are called to perform the verification task.

[0047] Following an anomaly, the initial assessment of monitoring data collected from static monitoring equipment is crucial to ensure its sufficiency in identifying the anomaly. If the data is incomplete, unclear, or insufficient to accurately identify the anomaly, supplementary data is required. Pre-defined sufficient conditions refer to standards ensuring the accuracy and completeness of the monitoring data. If these conditions are not met, the current monitoring data is considered inadequate to effectively support the anomaly assessment. These conditions include data clarity and spatiotemporal coverage. When the monitoring data does not meet the pre-defined sufficient conditions, supplementary data is determined based on the risk exposure difference and behavioral characteristics of the anomaly. A larger risk exposure difference indicates a higher probability of anomalies occurring in the area, thus requiring more data to compensate for the deficiencies in the original monitoring.

[0048] The difference characteristics before and after disturbance refer to the changes in monitoring data with and without the presence of cooperative interference factors. For example, a weather change might cause image blurring; analyzing the differences before and after this change clarifies the impact of the interference factors. Combining the disturbance characteristics of cooperative interference factors, a value index is assigned to the verification task based on data quality assessment. This verification value index reflects the importance and urgency of supplementing the data. For example, if an anomaly occurs in a high-risk area and there are significant gaps in the monitoring data, a high verification value index is given, indicating the need for immediate verification. The assessment of the verification value index is based on data quality, risk exposure difference, and the impact of interference factors.

[0049] Once the verification value index is assessed, if the index exceeds a preset threshold, it indicates that the verification task for the current anomaly event has sufficient urgency and importance. At this point, it is determined whether a panoramic verification task needs to be initiated. Spatiotemporal observation data compensation requirements refer to gaps or blind spots in the current monitoring data, requiring supplementary data from unmanned vehicles (UAVs) and drones. The different advantages of UAVs and drones enable them to effectively compensate for monitoring blind spots in specific scenarios. UAVs are suitable for acquiring close-range behavioral characteristics of targets, such as specific activities and behavioral details; drones are suitable for acquiring high-level spatial distribution information of target areas, such as monitoring large areas and global observation of targets. Based on the spatiotemporal observation data compensation requirements, suitable UAVs and / or drones are matched, and scheduling is performed based on the matching results to execute the verification task. This process ensures that the verification task can be completed in the shortest possible time and acquire sufficient data to supplement the original monitoring data, improving the accuracy and processing efficiency of anomalies.

[0050] Furthermore, by fusing multi-source data during the verification process, a comprehensive verification result is established, including: The spatiotemporal observation data compensation requirements include compensation for blind spots in the target area and detailed observation of target behavior. The unmanned vehicle is used to acquire close-range behavioral feature information of the target; the drone is used to acquire high-level spatial distribution information of the target area; spatiotemporal alignment and fusion are performed based on the data acquired by the unmanned vehicle and the drone to obtain fused compensation data; the fused compensation data is used to verify the consistency of static monitoring data and complete information to generate the panoramic verification result.

[0051] Monitoring systems have blind spots, which cannot be effectively observed by static monitoring equipment. The need for spatiotemporal observation data compensation primarily addresses how to fill these data gaps. Target areas may be subject to blind spots due to the equipment's field of view, equipment failure, or environmental factors; therefore, data compensation is necessary for these blind spots. Besides blind spots, static monitoring equipment has limitations in capturing detailed target behavior. For example, static cameras cannot capture fine behavioral characteristics such as subtle movements, facial expressions, or close-range interactions. Therefore, supplementary behavioral detail data is needed to comprehensively understand target behavior patterns. Autonomous vehicles (RVs) are used to acquire close-range behavioral characteristics of targets. Because RVs can approach target areas, they can acquire fine behavioral data such as gait, movement frequency, and interactions with the surrounding environment. The high mobility of RVs allows them to cover areas inaccessible to static monitoring equipment. Drones are used to acquire high-altitude spatial distribution information of target areas. The high-altitude perspective of drones helps to acquire broader spatial information, especially in large areas or environments with tall buildings where static monitoring equipment cannot fully cover. Drones supplement the blind spots of static equipment by taking pictures or collecting data from the air.

[0052] Since autonomous vehicles (RVs) and drones collect data at different times, their data needs to be synchronized to a unified timescale. This includes converting timestamps in the data to a uniform format or using interpolation techniques to fill in the time differences. Because RVs and drones have different perspectives and spatial positions, it's crucial to ensure that the data they acquire is aligned with the actual spatial coordinates of the target area. For example, spatially aligning close-range target information acquired by RVs with aerial view data provided by drones ensures data consistency. After completing spatiotemporal alignment, the data from RVs and drones can be fused to generate fused and compensated data.

[0053] Static monitoring data includes data collected by devices such as cameras and sensors. By comparing and fusing compensated data with static monitoring data, the accuracy and completeness of the static monitoring data are verified. If there are missing or inconsistent static monitoring data, supplementary data is used for correction and repair. Using the fused data, a panoramic verification result is generated. The panoramic verification result is a comprehensive analysis result integrating static monitoring equipment data and dynamic supplementary data, providing a comprehensive view of the target area.

[0054] Furthermore, based on the spatiotemporal observation data compensation requirements for the aforementioned abnormal event location, unmanned vehicles and / or drones are invoked to perform verification tasks according to the spatiotemporal observation data compensation requirements, including: Based on the event attributes of the abnormal events and the corresponding risk exposure differences, the verification requirements are analyzed to determine spatial coverage requirements and time response requirements. According to the spatial coverage requirements and time response requirements, and combined with the monitoring status of unmanned vehicles and drones, execution strategies are searched to match the drone attributes and quantity distribution constraints. The matching execution strategies are then used to call the corresponding unmanned vehicles and / or drones to perform the verification tasks.

[0055] Based on the event attributes and risk exposure differences of anomalous events, spatial coverage requirements are determined. These requirements ensure comprehensive monitoring of the target area during the verification process, including coverage of blind spots and areas with low observation capabilities, to ensure effective monitoring of all potentially anomalous areas. Time response requirements specify the timeframe required for verification and response after an event occurs. Different types of anomalous events require different response times; for example, high-risk events require a response within minutes, while low-risk events can tolerate longer response times.

[0056] Determining the monitoring status of unmanned vehicles and drones, including their battery level, current location, and availability, is crucial. These factors all influence the execution of the verification task; therefore, an execution strategy search must be conducted based on these factors. A key task in this strategy search is matching drone attributes, including their flight capabilities (altitude, speed, flight time, etc.) and their observation capabilities within the target area (camera resolution, sensor type, etc.). Based on these attributes, suitable drones can be selected to perform the verification task. The number and distribution of drones are also important. For example, for a large monitoring area, multiple drones may be needed to perform the task simultaneously; therefore, the appropriate number of drones must be calculated to ensure effective coverage of the entire target area.

[0057] To achieve execution strategy search, an optimal match is performed between given spatiotemporal observation data compensation requirements, the monitoring status of unmanned vehicles / drones, and the coverage requirements of the target area. First, based on the attributes of abnormal events and risk exposure differences, the spatial coverage requirements and time response requirements of each task are determined. Then, combined with the flight capabilities of the drone, such as flight altitude, speed, and battery level, as well as availability, tasks are allocated. Optimization search is performed using genetic algorithms or particle swarm optimization algorithms, and the optimal solution is found through crossover and mutation operations. The fitness function is used to evaluate the merits of each strategy, including factors such as task coverage, battery usage, and time response. Finally, the optimal execution strategy is output to ensure efficient task execution and reasonable resource allocation.

[0058] According to the matching execution strategy, appropriate unmanned vehicles and / or drones are dispatched to perform verification tasks. This process includes assigning tasks to qualified unmanned equipment and ensuring that they can reach the target area in a timely manner for verification.

[0059] Furthermore, it also includes: Based on the panoramic verification results, the effective observation capability and risk exposure gap in the anomaly identification network are corrected and updated; based on the updated anomaly identification network, the scheduling of subsequent anomaly event identification and verification tasks is optimized and adjusted.

[0060] Based on the panoramic verification results, the effective observation capabilities in the anomaly identification network are updated. This includes checking the actual performance of monitoring equipment under specific spatiotemporal conditions, especially after data supplementation in areas previously considered blind spots or with low observation capabilities, where the equipment's observation capabilities are improved. The monitoring capabilities of these areas are then reassessed and updated based on the newly acquired data. Simultaneously, based on the panoramic verification results, the risk exposure difference matrix is ​​updated. With the supplementation of monitoring data, areas previously assessed as having high risk exposure differences may be reassessed as lower-risk areas, or vice versa. The panoramic verification results allow for a more accurate assessment of risk exposure differences, thereby improving the accuracy of anomaly identification.

[0061] Based on the updated anomaly detection network, existing anomaly detection strategies are reassessed and adjusted. The updated network better reflects real-world monitoring capabilities and risk exposure gaps, making anomaly detection more accurate and efficient. With the optimization of the anomaly detection network, subsequent anomaly events can be identified more quickly and accurately. The anomaly detection algorithm is adjusted according to these changes to ensure accurate identification of potential abnormal behaviors under different environmental and spatiotemporal conditions. After accurate anomaly event identification, subsequent verification tasks are scheduled and optimized based on the updated network, ensuring timely and accurate handling of potential security issues in changing environments.

[0062] Example 2, Figure 2 This is a schematic diagram of the electronic device provided by the present invention based on the unmanned vehicle and drone linkage method for panoramic on-site verification of banks, and a block diagram of an exemplary electronic device suitable for implementing the embodiments of the present invention. Figure 2 The electronic device shown is merely an example and should not be construed as limiting the functionality or scope of the embodiments of the present invention. Figure 2 As shown, the electronic device includes a processor 21, a memory 22, an input device 23, and an output device 24; the number of processors 21 in the electronic device can be one or more. Figure 2 Taking a processor 21 as an example, the processor 21, memory 22, input device 23, and output device 24 in an electronic device can be connected via a bus or other means. Figure 2 Taking the example of a connection between China and Israel via a bus.

[0063] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0064] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for conducting on-site panoramic verification of banks based on the linkage of unmanned vehicles and drones, characterized in that: The method includes: Analyze the distribution characteristics of static monitoring equipment and, in conjunction with the spatiotemporal risks at the bank site, establish an anomaly identification network; Anomaly identification networks are used to perform spatiotemporal analysis on real-time monitoring data from static monitoring equipment to identify abnormal events. Based on the spatiotemporal observation data compensation requirements for the aforementioned abnormal events, unmanned vehicles and / or drones are invoked to perform verification tasks according to the spatiotemporal observation data compensation requirements. The multi-source data during the verification process is integrated to establish a comprehensive verification result, and corresponding handling procedures and data archiving are triggered.

2. The method for on-site panoramic verification of banks based on unmanned vehicles and drones according to claim 1, characterized in that, Analyze the distribution characteristics of static monitoring equipment, and combine this with the spatiotemporal risks at the bank site to establish an anomaly identification network, including: The bank site is spatially and temporally discretized to construct multiple spatiotemporal analysis units; Extract the functional attributes, behavioral triggering attributes, and historical anomaly attributes of each spatiotemporal analysis unit to determine the corresponding spatiotemporal risk value; The position parameters, field of view parameters, imaging parameters and environmental adaptation parameters of each static monitoring device are analyzed to establish an effective observation matrix of the device for each spatiotemporal analysis unit; Based on the spatiotemporal risk value and the effective observation matrix, a risk exposure difference matrix is ​​established, and an anomaly identification network is constructed based on the risk exposure difference matrix.

3. The method for on-site panoramic verification of banks based on unmanned vehicles and drones according to claim 2, characterized in that, Based on the spatiotemporal risk values ​​and the effective observation matrix, a risk exposure difference matrix is ​​established, including: Based on the effective observation matrix, a time-series statistical analysis of the observation capabilities of each spatiotemporal analysis unit at different time periods is performed to construct the equipment observation stability distribution. By combining historical anomaly identification results, the identification success rate, false alarm rate and false negative rate of each spatiotemporal analysis unit are reverse-mapped to generate an identification credibility distribution. Based on the observation stability distribution and the identification confidence distribution, the spatiotemporal analysis unit is subjected to observation degradation analysis to identify absolute blind zones, time-varying blind zones and pseudo-blind zones. Based on the absolute blind zone, time-varying blind zone, and identification pseudo-blind zone, the target imaging quality, illumination change trend, and dynamic occlusion mode are jointly analyzed, and the degradation characteristic curves of each spatiotemporal analysis unit are extracted. Based on the degradation characteristic curve and the spatiotemporal risk value, the risk exposure difference of each spatiotemporal analysis unit is calculated, and the risk exposure difference matrix is ​​constructed.

4. The method for on-site panoramic verification of banks based on unmanned vehicles and drones according to claim 3, characterized in that, Anomaly detection networks are used to perform spatiotemporal analysis on real-time monitoring data from static monitoring equipment to identify anomalous events, including: The real-time monitoring data from static monitoring equipment is mapped to the corresponding spatiotemporal analysis unit according to the spatial location and temporal status of the target. Based on the risk exposure difference corresponding to the spatiotemporal analysis unit, the mapped monitoring data is subjected to risk weighting processing to generate risk association features. Based on the aforementioned risk correlation characteristics, spatiotemporal correlation analysis is performed on the change process of target behavior in continuous time and multiple spatial units to extract behavioral evolution characteristics. Based on the behavioral evolution characteristics and the risk correlation characteristics, the determineability of each spatiotemporal analysis unit is quantified to generate a determineability index. The abnormal event is identified by comparing the determinable index with the abnormal threshold.

5. The method for on-site panoramic verification of banks based on unmanned vehicles and drones according to claim 4, characterized in that, Based on the aforementioned risk correlation characteristics, a spatiotemporal correlation analysis is performed on the change process of the target behavior in continuous time and multiple spatial units to extract behavioral evolution characteristics, including: Based on the aforementioned risk correlation characteristics, the behavioral states of the same target within a continuous time period are analyzed to extract dwell time, frequency of behavioral changes, and trajectory change characteristics, thereby determining the temporal evolution characteristics of the target. Correlation analysis is performed on the migration path of the target among multiple spatiotemporal analysis units to identify whether it moves along blind zone chains formed by low observation capability areas or false blind zones, and to determine the spatial avoidance characteristics of the target. By fusing the temporal evolution characteristics with the spatial avoidance characteristics, the spatiotemporal evolution features of the target behavior are generated.

6. The method for on-site panoramic verification of banks based on unmanned vehicles and drones according to claim 4, characterized in that, Before deploying unmanned vehicles and / or drones to perform verification tasks according to spatiotemporal observation data compensation requirements, the following steps are also included: The monitoring data corresponding to the abnormal event is evaluated to determine whether the current monitoring data meets the preset sufficient conditions. If the preset sufficient conditions are not met, the information compensation conditions are determined based on the risk exposure difference and behavioral characteristics corresponding to the abnormal event, and the collaborative interference factor is located. Based on the differences in characteristics before and after disturbance of the collaborative interference factor, and combined with abnormal events, the verification value index is evaluated. When the verification value index meets the requirements of panoramic verification, unmanned vehicles and drones are matched based on the compensation needs of spatiotemporal observation data. Using the matching results, unmanned vehicles and / or drones are called to perform verification tasks.

7. The method for on-site panoramic verification of banks based on unmanned vehicles and drones according to claim 3, characterized in that, The multi-source data from the verification process are integrated to establish a comprehensive verification result, including: The spatiotemporal observation data compensation requirements include the need to compensate for blind spots in the target area and the need to observe detailed target behavior. The unmanned vehicle is used to acquire close-range behavioral feature information of the target; the drone is used to acquire high-level spatial distribution information of the target area. Based on the data acquired by the unmanned vehicle and the drone, spatiotemporal alignment and fusion are performed to obtain fused compensation data; The fused compensation data is used to verify the consistency of the static monitoring data and complete the information, thereby generating the panoramic verification result.

8. The method for on-site panoramic verification of banks based on unmanned vehicles and drones according to claim 7, characterized in that, Based on the spatiotemporal observation data compensation requirements for the aforementioned abnormal event location, unmanned vehicles and / or drones are invoked to perform verification tasks according to the spatiotemporal observation data compensation requirements, including: Based on the event attributes of the abnormal events and the corresponding risk exposure differences, the verification requirements are analyzed to determine the spatial coverage requirement constraints and the time response requirement constraints. Based on the spatial coverage requirement constraints and time response requirement constraints, and combined with the monitoring status of unmanned vehicles and drones, an execution strategy search is performed to match the drone attributes and quantity distribution constraints. The matching execution strategy is used to invoke the corresponding unmanned vehicle and / or drone to perform the verification task.

9. The method for on-site panoramic verification of banks based on unmanned vehicles and drones according to claim 2, characterized in that, Also includes: Based on the panoramic verification results, the effective observation capability and risk exposure deficit in the anomaly identification network are corrected and updated. Based on the updated anomaly detection network, the scheduling of subsequent anomaly event identification and verification tasks is optimized and adjusted.

10. An electronic device, characterized in that, The electronic device includes: processor; Memory used to store the processor's executable instructions; The processor is used to execute the on-site panoramic verification method for banks based on unmanned vehicles and drones, as described in any one of claims 1 to 9.

Images