Smart city public safety management system based on video monitoring

Abstract

The invention discloses a smart city public safety management system based on video monitoring, and relates to the technical field of public safety management, original video data streams and a real-time road data set D are collected in real time through an API interface, a time sequence feature matrix TSM and a road congestion coefficient Ck are generated, and the data integration efficiency is improved. The behavior feature extraction module identifies group behavior features through target detection and an optical flow algorithm, and encryption transmission guarantees privacy. The risk prediction module calculates a cross-regional risk index RRI based on federated learning, triggers a response mechanism and accurately early warns risks, and the dynamic decision module combines real-time data to generate a resource scheduling scheme and selects a proper path and callable resources. Through the combination of the execution feedback module and the driving iterative optimization module, the algorithm is dynamically adjusted, closed-loop management is formed, the problems of data splitting, response lag and model stiffness of a traditional system are solved, and the real-time performance, the accuracy and the scene adaptability of public safety management are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of public safety management, and specifically to an intelligent city public safety management system based on video surveillance. Background Art

[0002] With the deepening of the construction of intelligent cities, the video surveillance network has become the core infrastructure for urban safety management, covering scenarios such as transportation hubs, commercial areas, and communities, and becoming the core perception layer of public safety management. With the abundance of material life, the flow of people in public places is increasing day by day. How to efficiently manage the safety of the crowd, effectively link cross-regional video collection and analysis, and build an effective public safety prediction model and management system has become an urgent task.

[0003] Currently, the linkage management systems for intelligent cities and video surveillance data mostly focus on video collection and simple behavior recognition in a single area, such as the out-of-bounds detection of pedestrians or vehicles running red lights. First, for risk prediction, it relies on manual experience and static rules, lacks quantitative modeling of group behavior states such as moving speed, direction chaos degree, and abnormal behavior index, and has a lag in response decision-making. Second, the existing systems lack the global analysis ability for regional risk prediction and management, and the data of cross-regional cameras cannot be integrated. When there are potential safety hazards in the flow of people in a certain area, it is impossible to give early warnings about the moving directions of people in other areas. Finally, the resource scheduling is rigid, and the emergency plan cannot be dynamically adjusted according to real-time risks, often resulting in regional response delays or resource redundancy in low-risk areas. These defects have led to the failure to form a full-process management framework of "data integration → dynamic modeling → resource scheduling → closed-loop optimization" in the city, resulting in the inability to further improve the prediction and prevention efficiency of potential safety hazards in the urban flow of people. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides an intelligent city public safety management system based on video surveillance, which solves the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: An intelligent city public safety management system based on video surveillance, characterized in that: it includes a data collection and analysis module, a behavior feature extraction module, a risk prediction module, a dynamic decision-making module, an execution feedback module, and an iterative optimization module;

[0006] The data collection and analysis module obtains the time series feature matrix TSM and the road congestion coefficient Ck by collecting the original video data stream and the real-time road data set D in real time from the public video surveillance terminals in each area of the city, the urban traffic management platform, and vehicle-mounted GPS through the API interface;

[0007] The behavior feature extraction module extracts and encrypts behavior features from the time series feature matrix TSM at each local node, generates a set of homomorphically encrypted real-time behavior feature data FV, and transmits the set of homomorphically encrypted real-time behavior feature data FV and the road congestion coefficient Ck to the central node in ciphertext form;

[0008] The risk prediction module receives the encrypted real-time behavior feature data set FV from each node, establishes a cross-regional risk prediction algorithm through federated learning, calculates the cross-regional risk index RRI, and compares the cross-regional risk index RRIk of region k with the preset risk threshold range T to execute the safety management plan;

[0009] The dynamic decision-making module establishes a dynamic resource scheduling algorithm based on the cross-regional risk index RRIk and the road congestion coefficient Ck of region k and generates a resource scheduling plan to be sent to the duty unit;

[0010] The execution feedback module obtains the feedback effect evaluation and generates a feedback data set F by quantifying the execution effect of the resource scheduling plan;

[0011] The iterative optimization module optimizes the cross-regional risk prediction algorithm and the dynamic resource scheduling algorithm based on the feedback data set F.

[0012] Preferably, the data acquisition and analysis module includes a data stream acquisition unit and a data analysis unit;

[0013] The data stream acquisition unit accesses the public video monitoring terminals in each region of the city through API calls to be compatible with the communication protocols of different terminals and obtains the original video stream;

[0014] Through API calls to the urban traffic management platform and vehicle-mounted GPS data, obtain the maximum traffic flow max(Tf), real-time traffic flow RTf, average historical vehicle speed Hv, and average real-time vehicle speed Rv of the roads within the jurisdiction of the monitoring, and obtain a set of real-time road data D.

[0015] Preferably, the data analysis unit analyzes the original video stream and the set of real-time road data D to obtain the time series feature matrix TSM and the road congestion coefficient Ck;

[0016] Use the YOLOv8 object detection algorithm to identify the person bounding box Bi frame by frame in the original video stream, count the number of people per unit area, calculate the crowd density CD in the video area, use an improved optical flow algorithm to track the displacement of people, and calculate the individual movement speed vector V;

[0017] Integrate the crowd density CD and the individual movement speed vector V in the original video stream to obtain the time series feature matrix TSM of the video area;

[0018] Based on the analysis of the real-time road data set D, the road congestion coefficient Ck is obtained, and the analysis expression of the road congestion coefficient Ck is as follows:

[0019] ;

[0020] In the formula, η represents the contribution weight coefficient of the ratio of the real-time traffic flow RTf of the road to the maximum traffic flow max(Tf) of the road, and ι represents the contribution weight coefficient of the ratio of the average driving speed Rv of the real-time vehicles on the road to the average driving speed Hv of the historical vehicles on the road.

[0021] Preferably, the behavior feature extraction module includes a behavior feature analysis unit and a data processing unit;

[0022] The behavior feature analysis unit extracts and encrypts behavior features from the time series feature matrix TSM at each local node, and converts the time series feature matrix TSM into a crowd density distribution coefficient CDM, a moving direction consistency coefficient MDC, and an abnormal behavior coefficient ABFV;

[0023] Among them, the analysis expression of the crowd density distribution coefficient CDM is as follows:

[0024] ;

[0025] In the formula, (x, y) represents the video pixel coordinates, (xi, yi) represents the human body coordinates of the i-th person detected, N represents the total number of people detected, π represents the constant 3.14, σ represents the preset Gaussian kernel bandwidth, and e represents the natural constant;

[0026] By analyzing the individual movement speed vector V, the individual movement direction angle θ is obtained, and the analysis expression of the moving direction consistency coefficient MDC is as follows:

[0027] ;

[0028] In the formula, θi represents the moving direction angle of the i-th person, and θj represents the moving direction angle of the j-th person, and j ≠ i;

[0029] Based on the personnel bounding box Bi, the number of people with squatting and lying down behaviors in the surveillance video is analyzed. Based on the individual movement speed vector V, the number of people with stationary and running behaviors in the video is analyzed. Based on the proportion of the number of people with squatting, lying down, stationary, and running behaviors in the surveillance video to the total number of people N detected, the abnormal behavior coefficient ABFV is obtained;

[0030] Among them, the detection criteria for squatting and lying down behaviors are as follows:

[0031] If the current height of the personnel bounding box Bi ≤ 0.5 times the original height of the personnel bounding box Bi, and the current width of the personnel bounding box Bi ≥ 1.25 times the original width of the personnel bounding box Bi, then it is determined that the current detection target is in a squatting state;

[0032] If the current height of the personnel bounding box Bi > 0.5 times the original height of the personnel bounding box Bi, and the current width of the personnel bounding box Bi < 1.25 times the original width of the personnel bounding box Bi, then it is determined that the current detection target is in a squatting state;

[0033] If the current height of the personnel bounding box Bi ≤ 0.25 times the original height of the personnel bounding box Bi, and the current width of the personnel bounding box Bi ≥ 1.5 times the original width of the personnel bounding box Bi, then it is determined that the current detection target is in a lying state;

[0034] If the current height of the personnel bounding box Bi > 0.25 times the original height of the personnel bounding box Bi, and the current width of the personnel bounding box Bi < 1.5 times the original width of the personnel bounding box Bi, then it is determined that the current detection target is not in a lying state;

[0035] The detection criteria for static and running behaviors are as follows:

[0036] If the individual movement speed vector V = 0 m / s, then it is determined that the current detection target is in a static state;

[0037] If the individual movement speed vector V ≠ 0 m / s, then it is determined that the current detection target is not in a static state;

[0038] If the individual movement speed vector V > 2.5 m / s, then it is determined that the current detection target is in a running state;

[0039] If the individual movement speed vector V ≤ 2.5 m / s, then it is determined that the current detection target is not in a running state.

[0040] Preferably, the data processing unit vertically merges the encrypted crowd density distribution coefficient CDM, the movement direction consistency coefficient MDC, and the abnormal behavior coefficient ABFV to generate a real-time behavior feature data set FV, homomorphically encrypts the obtained real-time behavior feature data set FV and the road congestion coefficient Ck, and transmits them to the central node in ciphertext form.

[0041] Preferably, the risk prediction module includes a risk modeling unit and an execution unit;

[0042] The risk modeling unit receives the encrypted real-time behavior feature data set FV from each node, designs a cross-regional risk prediction algorithm through federated learning based on the data in the real-time behavior feature data set FV, and obtains the cross-regional risk index RRIk of region k;

[0043] Among them, the expression of the cross-regional risk prediction algorithm is as follows:

[0044] ;

[0045] In the formula, k represents the region, RRIk represents the cross-regional risk index of region k, max(CDMk) represents the maximum value of the crowd density distribution coefficient CDM in region k, a represents the set density sensitivity coefficient, b represents the set curve density threshold, α represents the contribution weight coefficient of the maximum value max(CDMk) of the crowd density distribution coefficient in region k, MDCk represents the moving direction consistency coefficient of region k, β represents the contribution weight coefficient of the moving direction consistency coefficient MDCk of region k, ABFVk represents the abnormal behavior coefficient of region k, and γ represents the contribution weight coefficient of the abnormal behavior coefficient ABFVk of region k.

[0046] Preferably, the execution unit compares the cross-regional risk index RRIk of region k with the risk threshold range T, and executes a hierarchical response plan according to the comparison result;

[0047] If the cross-regional risk index RRIk of region k < the risk threshold Tmin, only record the current data as historical reference data;

[0048] If the risk threshold Tmin ≤ the cross-regional risk index RRIk of region k < the risk threshold Tmax, trigger a risk warning and push it to the duty units in the jurisdiction where region k is located and the public electronic screens in the nearby areas, and advise the people and vehicles that have not entered region k to choose their routes carefully;

[0049] If the cross-regional risk index RRIk of region k ≥ the risk threshold Tmax, trigger a risk warning, generate a prompt on the public electronic screens in the nearby areas, warn the people and vehicles that have not entered region k to avoid entering region k, and trigger a dynamic resource scheduling algorithm to automatically generate a resource scheduling plan and send it to the duty units in the jurisdiction where it is located;

[0050] Among them, the risk threshold Tmin represents the lower limit value in the risk threshold range T, and the risk threshold Tmax represents the upper limit value in the risk threshold range T.

[0051] Preferably, the dynamic decision-making module includes a dynamic resource scheduling decision-making unit;

[0052] The dynamic resource scheduling decision-making unit designs a dynamic resource scheduling algorithm based on the cross-regional risk index RRIk of region k and the road congestion coefficient Ck, and generates a preferred resource scheduling value PD;

[0053] Obtain the path n of the duty units in the jurisdiction where region k is located to reach region k from the urban traffic management platform by calling the API, and form a real-time candidate path set p by integrating the path n;

[0054] The expression of the dynamic resource scheduling algorithm is as follows:

[0055] ;

[0056] In the formula, argmin represents the parameter for finding the minimum value of the objective function, pn represents the nth path in the candidate path set p, Spnk represents the Manhattan distance S from the nth path in the candidate path set p to area k, Rpnm represents the Manhattan distance R from the available resource unit m on the nth path in the candidate path set p to area k, t represents the time difference from the start of the risk warning to the current moment, δ represents the path efficiency contribution weight coefficient of the nth path in the candidate path set p reaching area k, w represents the contribution weight coefficient of the Manhattan distance R from the available resource unit m on the nth path in the candidate path set p to area k, and μ represents the contribution weight coefficient of the time difference t from the start of the risk warning to the current moment;

[0057] Derive the parameters when the resource scheduling optimization value PD is at its minimum and generate a resource scheduling plan. The content of the plan includes the nth path in the selected candidate path set p and the available resource unit m called from the nth path in the selected candidate path set p, and send the resource scheduling plan to the on-duty units within the jurisdiction.

[0058] Preferably, the execution feedback module includes a feedback data collection unit;

[0059] The feedback data collection unit generates a feedback data set F by collecting the execution feedback data of the resource scheduling plan, and quantifies the execution effect of the resource scheduling plan;

[0060] Among them, the feedback data set F includes the response time RT, the resource utilization rate RC, and the risk prediction accuracy rate SR;

[0061] The response time RT represents the time difference calculated by the available resource unit m based on the timestamp obtained by the GPS device from the time when the resource scheduling plan is sent to the time of arrival at the scene;

[0062] The resource utilization rate RC represents the ratio of the actual number of resources arriving at the scene to the predicted number of required resources;

[0063] The risk prediction accuracy rate SR represents the ratio of the number of correct risk predictions to the total number of risk predictions.

[0064] Preferably, the iterative optimization module includes an iterative optimization unit;

[0065] The iterative optimization unit dynamically optimizes the cross-regional risk prediction algorithm and the dynamic resource scheduling algorithm based on the feedback data set F;

[0066] The optimization scheme of the cross-regional risk prediction algorithm is as follows:

[0067] Based on the risk prediction accuracy rate SR in the feedback data set F, if the risk prediction accuracy rate SR ≥ 95%, it is determined that the prediction result is reasonable, and there is no need to dynamically adjust the contribution weight coefficients α, β, and γ in the cross-regional risk prediction algorithm;

[0068] If the risk prediction accuracy rate SR < 95%, it is determined that the prediction result is unreasonable, and the contribution weight coefficients α, β, and γ in the cross-regional risk prediction algorithm are dynamically adjusted;

[0069] The optimization scheme of the dynamic resource scheduling algorithm is as follows:

[0070] Based on the response time RT and resource utilization rate RC in the feedback data set F;

[0071] Optimization direction 1: If the single response time RT ≤ 15 minutes, it is determined that the nth path in the selected candidate path set p is reasonable, and there is no need to dynamically adjust the contribution weight coefficients δ, w in the dynamic resource scheduling algorithm;

[0072] If the single response time RT > 15 minutes, it is determined that the nth path in the selected candidate path set p is unreasonable, and the contribution weight coefficients δ, w in the dynamic resource scheduling algorithm are dynamically adjusted;

[0073] Optimization direction 2: If the continuous three - time resource utilization rate RC ≥ 80%, it is determined that the selected available resource unit m is reasonable, and there is no need to dynamically adjust the contribution weight coefficients δ, w in the dynamic resource scheduling algorithm;

[0074] If there is one time when the resource utilization rate RC < 80% within three consecutive times, it is determined that the selected available resource unit m is unreasonable, and the contribution weight coefficients δ, w in the dynamic resource scheduling algorithm are dynamically adjusted.

[0075] The present invention provides a smart city public security management system based on video surveillance, which has the following beneficial effects:

[0076] (1)When the system is running, it accesses the time monitoring terminals in various regions of the city, the traffic management platform, and in-vehicle GPS data through the API, and obtains the original video stream, road traffic flow, and vehicle speed data in real time. It constructs the time series feature matrix TSM and the road congestion coefficient Ck, integrates and analyzes multi-source data, and uses homomorphic encryption technology to ensure the security of sensitive information transmission, providing multi-dimensional reliable data for subsequent analysis, and solving the problems of data islands and data security risks in traditional systems. Based on the time series feature matrix TSM, the YOLOv8 object detection algorithm and the improved optical flow algorithm are used to identify the person bounding box Bi frame by frame and track the individual displacement, obtaining the crowd density distribution coefficient CDM, the moving direction consistency coefficient MDC, and the abnormal behavior coefficient ABFV, generating the real-time behavior feature data set FV, and solving the problem that traditional methods lack quantification of the group behavior state. These data provide a key basis for risk prediction and resource scheduling, realizing the deep coordination of the data collection and analysis processing links, and enhancing the overall stability of the system.

[0077] (2)By receiving the encrypted real-time behavior feature data set FV of each node, a risk prediction algorithm is designed based on federated learning and the real-time behavior feature data set FV. This algorithm considers the comprehensive influence of the maximum value of the crowd density distribution coefficient max(CDMk) in region k, the moving direction consistency coefficient MDCk in region k, and the abnormal behavior coefficient ABFVk in region k, generates the cross-regional risk index RRIk, and through a hierarchical response mechanism, realizes precise risk prevention and control, avoiding the problem of delayed response caused by manual experience. By combining the cross-regional risk index RRIk and the road congestion coefficient Ck, a dynamic source scheduling algorithm is designed. By comprehensively considering the efficiency of each path n in the candidate path set p and the Manhattan distance R from the available resource unit m on the nth path in the candidate path set p to region k, a resource scheduling scheme is obtained, avoiding resource redundancy in low-risk regions. The federated learning mechanism realizes the combination of the cross-regional risk prediction algorithm and the dynamic resource scheduling algorithm, breaking through the problems of global analysis ability, regional response delay, and resource redundancy in low-risk regions caused by data islands in traditional systems, and improving the cross-regional collaboration efficiency.

[0078] (3) By collecting the execution feedback data of the resource scheduling scheme, a feedback data set F is generated to quantify the execution effect of the resource scheduling scheme. Based on the global analysis of the feedback data set F, the system can automatically identify decision-making biases and algorithm optimization trends, and optimize the risk prediction algorithm and resource scheduling algorithm through an adaptive learning mechanism. For the response time RT and resource utilization rate RC, the contribution weight coefficients of path correction and call resource unit selection are corrected to improve the real-time performance and accuracy of resource scheduling. For the risk prediction error, the contribution weight coefficients α, β, and γ in the regional risk prediction algorithm are dynamically balanced, improving the robustness of the algorithm. This closed-loop mechanism not only avoids the long-term performance degradation caused by static rules in traditional systems but also significantly improves the prevention and control efficiency and long-term stability through continuous data-driven optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 It is a schematic block diagram of a smart city public safety management system based on video surveillance according to the present invention;

[0080] Figure 2 It is a schematic block diagram of the public safety management resource planning process of a smart city public safety management system based on video surveillance according to the present invention;

[0081] Figure 3 It is a schematic diagram of data collection, processing, and flow direction. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0082] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0083] Embodiment 1

[0084] The present invention provides a smart city public safety management system based on video surveillance. Please refer to Figure 1 and Figure 3 , which includes a data collection and analysis module, a behavior feature extraction module, a risk prediction module, a dynamic decision-making module, an execution feedback module, and an iterative optimization module;

[0085] The data collection and analysis module collects the original video data stream and the real-time road data set D from the public video surveillance terminals in each region of the city, the urban traffic management platform, and the vehicle-mounted GPS through the API interface, and obtains the time series feature matrix TSM and the road congestion coefficient Ck;

[0086] The behavior feature extraction module extracts and encrypts behavior features from the time series feature matrix TSM at each local node, generates a set of homomorphically encrypted real-time behavior feature data FV, and transmits the set of homomorphically encrypted real-time behavior feature data FV and the road congestion coefficient Ck to the central node in ciphertext form;

[0087] The risk prediction module receives the encrypted real-time behavior feature data set FV from each node, establishes a cross-regional risk prediction algorithm through federated learning, calculates the cross-regional risk index RRI, and compares the cross-regional risk index RRIk of region k with the preset risk threshold range T to execute the security management plan;

[0088] The dynamic decision-making module establishes a dynamic resource scheduling algorithm based on the cross-regional risk index RRIk and the road congestion coefficient Ck of region k and generates a resource scheduling plan to be sent to the duty unit;

[0089] The execution feedback module obtains the feedback effect evaluation and generates a feedback data set F by quantifying the execution effect of the resource scheduling plan;

[0090] The iterative optimization module optimizes the cross-regional risk prediction algorithm and the dynamic resource scheduling algorithm based on the feedback data set F.

[0091] In this embodiment, by accessing the video surveillance terminals, traffic management platforms, and vehicle-mounted GPS data in various regions of the city in real time through the API interface, multi-dimensional data such as video streams, road traffic flows, and vehicle speeds are integrated to generate a time-series feature matrix TSM and a road congestion coefficient Ck, solving the traditional problem of data islands. At the same time, the homomorphic encryption technology is used to ensure the security of data transmission and avoid the risk of data leakage. The behavior feature extraction module uses object detection and optical path algorithms to identify the person bounding box Bi frame by frame and track the individual displacement, accurately extracting the real-time behavior feature data set FV, which is encrypted and transmitted to the central node, overcoming the defect of the traditional method's lag in detecting complex behaviors. The risk prediction module fuses encrypted data from multiple regions based on the federated learning framework, dynamically models to obtain the cross-regional risk index RRI, and triggers a hierarchical response in combination with the preset threshold range T, solving the problem of the traditional system's lag in response due to relying on manual experience and static rules. The dynamic decision-making module combines the cross-regional risk index RRIk and the road congestion coefficient Ck of region k to generate a resource scheduling plan, allocating paths and duty units in real time, optimizing the emergency response efficiency, and avoiding the bottlenecks of the traditional system's rigid pre-plan and low resource utilization rate. The execution feedback module quantifies the response time RT, resource utilization rate RC, and risk prediction accuracy SR, generates a feedback data set F to drive the iterative optimization module to dynamically adjust the algorithm contribution weight coefficient, forming a closed-loop management mechanism of "data collection and analysis → risk prediction → resource scheduling → feedback optimization", solving the problem of the traditional system's long-term performance degradation due to lack of adaptive optimization. Through modular collaboration and closed-loop data circulation, a public safety management system with privacy, real-time, and self-adaptability is constructed, solving the pain points of the traditional system's rigid resource scheduling and the inability of the emergency plan to be dynamically adjusted according to real-time risks, resulting in regional response delays or resource redundancy in low-risk regions, providing an extensible technical paradigm for smart city security management.

[0092] Embodiment 2

[0093] This embodiment is an explanatory description based on Embodiment 1. Please refer to Figure 1 and Figure 3 , specifically: The data collection and analysis module includes a data stream collection unit and a data analysis unit;

[0094] The data stream collection unit accesses the public video surveillance terminals in various regions of the city by calling the communication protocols compatible with different terminals through the API, and obtains the original video stream;

[0095] By calling the city traffic management platform and vehicle-mounted GPS data through the API, the maximum traffic flow max(Tf), real-time traffic flow RTf, historical average vehicle speed Hv, and real-time average vehicle speed Rv of the roads in the monitored jurisdiction are obtained, and a real-time road data set D is obtained.

[0096] The data analysis unit analyzes the original video stream and the implemented road data set D to obtain the time series feature matrix TSM and the road congestion coefficient Ck;

[0097] Using the YOLOv8 object detection algorithm, the person bounding box Bi is recognized frame by frame from the original video stream, the number of people per unit area is counted, the crowd density CD in the video area is calculated, and the improved optical flow algorithm is used to track the displacement of people to calculate the individual movement speed vector V;

[0098] Integrate the crowd density CD and the individual movement speed vector V in the original video stream to obtain the time series feature matrix TSM of the video area;

[0099] Based on the analysis of the real-time road data set D, the road congestion coefficient Ck is obtained. The analysis expression of the road congestion coefficient Ck is as follows:

[0100] ;

[0101] In the formula, η represents the contribution weight coefficient of the ratio of the real-time road traffic flow RTf to the maximum traffic flow max(Tf) of the road, and ι represents the contribution weight coefficient of the ratio of the average driving speed Rv of the real-time vehicles on the road to the average driving speed Hv of the historical vehicles on the road.

[0102] The purpose achieved by this formula: By the ratio of the real-time traffic flow RTf to the maximum traffic flow max(Tf) of the road, the current traffic complexity is quantified, and combined with the deviation degree of the average speed Rv of the real-time vehicles and the historical average speed Hv, the road traffic efficiency is reflected, providing a basis for the real-time road state for dynamic resource scheduling.

[0103] The behavior feature extraction module includes a behavior feature analysis unit and a data processing unit;

[0104] The behavior feature analysis unit extracts and encrypts the behavior features of the time series feature matrix TSM at each local node, and converts the time series feature matrix TSM into a crowd density distribution coefficient CDM, a moving direction consistency coefficient MDC, and an abnormal behavior coefficient ABFV;

[0105] Among them, the analysis expression of the crowd density distribution coefficient CDM is as follows:

[0106] ;

[0107] In the formula, (x, y) represents the video pixel coordinates, (xi, yi) represents the human body coordinates of the i-th person detected, N represents the total number of people detected, π represents the constant 3.14, σ represents the preset Gaussian kernel bandwidth, specifically a value set according to experience and experimental rules, and e represents the natural constant;

[0108] By analyzing the individual movement speed vector V, the individual movement direction angle θ is obtained, and the analysis expression of the movement direction consistency coefficient MDC is as follows:

[0109] ;

[0110] In the formula, θi represents the movement direction angle of the i-th person, θj represents the movement direction angle of the j-th person, and j ≠ i;

[0111] The purpose of this formula is achieved: by dynamically adjusting the Gaussian kernel bandwidth σ, adapting to the changes in crowd density in different scenarios, accurately quantifying the spatial distribution of the crowd in the video area, quantifying the consistency of the group movement direction by calculating the cosine similarity of the movement direction angles of all individuals, introducing the calculation of the differences between all individuals, comprehensively evaluating the coordination of the group movement, providing key inputs for the calculation of the cross-regional risk index RRI, and enhancing the accuracy of risk prediction.

[0112] Based on the personnel bounding box Bi, the number of people with squatting and lying-down behaviors in the surveillance video is analyzed. Based on the individual movement speed vector V, the number of people with stationary and running behaviors in the video is analyzed. Based on the proportion of the number of people with squatting, lying-down, stationary, and running behaviors in the surveillance video to the total number of detected people N, the abnormal behavior coefficient ABFV is obtained;

[0113] Among them, the detection criteria for squatting and lying-down behaviors are as follows:

[0114] If the current height of the personnel bounding box Bi ≤ 0.5 times the original height of the personnel bounding box Bi, and the current width of the personnel bounding box Bi ≥ 1.25 times the original width of the personnel bounding box Bi, it is determined that the current detection target is in a squatting state;

[0115] If the current height of the personnel bounding box Bi > 0.5 times the original height of the personnel bounding box Bi, and the current width of the personnel bounding box Bi < 1.25 times the original width of the personnel bounding box Bi, it is determined that the current detection target is in a squatting state;

[0116] If the current height of the personnel bounding box Bi ≤ 0.25 times the original height of the personnel bounding box Bi, and the current width of the personnel bounding box Bi ≥ 1.5 times the original width of the personnel bounding box Bi, it is determined that the current detection target is in a lying-down state;

[0117] If the current height of the personnel bounding box Bi > 0.25 times the original height of the personnel bounding box Bi, and the current width of the personnel bounding box Bi < 1.5 times the original width of the personnel bounding box Bi, it is determined that the current detection target is not in a lying-down state;

[0118] The detection criteria for stationary and running behaviors are as follows:

[0119] If the individual's movement speed vector V = 0 m / s, it is determined that the current detection target is in a stationary state;

[0120] If the individual's movement speed vector V ≠ 0 m / s, it is determined that the current detection target is not in a stationary state;

[0121] If the individual's movement speed vector V > 2.5 m / s, it is determined that the current detection target is in a running state;

[0122] If the individual's movement speed vector V ≤ 2.5 m / s, it is determined that the current detection target is not in a running state.

[0123] The data processing unit vertically combines the encrypted crowd density distribution coefficient CDM, movement direction consistency coefficient MDC, and abnormal behavior coefficient ABFV to generate a real-time behavior feature data set FV, homomorphically encrypts the obtained real-time behavior feature data set FV and the road congestion coefficient Ck, and transmits them to the central node in ciphertext form.

[0124] In this embodiment, the multi-terminal access protocol is accessed through the API interface, and the video monitoring terminals, teaching management platforms, and vehicle-mounted GPS data in various regions of the city are accessed in real time. The original video stream, the maximum traffic flow max(Tf) of the road, the real-time traffic flow RTf, the real-time average vehicle speed Rv, and the historical average vehicle speed Hv are synchronously obtained to generate a real-time road data set D, solving the problem of fragmented data collection caused by information islands in the traditional system. For the collected data, the YOLOv8 algorithm is used to identify the person bounding box Bi frame by frame, the crowd density CD is counted, and the individual movement speed vector V is tracked. Combining the improved optical flow algorithm, a temporal feature matrix TSM is generated, and the road congestion coefficient Ck is dynamically calculated, improving the real-time performance and accuracy of traffic state analysis. The behavior feature analysis unit calculates the crowd density distribution coefficient CDM through the Gaussian kernel function, quantifies the movement direction consistency coefficient MDC by cosine similarity, and detects squatting and lying down behaviors based on the changes in the height and width of the person bounding box Bi, and detects stationary and running behaviors based on the individual movement speed vector V to generate an abnormal behavior coefficient ABFV, solving the problem of lagging response of traditional detection methods to complex behavior detection. The data processing unit vertically combines the crowd density distribution coefficient CDM, the movement direction consistency coefficient MDC, and the abnormal behavior coefficient ABFV for homomorphic encryption, generates a real-time behavior feature data set FV for encrypted transmission to the central node, and through the localization processing and privacy protection mechanism, solves the data leakage risk caused by centralized calculation in the traditional system. This deep integration of multi-source data and refined behavior modeling provides high-precision input for risk prediction and resource scheduling, solves the problems of decision-making lag and error accumulation caused by data fragmentation, rough analysis, and privacy risks in the traditional system, and significantly improves the reliability and scenario adaptation ability of public safety management.

[0125] Example 3

[0126] This example is an explanatory note based on Example 2. Please refer to Figure 1 and Figure 2 . Specifically, the risk prediction module includes a risk modeling unit and an execution unit;

[0127] The risk modeling unit receives the encrypted real-time behavior feature data set FV from each node. Based on the data in the real-time behavior feature data set FV, it designs a cross-regional risk prediction algorithm through federated learning to obtain the cross-regional risk index RRIk of region k;

[0128] Among them, the expression of the cross-regional risk prediction algorithm is as follows:

[0129] ;

[0130] In the formula, k represents the region, RRIk represents the cross-regional risk index of region k, max(CDMk) represents the maximum value of the crowd density distribution coefficient CDM in region k, a represents the set density sensitivity coefficient, b represents the set curve density threshold, α represents the contribution weight coefficient of the maximum value max(CDMk) of the crowd density distribution coefficient in region k, MDCk represents the moving direction consistency coefficient of region k, β represents the contribution weight coefficient of the moving direction consistency coefficient MDCk of region k, ABFVk represents the abnormal behavior coefficient of region k, and γ represents the contribution weight coefficient of the abnormal behavior coefficient ABFVk of region k.

[0131] This formula achieves the purpose: through the logical function maps the maximum value max(CDMk) of the crowd density distribution coefficient to the interval [0,1], avoiding the sensitivity of the linear model to extreme density changes, and combines the direction consistency coefficient MDCk and the abnormal behavior coefficient ABFVk to accurately quantify the risk index of region k, providing a scientific basis for the hierarchical response mechanism.

[0132] The execution unit compares the cross-regional risk index RRIk of region k with the risk threshold range T, and executes the hierarchical response plan according to the comparison result;

[0133] If the cross-regional risk index RRIk of region k < the risk threshold Tmin, only record the current data as historical reference data;

[0134] If the risk threshold Tmin ≤ the cross-regional risk index RRIk of region k < the risk threshold Tmax, trigger a risk warning and push it to the duty units in the jurisdiction where region k is located and the public electronic screens in the nearby areas, advising the people and vehicles that have not entered region k to choose their routes carefully;

[0135] If the cross - regional risk index RRIk of area k ≥ the risk threshold Tmax, a risk warning is triggered. A prompt is generated on the public electronic screen in the nearby area to warn people and vehicles not entering area k to avoid entering area k, and the dynamic resource scheduling algorithm is triggered to automatically generate a resource scheduling plan and send it to the on - duty units in the jurisdiction.

[0136] Among them, the risk threshold Tmin represents the lower limit value in the risk threshold range T, and the risk threshold Tmax represents the upper limit value in the risk threshold range T.

[0137] In this embodiment, the encrypted real - time behavior feature data set FV from each node is received through the federated learning framework, and a cross - regional risk prediction algorithm is designed based on these data to calculate the cross - regional risk index RRI. The cross - regional risk prediction algorithm takes into account the maximum value of the population density distribution coefficient max(CDMk), the moving direction consistency coefficient MDCk, and the abnormal behavior coefficient ABFVk in area k. Through the set density - sensitive coefficient a, curve density threshold b, and their respective contribution weight coefficients α, β, and γ, a non - linear mapping of the population density is performed, enhancing the sensitivity of the model to high - density areas. The cross - regional risk index RRIk of area k is comprehensively evaluated. This comprehensive analysis method can not only quantify the degree of population aggregation in a specific area, but also measure the consistency of group movement and the frequency of abnormal behaviors, thus providing comprehensive data support for risk assessment. The execution unit then compares the calculated cross - regional risk index RRIk with the preset risk threshold range T, and then triggers corresponding hierarchical response plans, such as recording historical reference data, pushing early warning information, or generating an emergency resource scheduling plan, etc. In this way, the system can not only quickly respond to potential security threats, but also achieve precise prevention and control on the premise of protecting privacy, effectively avoiding the problem of response lag caused by traditional reliance on manual experience and static rules. In addition, this method enhances the predictability and response ability to complex public security events, enabling urban managers to take the most appropriate measures in the first time, improving the utilization efficiency of public resources and the overall level of public security management. This mechanism particularly emphasizes the deep integration and intelligent analysis of data in different dimensions, laying a solid foundation for building a more intelligent and safe urban environment.

[0138] Embodiment 4

[0139] This embodiment is an explanatory description based on Embodiment 3. Please refer to Figure 1 and Figure 2 , specifically: The dynamic decision - making module includes a dynamic resource scheduling decision - making unit;

[0140] The dynamic resource scheduling decision - making unit designs a dynamic resource scheduling algorithm based on the cross - regional risk index RRIk of area k and the road congestion coefficient Ck, and generates a resource scheduling optimization value PD;

[0141] The path n of the duty units in the jurisdiction where area k is located obtained from the urban traffic management platform by calling the API, and by integrating the path n, a real-time candidate path set p is formed;

[0142] The expression of the dynamic resource scheduling algorithm is as follows:

[0143] ;

[0144] In the formula, argmin represents the parameter for finding the minimum value of the objective function, pn represents the nth path in the candidate path set p, and Spnk represents the Manhattan distance S of the nth path in the candidate path set p to area k. represents the path efficiency of the nth path in the selected path set p, Rpnm represents the Manhattan distance R from the available resource unit m on the nth path in the candidate path set p to area k, t represents the time difference from the start of the risk warning to the current moment, δ represents the contribution weight coefficient of the path efficiency of the nth path in the candidate path set p to area k, w represents the contribution weight coefficient of the Manhattan distance R from the available resource unit m on the nth path in the candidate path set p to area k, μ represents the contribution weight coefficient of the time difference t from the start of the risk warning to the current moment, e μt The time factor of the cross-regional risk index RRIk of area k;

[0145] Export the parameter when the resource scheduling optimization value PD is the minimum and generate a resource scheduling plan. The content of the plan includes the nth path in the selected candidate path set p and the available resource unit m called from the nth path in the selected candidate path set p, and send the resource scheduling plan to the duty units in the jurisdiction.

[0146] In this embodiment, the dynamic resource scheduling decision unit designs a dynamic resource scheduling algorithm based on the cross-regional risk index RRIk and the road congestion coefficient Ck of area k, generates a candidate path set p by calling the factual path data of the urban traffic management platform, calculates the resource scheduling optimization value PD of each path, and selects the parameter when the resource scheduling optimization value PD takes the minimum value as the resource scheduling plan. This algorithm reduces the negative impact of high congestion or other high-risk areas on path selection by integrating the cross-regional risk index RRIk and the road congestion coefficient Ck of area k, and introduces the time factor e μt, Dynamically enhance the sensitivity to response timeliness to ensure that the priority of resource scheduling automatically adjusts over time. This mechanism breaks through the limitations of traditional reliance on static path planning or a single distance metric, achieving multi-dimensional real-time integration of path efficiency, resource accessibility, and time urgency, significantly shortening the emergency response time and improving resource utilization. By dynamically optimizing the optimal path in the candidate path set p and the available resource unit m, the system solves the problems of resource redundancy in low-risk areas or response delays in high-risk areas caused by the rigidity of traditional plans, providing a quantifiable and reusable technical path for the precise handling of urban public safety incidents, and further strengthening the agility and scenario adaptation ability of smart city management.

[0147] Example 5

[0148] This example is an explanatory note based on Example 4. Please refer to Figure 1 and Figure 2 , specifically: The execution feedback module includes a feedback data collection unit;

[0149] The feedback data collection unit generates a feedback data set F by collecting the execution feedback data of the resource scheduling plan to quantify the execution effect of the resource scheduling plan;

[0150] Among them, the feedback data set F includes the response time RT, the resource utilization rate RC, and the risk prediction accuracy rate SR;

[0151] The response time RT represents the time difference between the time when the available resource unit m obtains the time stamp based on the GPS device and the time when the resource scheduling plan is sent to arrive at the scene;

[0152] The resource utilization rate RC represents the ratio of the actual number of resources arriving at the scene to the predicted number of required resources;

[0153] The risk prediction accuracy rate SR represents the ratio of the number of correct risk predictions to the total number of risk predictions.

[0154] The iterative optimization module includes an iterative optimization unit;

[0155] The iterative optimization unit dynamically optimizes the cross-regional risk prediction algorithm and the dynamic resource scheduling algorithm based on the feedback data set F;

[0156] The optimization plan for the cross-regional risk prediction algorithm is as follows:

[0157] Based on the risk prediction accuracy rate SR in the feedback data set F, if the risk prediction accuracy rate SR ≥ 95%, it is determined that the prediction result is reasonable, and there is no need to dynamically adjust the contribution weight coefficients α, β, and γ in the cross-regional risk prediction algorithm;

[0158] If the risk prediction accuracy rate SR < 95%, it is determined that the prediction result is unreasonable, and the contribution weight coefficients α, β, and γ in the cross-regional risk prediction algorithm are dynamically adjusted;

[0159] The optimization plan for the dynamic resource scheduling algorithm is as follows:

[0160] Based on the response time RT and resource utilization rate RC in the feedback data set F;

[0161] Optimization direction 1: If the single response time RT ≤ 15 minutes, it is determined that the nth path in the selected candidate path set p is reasonable, and there is no need to dynamically adjust the contribution weight coefficients δ and w in the dynamic resource scheduling algorithm;

[0162] If the single response time RT > 15 minutes, it is determined that the nth path in the selected candidate path set p is unreasonable, and the contribution weight coefficients δ and w in the dynamic resource scheduling algorithm are dynamically adjusted;

[0163] Optimization direction 2: If the resource utilization rate RC ≥ 80% for three consecutive times, it is determined that the selected available resource unit m is reasonable, and there is no need to dynamically adjust the contribution weight coefficients δ and w in the dynamic resource scheduling algorithm;

[0164] If there is one time when the resource utilization rate RC < 80% within three consecutive times, it is determined that the selected available resource unit m is unreasonable, and the contribution weight coefficients δ and w in the dynamic resource scheduling algorithm are dynamically adjusted.

[0165] In this embodiment, by introducing an execution feedback module and an iterative optimization module, while enhancing the system's adaptive adjustment ability, the stability and flexibility of its long-term operation are significantly enhanced. The feedback data collection unit in the execution feedback module can collect the system response time RT, resource utilization rate RC, and risk prediction accuracy SR, providing solid data support for subsequent optimization decisions. This feedback mechanism ensures that the system is always in an efficient operation state and can be dynamically adjusted when needed. At the same time, the iterative optimization module, through the iterative optimization unit, continuously adjusts and optimizes the core algorithms in the system according to the collected feedback data. For the emergency response requirements in different scenarios, the optimization unit will adjust the dynamic resource scheduling strategy according to the real-time feedback to ensure that resources can be accurately and efficiently allocated in the face of emergencies, avoiding resource waste or shortage. In addition, for the optimization of cross-regional risk prediction, the optimization unit can analyze historical data and real-time feedback, adjust the parameters of the risk assessment model, and improve the prediction accuracy and response speed. Through this closed-loop feedback and iterative optimization mechanism, the system can not only quickly respond to and adapt to emergencies, but also continuously improve the decision-making accuracy and resource allocation efficiency during the long-term operation process, thus ensuring the efficiency and reliability of the entire public safety management system in different environments and conditions. Generally speaking, the introduction of this solution significantly improves the system's dynamic adaptation ability and decision-making intelligence, making public safety management more accurate, efficient, and sustainable.

[0166] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A smart city public safety management system based on video surveillance, characterized by: It includes data collection and analysis module, behavior feature extraction module, risk prediction module, dynamic decision-making module, execution feedback module and iterative optimization module; The data collection and analysis module collects raw video data streams and real-time road data sets D from public video surveillance terminals in various areas of the city, urban traffic management platforms and vehicle-mounted GPS through API interfaces, and obtains the time series feature matrix TSM and road congestion coefficient Ck; The behavior feature extraction module extracts and encrypts the time series feature matrix TSM at each local node, generates a homomorphically encrypted real-time behavior feature data set FV, and transmits the homomorphically encrypted real-time behavior feature data set FV and the road congestion coefficient Ck to the central node in ciphertext form; The risk prediction module receives the encrypted real-time behavior feature data set FV of each node, establishes a cross-regional risk prediction algorithm through federated learning, calculates the cross-regional risk index RRI, and executes the safety management plan based on the comparison between the cross-regional risk index RRIk of region k and the preset risk threshold range T; The dynamic decision-making module establishes a dynamic resource scheduling algorithm based on the cross-region risk index RRIk and road congestion coefficient Ck of region k and generates a resource scheduling plan and sends it to the duty unit; The execution feedback module obtains feedback effect evaluation and generates feedback data set F by quantifying the execution effect of the resource scheduling solution; The iterative optimization module optimizes the cross-regional risk prediction algorithm and dynamic resource scheduling algorithm based on the feedback dataset F.

2. According to claim 1, a smart city public safety management system based on video surveillance is characterized in that: The data acquisition and analysis module includes a data stream acquisition unit and a data analysis unit; The data stream acquisition unit uses API to call communication protocols compatible with different terminals, access public video surveillance terminals in various areas of the city, and obtain the original video stream; By calling the urban traffic management platform and the vehicle GPS data through the API, the maximum road traffic flow max (Tf), the real-time road traffic flow RTf, the historical average vehicle speed Hv and the real-time average vehicle speed Rv in the monitored area are obtained to obtain the real-time road data set D.

3. A smart city public safety management system based on video surveillance according to claim 2, characterized in that: The data analysis unit obtains the time series feature matrix TSM and the road congestion coefficient Ck by analyzing the original video stream and the implementation road data set D; Use the YOLOv8 target detection algorithm to identify the person bounding box Bi frame by frame in the original video stream, count the number of people in the unit area, calculate the crowd density CD in the video area, use the improved optical flow algorithm to track the displacement of people, and calculate the individual movement speed vector V; Integrate the crowd density CD and individual moving speed vector V in the original video stream to obtain the temporal feature matrix TSM of the video area; Based on the analysis of the real-time road data set D, the road congestion coefficient Ck is obtained. The analysis expression of the road congestion coefficient Ck is as follows: ； Where η represents the contribution weight coefficient of the ratio of the real-time traffic flow RTf to the maximum traffic flow max (Tf), and ι represents the contribution weight coefficient of the ratio of the real-time average vehicle speed Rv to the historical average vehicle speed Hv.

4. A smart city public safety management system based on video surveillance according to claim 3, characterized in that: The behavior feature extraction module includes a behavior feature analysis unit and a data processing unit; The behavior feature analysis unit extracts and encrypts the behavior features of the time series feature matrix TSM at each local node, and converts the time series feature matrix TSM into a crowd density distribution coefficient CDM, a moving direction consistency coefficient MDC, and an abnormal behavior coefficient ABFV; Among them, the crowd density distribution coefficient CDM analysis expression is as follows: ； Where (x, y) represents the video pixel coordinates, (xi, yi) represents the detected human coordinates of the i-th person, N represents the total number of people detected, π represents the constant 3.14, σ represents the preset Gaussian kernel bandwidth, and e represents a natural constant; By analyzing the individual moving velocity vector V, the individual moving direction angle θ is obtained, and the moving direction consistency coefficient MDC analysis expression is as follows: ； In the formula, θi represents the moving direction angle of the i-th person, θj represents the moving direction angle of the j-th person, and j≠i; The number of people squatting and lying down in the surveillance video is analyzed based on the personnel bounding box Bi, the number of people standing still and running in the video is analyzed based on the individual movement speed vector V, and the abnormal behavior coefficient ABFV is obtained based on the ratio of the number of people squatting, lying down, standing still and running in the surveillance video to the total number of people detected N; Among them, the detection standards for squatting and lying down are as follows: If the current height of the person boundary box Bi is ≤ 0.5 times the original height of the person boundary box Bi, and the current width of the person boundary box Bi is ≥ 1.25 times the original width of the person boundary box Bi, then it is determined that the current detection target is in a squatting state; If the current height of the person boundary box Bi is greater than 0.5 times the original height of the person boundary box Bi, and the current width of the person boundary box Bi is less than 1.25 times the original width of the person boundary box Bi, then it is determined that the current detection target is in a squatting state; If the current height of the person boundary box Bi is ≤ 0.25 times the original height of the person boundary box Bi, and the current width of the person boundary box Bi is ≥ 1.5 times the original width of the person boundary box Bi, then it is determined that the current detection target is in a lying state; If the current height of the person boundary box Bi is greater than 0.25 times the original height of the person boundary box Bi, and the current width of the person boundary box Bi is less than 1.5 times the original width of the person boundary box Bi, it is determined that the current detection target is not in a lying state; The stationary and running behavior detection criteria are as follows: If the individual moving speed vector V = 0m / s, it is determined that the current detection target is in a stationary state; If the individual moving speed vector V≠0m / s, it is determined that the current detection target is not in a stationary state; If the individual moving speed vector V>2.5m / s, it is determined that the current detection target is in a running state; If the individual moving speed vector V≤2.5m / s, it is determined that the current detection target is not in a running state.

5. A smart city public safety management system based on video surveillance according to claim 4, characterized in that: The data processing unit generates a real-time behavior feature data set FV by vertically merging the encrypted crowd density distribution coefficient CDM, moving direction consistency coefficient MDC and abnormal behavior coefficient ABFV, homomorphically encrypts the acquired real-time behavior feature data set FV and road congestion coefficient Ck, and transmits them to the central node in ciphertext form.

6. A smart city public safety management system based on video surveillance according to claim 5, characterized in that: The risk prediction module includes a risk modeling unit and an execution unit; The risk modeling unit receives the encrypted real-time behavior feature data set FV from each node, designs a cross-regional risk prediction algorithm through federated learning based on the data in the real-time behavior feature data set FV, and obtains the cross-regional risk index RRIk of region k; Among them, the cross-regional risk prediction algorithm expression is as follows: ； In the formula, k represents region, RRIk represents the cross-regional risk index of region k, max(CDMk) represents the maximum value of the population density distribution coefficient CDM in region k, a represents the set density sensitivity coefficient, b represents the set curve density threshold, α represents the contribution weight coefficient of the maximum value max(CDMk) of the population density distribution coefficient in region k, MDCk represents the moving direction consistency coefficient of region k, β represents the contribution weight coefficient of the moving direction consistency coefficient MDCk of region k, ABFVk represents the abnormal behavior coefficient of region k, and γ represents the contribution weight coefficient of the abnormal behavior coefficient ABFVk of region k.

7. A smart city public safety management system based on video surveillance according to claim 6, characterized in that: The execution unit compares the cross-region risk index RRIk of region k with the risk threshold range T, and executes a graded response plan based on the comparison result; If the cross-region risk index RRIk of region k is less than the risk threshold Tmin, only the current data is recorded as historical reference data; If the risk threshold Tmin≤the cross-regional risk index RRIk of region k<the risk threshold Tmax, a risk warning is triggered and pushed to the duty units in the jurisdiction of region k and the public electronic screens in the nearby areas, advising people and vehicles that have not entered region k to choose routes carefully; If the cross-region risk index RRIk of region k ≥ the risk threshold Tmax, a risk warning is triggered, and a prompt is generated on the public electronic screen in the nearby area to warn people and vehicles that have not entered region k to avoid entering region k, and the dynamic resource scheduling algorithm is triggered to automatically generate a resource scheduling plan and send it to the duty unit in the jurisdiction; The risk threshold Tmin represents the lower limit value in the risk threshold range T, and the risk threshold Tmax represents the upper limit value in the risk threshold range T.

8. A smart city public safety management system based on video surveillance according to claim 7, characterized in that: The dynamic decision module includes a dynamic resource scheduling decision unit; The dynamic resource scheduling decision unit designs a dynamic resource scheduling algorithm based on the cross-region risk index RRIk and road congestion coefficient Ck of region k, and generates a resource scheduling optimization value PD; By calling the API from the urban traffic management platform, we obtain the path n from the duty units in the jurisdiction of area k to area k, and integrate the paths n to form a real-time candidate path set p; The dynamic resource scheduling algorithm expression is as follows: ； Wherein, argmin represents the parameter for finding the minimum value of the objective function, pn represents the nth path in the candidate path set p, Spnk represents the Manhattan distance S of the nth path in the candidate path set p to the area k, Rpnm represents the Manhattan distance R of the available resource unit m on the nth path in the candidate path set p to the area k, t represents the time difference from the start of the risk warning to the current moment, δ represents the path efficiency contribution weight coefficient of the nth path in the candidate path set p to the area k, w represents the contribution weight coefficient of the Manhattan distance R of the available resource unit m on the nth path in the candidate path set p to the area k, and μ represents the contribution weight coefficient of the time difference t from the start of the risk warning to the current moment; The parameters when the resource scheduling preference value PD is minimum are exported and a resource scheduling plan is generated. The plan includes the nth path in the selected candidate path set p and the available resource unit m called from the nth path in the selected candidate path set p. The resource scheduling plan is sent to the on-duty units within the jurisdiction.

9. A smart city public safety management system based on video surveillance according to claim 8, characterized in that: The execution feedback module includes a feedback data collection unit; The feedback data collection unit generates a feedback data set F by collecting the execution feedback data of the resource scheduling scheme to quantify the execution effect of the resource scheduling scheme; Among them, the feedback data set F includes response time RT, resource utilization RC and risk prediction accuracy SR; The response time RT represents the time difference between the available resource unit m and the time difference between the resource scheduling plan sent to the arrival site based on the timestamp obtained by the GPS device; Resource utilization rate RC represents the ratio of the actual number of resources on site to the predicted number of required resources; The risk prediction accuracy rate SR represents the ratio of the number of correct risk predictions to the total number of risk predictions.

10. A smart city public safety management system based on video surveillance according to claim 8, characterized in that: The iterative optimization module includes an iterative optimization unit; The iterative optimization unit dynamically optimizes the cross-regional risk prediction algorithm and the dynamic resource scheduling algorithm based on the feedback data set F; The cross-regional risk prediction algorithm optimization scheme is as follows: Based on the risk prediction accuracy SR in the feedback data set F, if the risk prediction accuracy SR ≥ 95%, the prediction result is judged to be reasonable, and there is no need to dynamically adjust the contribution weight coefficients α, β and γ in the cross-regional risk prediction algorithm; If the risk prediction accuracy rate SR is less than 95%, the prediction result is judged to be unreasonable, and the contribution weight coefficients α, β and γ in the cross-regional risk prediction algorithm are dynamically adjusted; The optimization scheme of dynamic resource scheduling algorithm is as follows: Based on the response time RT and resource utilization RC in the feedback data set F; Optimization direction 1: If the single response time RT is ≤ 15 minutes, the nth path in the selected candidate path set p is determined to be reasonable, and there is no need to dynamically adjust the contribution weight coefficients δ and w in the dynamic resource scheduling algorithm; If the single response time RT is greater than 15 minutes, the nth path in the selected candidate path set p is determined to be unreasonable, and the contribution weight coefficients δ and w in the dynamic resource scheduling algorithm are dynamically adjusted; Optimization direction 2: If the resource utilization rate RC is ≥ 80% for three consecutive times, it is determined that the available resource unit m is reasonably selected, and there is no need to dynamically adjust the contribution weight coefficients δ and w in the dynamic resource scheduling algorithm; If the resource utilization rate RC is less than 80% for three consecutive times, it is determined that the selection of the available resource unit m is unreasonable, and the contribution weight coefficients δ and w in the dynamic resource scheduling algorithm are dynamically adjusted.

Images