Video monitoring safety control system and method based on Internet of Things

Abstract

The invention discloses a video monitoring safety control system and method based on the Internet of Things, particularly relates to the technical field of video monitoring safety control, and is used for solving the problem of low video monitoring control efficiency. Multi-source data are collected through an Internet of Things sensor network, the time sequence consistency is ensured by adopting a clock synchronization protocol, the data are preprocessed, the accuracy is ensured, then correlation analysis is carried out by utilizing a Bayesian network and principal component analysis, historical information and scene modes are extracted, risk levels of all areas are determined, and the risk levels of all the areas are determined. And determining a resource scheduling priority according to the risk level. And by simulating scene change, the risk early warning capability of the system is improved. According to the system, camera resources are dynamically scheduled through a collaborative optimization algorithm, monitoring resources of a high-risk area are preferentially allocated, redundant monitoring is reduced, power consumption and bandwidth use are optimized, and therefore the response speed and precision of video monitoring in a complex environment are improved, and the safety of video monitoring is improved.

Description

Technical Field

[0001] The present invention relates to the field of video surveillance security control technology, and more specifically, to a video surveillance security control system and method based on the Internet of Things. Background Art

[0002] As a technology that connects physical devices through the network to achieve data exchange, the Internet of Things (IoT) has developed rapidly in recent years and is widely used in smart cities, traffic management, environmental monitoring and other fields. By integrating sensors, devices, and communication technologies, the IoT enables devices to share data in real time and provide more comprehensive perception and control capabilities. Video surveillance technology, as an important means of security management, is also constantly evolving. It is possible to integrate sensors and cameras into the monitoring system. The IoT allows real-time monitoring of environmental data such as temperature, humidity, smoke, and human flow through sensor networks, and combines these data with video data to provide more comprehensive environmental perception capabilities.

[0003] Deficiencies in existing technologies: In security monitoring, sensors and cameras work independently, and environmental data (such as temperature, humidity, and crowd density) cannot be effectively combined with video surveillance data, resulting in the system being unable to promptly identify potential risks such as fire or crowd gathering. For example, areas with dense crowds may not be marked as high risk, which in turn affects the real-time response of video surveillance. At the same time, inefficient resource scheduling also results in the inability of camera monitoring resources to be dynamically adjusted according to risk changes. Cameras are prone to monitoring low-risk areas with high resource usage for a long time, while monitoring resources for high-risk areas are insufficient, which easily increases regional safety risks. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a video surveillance security control system and method based on the Internet of Things to solve the problem of low efficiency of video surveillance control in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The video surveillance security control method based on the Internet of Things includes the following steps:

[0007] Collect multi-source sensor data, transmit the data through the IoT sensor network, synchronize the data using the clock synchronization protocol, and pre-process the collected data;

[0008] Perform correlation analysis on the preprocessed data, extract historical information and scenario patterns from the collected data, determine the risk level of each area, determine the priority of regional resource scheduling based on the risk level of each area, and use Bayesian networks to simulate scenario changes in different areas;

[0009] Based on the results of real-time regional risk assessment and priority allocation, a collaborative optimization algorithm is used to dynamically schedule camera resources for different areas, and monitoring resources are prioritized in high-risk areas.

[0010] In a preferred embodiment, the sensor includes an environmental sensor, a crowd detector, a noise sensor, an infrared sensor, a smoke alarm, and a camera;

[0011] Environmental sensors collect environmental data, crowd detectors, noise sensors, infrared sensors, and smoke alarms collect data as sensor data, and cameras collect data as video data;

[0012] Use Kalman filtering to denoise environmental data and sensor data, use the isolation forest algorithm to detect outliers in the data, and use the non-local mean algorithm to denoise video data;

[0013] Use Z-score standardization to standardize environmental data and sensor data;

[0014] Use dynamic time warping algorithm to time align environmental data, sensor data, and video data.

[0015] In a preferred embodiment, the preprocessed data is subjected to association analysis, and the specific process is as follows:

[0016] Integrate various types of collected data into a feature matrix, where columns represent data or feature dimensions;

[0017] Use Bayesian networks to perform conditional probability calculations to infer the probability of occurrence of specific features;

[0018] Use PCA to reduce the dimension of the feature matrix, identify the principal components, and calculate the correlation coefficient or mutual information between different data features to quantify the correlation between each data;

[0019] Based on the results of association analysis, and combined with Bayesian network inference and principal components extracted by PCA, a multi-dimensional scenario model is created.

[0020] In a preferred embodiment, historical information and scenario patterns are extracted from the collected data to determine the risk level of each area, the priority of regional resource scheduling is determined according to the risk level of each area, and the Bayesian network is used to simulate the scenario changes in different areas. The specific process is as follows:

[0021] The construction of historical information datasets involves extracting information of different dimensions from multiple data sources and constructing a joint dataset;

[0022] On the joint dataset, association rule mining algorithms are used to identify high-frequency scene patterns;

[0023] Calculate the support and confidence of each event combination, which is defined as the frequency of occurrence of a certain event combination;

[0024] Use time series pattern mining to identify recurring patterns of events over time;

[0025] After identifying historical high-risk areas and scenario patterns, causal reasoning and modeling of different risk factors are performed through Bayesian networks, and different risk factors are constructed into a causal relationship graph to simulate scenario changes under sensor values ​​in different areas.

[0026] In a preferred embodiment, time series pattern mining is used to identify the repetitive patterns of events in the time dimension. The specific process is as follows:

[0027] Perform sliding window construction to divide the historical time series data into multiple subsequences in a sliding window manner;

[0028] In each subsequence, an autoregressive model is used to model the temporal characteristics of event occurrence;

[0029] Cluster analysis was used to divide historical data into regions, and the K-Means clustering algorithm was used to aggregate data from different regions and identify high-risk areas as cluster centers;

[0030] Select the cluster center, and the Euclidean distance measures the distance from each data point to the cluster center;

[0031] According to the distance between the data point and the cluster center, the data point is assigned to the nearest cluster center, and the cluster center is recalculated until the clustering result converges;

[0032] Cluster analysis is used to identify areas with similar risk patterns, and high-risk areas are marked as priority areas for resource scheduling.

[0033] In a preferred embodiment, according to the results of real-time regional risk assessment and priority allocation, a collaborative optimization algorithm is used to dynamically schedule camera resources for different regions, and monitoring resources are preferentially allocated to high-risk areas. The specific process is as follows:

[0034] Conduct task priority analysis for the monitoring area. The task priority of the monitoring area is determined by risk level, event urgency, and camera status;

[0035] The risk level indicates the current security risk of the area, the event urgency is the urgency assessment of the current task, and the camera status is the current status of the monitoring device, including available resources, bandwidth, and power consumption;

[0036] The event urgency is calculated based on motion detection and behavior anomaly detection to obtain the matching degree between the current event and similar historical events;

[0037] Determine the current status of each camera based on task priority, camera resource status, and environmental restrictions, and determine whether the working mode needs to be adjusted. The task priority is the urgency of the event required for monitoring in the current area. The camera resource status includes the power consumption, bandwidth, and network transmission status of the current camera. Environmental restrictions include network bandwidth and energy consumption restrictions.

[0038] Coordinate and manage cameras in the area based on determined task priorities and camera working modes.

[0039] In a preferred embodiment, the cameras in the area are coordinated and managed according to the determined task priorities and camera working modes. The specific steps are as follows:

[0040] Obtain the camera overlap in the coverage area. When the camera overlap is higher than the set threshold, collaborative scheduling is performed.

[0041] After the cameras to be co-scheduled are determined, the synergy between the cameras is calculated. The synergy calculation is quantified based on the spatial position, task priority, and current working mode of the cameras.

[0042] Calculate the synergy effect by the spatial distance and coverage radius between cameras;

[0043] When there is a synergy between cameras, resource scheduling and control scheduling of the angle, focal length and rotation speed between cameras are performed through collaborative resource allocation.

[0044] The video surveillance security control system based on the Internet of Things is used to implement the above-mentioned video surveillance security control method based on the Internet of Things, including:

[0045] The data processing module is used to collect multi-source sensor data, transmit the data through the IoT sensor network, synchronize the data using the clock synchronization protocol, and pre-process the collected data;

[0046] The scenario demand analysis module is used to perform correlation analysis on the preprocessed data, extract historical information and scenario patterns from the collected data, determine the risk level of each area, determine the priority of regional resource scheduling based on the risk level of each area, and use the Bayesian network to simulate scenario changes in different areas;

[0047] The monitoring control module is used to dynamically schedule camera resources for different areas using a collaborative optimization algorithm based on the results of real-time regional risk assessment and priority allocation, and to prioritize monitoring resources in high-risk areas.

[0048] Technical effects and advantages of the present invention:

[0049] The technical effect of the present invention is to collect multi-source data through the Internet of Things sensor network, combine the clock synchronization protocol to ensure the time consistency of the data, and pre-process the data collected by the sensors and cameras. In the data pre-processing process, the accuracy and integrity of the data are ensured by using technologies such as denoising and outlier detection. Subsequently, the multi-source data is analyzed for association using Bayesian networks and principal component analysis, scene patterns are extracted from historical data, and resource scheduling strategies are dynamically adjusted according to the risk levels under different environmental conditions, thereby enhancing the system's early warning capabilities for risk events, and can also perform intelligent monitoring and scheduling for specific environmental changes in different regions, thereby improving the accuracy and timeliness of monitoring.

[0050] In addition, through the collaborative optimization algorithm to dynamically schedule camera resources, based on the results of real-time regional risk assessment and priority allocation, the system can intelligently identify high-risk areas and prioritize the allocation of monitoring resources to avoid excessive consumption of resources in low-risk areas. At the same time, when multiple cameras are covered, it reduces redundant monitoring and resource waste, improves monitoring coverage and resource utilization efficiency, improves the emergency response capability and risk prediction accuracy of video surveillance, and enables video surveillance to effectively respond to various security threats in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 The present invention is a flow chart of a video surveillance security control method based on the Internet of Things.

[0052] Figure 2 It is a structural schematic diagram of the video surveillance security control system based on the Internet of Things of the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] Example 1: Figure 1 As shown, the video surveillance security control method based on the Internet of Things includes the following steps:

[0055] Collect multi-source sensor data, transmit the data through the IoT sensor network, synchronize the data using the clock synchronization protocol, and pre-process the collected data;

[0056] Perform correlation analysis on the preprocessed data, extract historical information and scenario patterns from the collected data, determine the risk level of each area, determine the priority of regional resource scheduling based on the risk level of each area, and use Bayesian networks to simulate scenario changes in different areas;

[0057] Based on the results of real-time regional risk assessment and priority allocation, a collaborative optimization algorithm is used to dynamically schedule camera resources for different areas, and monitoring resources are prioritized in high-risk areas.

[0058] Multi-source sensor data collection is performed to collect data from multiple sensors and cameras. Sensors may include environmental sensors (such as temperature, humidity, and light sensors), crowd detectors, noise sensors, and specific safety equipment (such as infrared sensors, smoke alarms, etc.). Video data mainly comes from surveillance cameras, combined with real-time event data in a specific area (such as intrusion, fire, etc.). This data collection step is performed through the IoT sensor network to ensure that all sensors and cameras synchronously collect data and transmit it to the central processing system for data processing. To ensure clock synchronization, a high-precision clock synchronization protocol can be used to ensure that the timestamps of different devices are consistent;

[0059] The collected data often contains noise, redundancy or outliers, which require data cleaning and denoising. In this step, the environmental data and sensor data are denoised through Kalman filtering or adaptive filtering algorithms to eliminate the impact of environmental interference on the data. In addition, outlier detection algorithms (such as the 3σ principle or isolation forest algorithm based on statistics) are used to detect and remove outliers in sensor data to ensure the validity and accuracy of input data. Especially in terms of video data, the system can remove noise in the video through image denoising algorithms (such as the non-local mean algorithm NL-Means) to ensure the quality of subsequent processed images;

[0060] Before data from different sources are integrated, data standardization is required. Since the acquisition frequencies and data formats of sensors and cameras are different, environmental data, sensor data, and video data are first standardized through Z-score standardization or minimum-maximum normalization to unify all data to the same scale.

[0061] The processing of video data involves the standardization of image formats and the unification of frame rates. In order to ensure the consistency of multi-source data in the time dimension, the dynamic time warping (DTW) algorithm is used to time-align environmental data, sensor data, and video data of different frequencies to ensure comprehensive analysis within the same time segment.

[0062] After data standardization is completed, the environmental data, sensor data, and video data are correlated and analyzed through multi-source data fusion technology (such as Bayesian network or principal component analysis (PCA)). The purpose of this step is to unify various data sources (for example, environmental data, sensor data, and video data as different data sources) in an overall framework to form a multi-dimensional scene description and serve as a scene model. For example, the system can associate crowd density, ambient light intensity, and motion trajectory in real-time video to create a comprehensive risk assessment data structure. The multi-source data fusion algorithm used ensures that the correlation of various types of data is effectively identified. The specific steps are as follows:

[0063] Integrate various types of collected data (such as crowd density, ambient light intensity, video motion trajectory, etc.) into a feature matrix X, where each column represents a data source or feature dimension. Assuming there are m sensor data features, n video data features, and p environmental data features, the dimension of the feature matrix is ​​(m+n+p)×T, where T is the time step.

[0064] In a Bayesian network, each node represents a data feature (such as crowd density, light intensity, motion detection results in a video), and the edge represents the causal relationship between features. Through conditional probability calculation, the probability of a specific feature (such as abnormal behavior) under specific conditions can be inferred. For example, a Bayesian network can infer "the probability of abnormal behavior occurring when the crowd density is high and the light intensity is low";

[0065] Use PCA to reduce the dimension of the feature matrix and identify the most representative principal components. For example, extract the first few principal components from all sensor and video data to represent key monitoring environment features, such as "the main cause of high-risk activities in the area". The principal components can retain most of the relevant information of the data and reduce noise interference;

[0066] By calculating the correlation coefficient or mutual information between different data features, the system can quantify the correlation between various data sources. For example, it can calculate the correlation coefficient between the motion trajectory in the video and the ambient light intensity, and identify the correlation. A higher correlation indicates that abnormal behavior is more likely to occur under specific environmental conditions.

[0067] Based on the results of association analysis, the system combines Bayesian network inference and principal components extracted by PCA to create multi-dimensional scenario patterns. For example, the system can describe "the risk of illegal intrusion is higher in areas with dense traffic and insufficient light", thereby generating a comprehensive risk assessment data structure for subsequent risk monitoring and early warning.

[0068] The association analysis process helps the system extract hidden relationships from multi-source data and accurately identify the occurrence patterns of risk events in different environments.

[0069] Extract historical information and scenario patterns from the data collected and monitored over a long period of time, identify high-risk areas or periodic risk events, and provide a basis for real-time risk assessment and resource scheduling. The specific steps are as follows:

[0070] Constructing a historical information dataset involves extracting information of different dimensions from multiple data sources, mainly including the following categories: environmental data including temperature, humidity, etc., sensor data including light, personnel density, etc., video data including behavior trajectories and abnormal event markers (such as wandering, running, gathering, etc.) in surveillance videos; external data such as historical weather data, emergency reports, etc.

[0071] Assume there is a data set of n time points, and the sensor data is D s ={s1, s2, ..., s n}, video data is D v = {v1, v2, ..., v n}, external data is D e ={e1, e2, ..., e n Each data point is provided by different data sources, so it is necessary to construct a joint dataset D = {(s i , v i , e i )|i=1, 2, ..., n}, used for subsequent data mining, i represents the time point of data collection;

[0072] On the joint dataset, an association rule mining algorithm is used to identify high-frequency scene patterns. The goal of association rule mining is to discover "if-then" patterns, such as "if the crowd density is high and the light intensity is low, abnormal behavior is frequent." The April algorithm is used to mine frequent itemsets. The steps are as follows:

[0073] Calculate the support of each event combination, which is defined as the frequency of occurrence of a certain event combination: Support(X) = total number of data points (X) / total number of data points, where X is the event item set, such as the combination of high temperature in sensor data and wandering behavior in video data;

[0074] Calculating event combinations The probability of event Y occurring when event X occurs is defined as the probability of event Y occurring. Through this process, high-frequency patterns associated with risks can be identified, such as the association between certain weather conditions and specific abnormal behaviors.

[0075] Use time series pattern mining technology to identify repeated patterns of events in the time dimension. Time series pattern mining is based on sliding window technology. The specific steps are as follows:

[0076] Sliding window construction is performed to divide the historical time sequence data d into multiple subsequences in a sliding window manner. Assuming that the sliding window size is w, the data set F is divided into {(D1, D2, ..., D w ), (D2, D3, ..., D w+1 ), ...}. Each subsequence represents the risk data for a fixed period of time;

[0077] In each subsequence, an autoregressive model is used to model the temporal characteristics of the event. The autoregressive model formula is: Among them, x t is the observed value at the current time t, c is the constant term (or intercept), which represents the fixed part of the sequence, and φ i is the autoregression coefficient, which measures the previous data x t-i For the current time x t The influence of t is the white noise term, representing the error or unexplained part;

[0078] The role of the constant term c is to provide a base level or trend for the time series, so that the model can still capture the overall level of the series when there are no other influencing factors. For example, in the autoregressive model of temperature data, it can represent the seasonal effect or the average value of the basic climate conditions.

[0079] In order to find high-risk areas in the monitoring area, cluster analysis is used to divide the historical data into regions, and the K-Means clustering algorithm is used to aggregate the data of different regions into several categories, among which the high-risk areas will be identified as cluster centers. The specific steps are as follows:

[0080] First, K initial cluster centers are selected to represent the initial risk status of K regions;

[0081] Calculate the distance from each data point to the cluster center using the Euclidean distance metric: Among them, x i is a data point, c k is the kth cluster center;

[0082] According to the distance between the data point and the cluster center, the data point is assigned to the nearest cluster center and the cluster center is recalculated: Among them, c k Repeat this step for all data points belonging to the kth cluster until the clustering results converge;

[0083] Cluster analysis is used to identify areas with similar risk patterns and mark high-risk areas as priority areas for future resource dispatch;

[0084] After identifying historical high-risk areas and scenario patterns, causal reasoning and modeling of different risk factors are performed through Bayesian networks. Bayesian networks can construct different risk factors (such as temperature, traffic flow, abnormal behavior, etc.) into a causal relationship graph, which is represented as a structure of nodes and edges, where each node represents a risk factor and the edge represents a causal relationship:

[0085] Conditional probability calculations are performed, and Bayes' theorem is used to infer the risk probability of a certain area under given conditions: Where A is a risk event, B is a trigger condition, P(A|B) is the probability of risk event A occurring under condition B, that is, the risk assessment result under the current conditions, P(B|A) represents the probability of trigger condition B occurring when risk event A occurs, P(A) is the prior probability of risk event A occurring, and P(B) is the prior probability of trigger condition B occurring. Based on the statistical results of historical data, the Bayesian network can infer the real-time risk level of a certain area based on different sensor data and historical events;

[0086] Risk Level L i According to the conditional probability P(A|B), the system usually divides the risk into multiple levels (such as low, medium, and high risk), and divides P(A|B) according to the preset threshold. T1 and T2 are the thresholds for risk level classification, with level L1 representing low risk, L2 representing medium risk, and L3 representing high risk;

[0087] According to the risk level L of each area i , generate the corresponding resource scheduling priority P i ,formula: Among them, λ i is the priority coefficient of region i, which is set according to the importance of the region or the frequency of historical risk events. i is the risk level of area i, C i The available resources of the current camera (such as bandwidth, power consumption, etc.) ensure that high-risk areas are allocated first when resources are insufficient;

[0088] The Bayesian network is used to simulate the scenario changes under sensor values ​​in different areas, such as "how the risk of crowded areas changes under high temperature and high humidity conditions", thus providing a basis for future resource scheduling;

[0089] Assume that the conditions in a certain monitoring area are high temperature (above 35°C), high humidity (above 70% humidity) and dense traffic (more than 30 people per minute). At this time, the risk event A of concern is illegal intrusion. Under this condition, Bayesian theorem is used to infer the risk probability of the area.

[0090] Assume that in the historical data, the probability of illegal intrusion P(A) is 0.05 (that is, 5% of the events are illegal intrusions). When illegal intrusion occurs, it is usually accompanied by high temperature, high humidity and dense traffic. The estimated probability P(B|A) is 0.8. Regardless of whether illegal intrusion occurs, the probability of the current condition P(B) is 0.6. The calculation formula is:

[0091] According to the calculation results, the probability of risk event A under given condition B is 0.0667 (about 6.67%);

[0092] Assuming the risk level thresholds are T1 = 0.05 and T2 = 0.1, P(A|B) = 0.0667 falls between T1 and T2, so the risk level of the area is L2;

[0093] If the priority coefficient of region i is λ i =1.5, the current available resources of the camera are C i =0.8, Therefore, the resource scheduling priority of this area is 3.75, which means that when allocating camera resources, this area will be given priority to obtain more resources to ensure that the occurrence of medium-risk events can be effectively monitored;

[0094] Through the Bayesian network, scene changes in different areas can be simulated. For example, in the case of high temperature and high humidity but sparse traffic, it can be inferred that the risk is low and the scheduling priority is reduced accordingly. Conversely, in the case of high temperature and high humidity with dense traffic, the risk probability increases, the priority is increased, and the camera's resolution, frame rate and coverage will also be adjusted accordingly. This type of scheduling strategy is based on the results of Bayesian reasoning and can dynamically adjust camera resources in different scenarios to ensure real-time response to environmental changes and reasonable allocation of resources.

[0095] Based on the results of the real-time regional risk assessment and priority allocation in the previous step, camera resources are dynamically scheduled to ensure enhanced monitoring in high-risk areas while saving resources in low-risk areas. The specific steps are as follows:

[0096] Perform task priority analysis on the monitoring area. The task priority of each monitoring area i is determined by three main factors: risk level, event urgency, and camera status. i Reflects the current security risk in the area and the urgency of the incident. i According to the urgency assessment of the current task, such as the severity of fire, theft and other related events, the camera status S i It is the current status of the monitoring device, including available resources, bandwidth, power consumption, etc. The calculation formula of task priority is: T i =w1·F1(L i)+w2·F1(E i )+w3·F1(S i ), where T i is the final task priority of region i, w1, w2, w3 are weight coefficients of the three factors on the priority, F1(L i )、F1(E i )、F1(S i ) is the processing function of different factors, which is used to quantify the data;

[0097] Weight coefficient w k The setting directly affects the weights of different factors in the priority calculation. The Bayesian optimization algorithm is used to dynamically adjust the weight coefficients to minimize the delay and failure rate of task execution. The objective function of the optimization problem is: Among them, L is the loss function, which measures the deviation between the task scheduling result and the actual demand; D i is the actual execution delay of the ith task, Delayed results predicted by the model;

[0098] By continuously adjusting the weight coefficients of various factors through Bayesian optimization, the priority calculation can adapt to the resource requirements in different scenarios, thereby achieving the optimal task scheduling effect.

[0099] It should be noted that the urgency of an event is mainly obtained by analyzing and judging real-time monitoring data (video, sensors, etc.), a process that can combine multiple aspects such as sensor networks, camera monitoring data, and historical data models;

[0100] The sensor network can monitor various factors in the environment in real time, such as temperature, humidity, smoke, noise and other abnormal conditions, and judge the urgency of the event based on the real-time changes in sensor data;

[0101] Event analysis is performed through video data monitored in real time by cameras to detect abnormal behaviors or emergencies. The urgency of the event is judged based on visual processing and target recognition technology. The urgency of the event is calculated through motion detection and behavioral anomaly detection. The matching degree between the current event and similar historical events is identified through time series analysis and anomaly detection technology. If the current event is highly similar to the pattern of historical emergency events, the urgency is high.

[0102] Adaptively adjust the camera working mode, judge the current status of each camera, and determine whether its working mode needs to be adjusted. The working mode of each camera depends on the following key factors: task priority, which indicates the urgency of the event in the current area, camera resource status, including the power consumption, bandwidth, network transmission status of the current camera, and environmental constraints, including network bandwidth and energy consumption constraints;

[0103] The camera working mode mainly affects two core parameters: resolution and frame rate. In order to cope with tasks of different priorities, the following formula is used to calculate the resolution R i and frame rate F i To make dynamic adjustments: Among them, R min and R max are the lowest and highest resolutions of the camera, respectively, T i is the priority of the i-th task; the frame rate adjustment formula is: F i =F min +log(1+T i )·(F max -F min ), F min 、F max are the minimum and maximum frame rates respectively;

[0104] When adjusting the camera working mode, it is also necessary to manage bandwidth and power consumption. Especially when the task priority is low, the system should save resources by reducing power consumption and bandwidth usage to ensure that bandwidth usage is proportional to the camera working mode. Cameras with higher task priority will get more bandwidth, while cameras with lower priority will use less bandwidth to ensure reasonable allocation of network resources.

[0105] During the operation of the camera, feedback adjustments are continuously made based on the actual monitoring effect, network status and device power consumption. The camera's operating data (such as clarity, latency, network congestion, etc.) is collected through the real-time monitoring feedback module to ensure that the control system can adjust the camera's working mode according to real-time changes to avoid performance degradation or resource waste under high load.

[0106] Coordinate the scheduling of camera resources. Based on the determined task priorities and camera working modes, the coordinated use of resources by multiple cameras is achieved through collaborative optimization algorithms. Multiple cameras are effectively coordinated for monitoring, covering key areas, reducing resource waste, and improving the comprehensiveness and effectiveness of monitoring.

[0107] Obtain the camera overlap in the coverage area, that is, multiple cameras may cover the same area or adjacent areas, avoid repeated monitoring, reasonably allocate resources to priority areas, and meet the priorities of different areas. It is necessary to ensure that the resources of high-priority areas are met first; according to the coverage of the camera, determine whether the two need to be coordinated. Assume that the coverage area of ​​camera i and j is FG i and FG j , define the overlap O ij for:

[0108] When the overlap is higher than the set threshold, it is considered that cameras i and j have overlapping resources and need to be coordinated.

[0109] After determining the cameras that need to be coordinated, calculate the synergy effect between the cameras. The synergy effect is quantified based on the spatial position of the camera, task priority, and current working mode. The specific synergy effect formula is: Where, d ij is the distance between cameras i and j, η i , η j is the coverage radius of the camera, M i 、M j are the current working modes of cameras i and j respectively;

[0110] The synergy effect is calculated by the spatial distance and coverage radius between cameras to ensure that cameras that are close to each other can work together to avoid repeated monitoring. When the task priorities and working modes of two cameras are close, the synergy effect will be enhanced, and collaborative monitoring can be achieved by allocating bandwidth and power consumption.

[0111] When there is a synergistic effect between cameras i and j, resources are scheduled through collaborative resource allocation. The core goal of collaborative resource allocation is to ensure that high-priority areas obtain sufficient resources while reducing resource consumption in low-priority areas. The collaborative resource allocation formula for camera j is as follows: In the formula, R j is the resource allocation obtained for camera j, T i , T j Respectively represent the task priorities of cameras i and j, B ij For the total resource budget, allocate it between the two cameras;

[0112] Through the combination of priority and working mode, high-priority cameras are given priority to obtain more resources, while low-priority cameras have their resource allocation reduced. When resources are limited, high-priority tasks are given priority to obtain sufficient bandwidth and power consumption.

[0113] While ensuring that high-priority tasks get priority in resources, the system also needs to dynamically adjust the physical properties of the camera. By monitoring the working status of the camera in real time, the system can automatically adjust the camera's shooting angle, rotation direction, focal length and zoom speed to meet the monitoring needs of different areas. For example, when the system detects an emergency in a high-risk area, the camera can automatically rotate to the area where the incident occurred and adjust the focal length according to the situation to ensure the clarity of the picture. This physical control can be accurately performed by a stepper motor or servo system to ensure that there are no blind spots in the monitoring area.

[0114] In real-time regional risk assessment, the system dynamically adjusts the camera's viewing angle according to the risk level and task priority. For example, in low-risk areas, the system can adjust the camera to wide-angle mode to cover a larger area and save power consumption; in high-risk areas, the system adjusts the camera to narrow-angle mode and focuses on specific targets to capture details. This adjustment process can be automatically performed through algorithms, combined with machine learning and environmental perception technology to achieve more accurate monitoring results;

[0115] Through the intelligent control system, the camera's coverage can be flexibly adjusted according to real-time needs. For example, when monitoring areas with dense traffic, the camera can automatically expand the monitoring range; when it is necessary to monitor key targets, the camera's monitoring range can be narrowed and accurately located to the area where the incident occurred. In this way, the system can flexibly adjust the camera's monitoring range in a dynamic environment and improve monitoring efficiency.

[0116] During the collaborative monitoring process of cameras, the working status and collaborative effect of each camera are monitored in real time, and the collaborative scheduling strategy is dynamically adjusted based on the feedback. The feedback mechanism optimizes the collaborative effect by collecting the real-time operating data of the camera (such as monitoring clarity, frame rate, delay, etc.);

[0117] Through the coordinated scheduling of camera resources, when multiple cameras cover the same area or adjacent areas, resources can be reasonably allocated to avoid resource waste. For example, when multiple cameras work together, the frame rate can be reduced, data redundancy can be reduced, and bandwidth and power consumption allocation can be optimized to ensure that high-priority tasks are supported by sufficient resources. This directly reduces the power consumption and bandwidth pressure of the system and improves resource utilization efficiency. By reducing repeated monitoring and data redundancy, resources can be saved in the monitoring network, allowing high-risk areas to be monitored more accurately and timely. Especially in the case of tight resources, coordinated scheduling ensures that the monitoring quality of important areas will not be affected, and the overall resource usage of the system is more reasonable, which improves the sustainability and stability of the system.

[0118] To cope with complex surveillance scenarios, multiple cameras can work together to achieve more accurate surveillance coverage. The system can coordinate the physical movement between different cameras, and ensure that all high-risk areas are fully monitored by adjusting the angle, focal length and rotation speed of multiple cameras. Coordinated scheduling is not limited to resource allocation, but can also be extended to physical movement coordination between cameras to make their perspectives complementary and form a more comprehensive surveillance view.

[0119] Collaborative scheduling ensures coordination between multiple cameras, expands monitoring coverage, and reduces blind spots. For example, through collaborative scheduling, multiple cameras in the same area can reasonably adjust the viewing angle and coverage to ensure that there are no unmonitored areas. At the same time, the system can dynamically adjust the camera's resolution, frame rate and other parameters to ensure high-quality monitoring of key areas at critical moments;

[0120] By covering blind spots and coordinating camera perspectives, we can more comprehensively grasp the activities within the monitoring area, significantly reducing monitoring loopholes and potential safety hazards. Especially in the event of an emergency, multiple cameras can work together to focus on key areas and provide clearer and more timely monitoring data, thereby effectively improving the comprehensiveness and security of video surveillance.

[0121] It should be noted that the thresholds involved in the embodiments may be determined according to specific scenarios and requirements, and are usually adjusted and optimized based on factors such as historical data and real-time data.

[0122] The technical effect of the present invention is to collect multi-source data through the Internet of Things sensor network, combine the clock synchronization protocol to ensure the time consistency of the data, and pre-process the data collected by the sensors and cameras. In the data pre-processing process, the accuracy and integrity of the data are ensured by using technologies such as denoising and outlier detection. Subsequently, the multi-source data is analyzed for association using Bayesian networks and principal component analysis, scene patterns are extracted from historical data, and resource scheduling strategies are dynamically adjusted according to the risk levels under different environmental conditions, thereby enhancing the system's early warning capabilities for risk events, and can also perform intelligent monitoring and scheduling for specific environmental changes in different regions, thereby improving the accuracy and timeliness of monitoring.

[0123] In addition, through the collaborative optimization algorithm to dynamically schedule camera resources, based on the results of real-time regional risk assessment and priority allocation, the system can intelligently identify high-risk areas and prioritize the allocation of monitoring resources to avoid excessive consumption of resources in low-risk areas. At the same time, when multiple cameras are covered, it reduces redundant monitoring and resource waste, improves monitoring coverage and resource utilization efficiency, improves the emergency response capability and risk prediction accuracy of video surveillance, and enables video surveillance to effectively respond to various security threats in complex environments.

[0124] Embodiment 2: This embodiment is a system embodiment of embodiment 1, which is used to implement the video surveillance security control method based on the Internet of Things introduced in embodiment 1, such as Figure 2 As shown, specifically including:

[0125] The data processing module is used to collect multi-source sensor data, transmit the data through the IoT sensor network, synchronize the data using the clock synchronization protocol, and pre-process the collected data;

[0126] The scenario demand analysis module is used to perform correlation analysis on the preprocessed data, extract historical information and scenario patterns from the collected data, determine the risk level of each area, determine the priority of regional resource scheduling based on the risk level of each area, and use the Bayesian network to simulate scenario changes in different areas;

[0127] The monitoring control module is used to dynamically schedule camera resources for different areas using a collaborative optimization algorithm based on the results of real-time regional risk assessment and priority allocation, and to prioritize monitoring resources in high-risk areas.

[0128] The above formulas are all dimensionless and calculated numerically. Specific dimension removal can be achieved by various means such as standardization, which will not be elaborated here. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0129] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, an ATA hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state ATA hard disk.

[0130] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0131] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0132] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0133] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0134] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0135] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A video surveillance security control method based on the Internet of Things, characterized in that: The steps include: Collect multi-source sensor data, transmit the data through the IoT sensor network, synchronize the data using the clock synchronization protocol, and pre-process the collected data; Perform correlation analysis on the preprocessed data, extract historical information and scenario patterns from the collected data, determine the risk level of each area, determine the priority of regional resource scheduling based on the risk level of each area, and use Bayesian networks to simulate scenario changes in different areas; Based on the results of real-time regional risk assessment and priority allocation, a collaborative optimization algorithm is used to dynamically schedule camera resources for different areas, and monitoring resources are prioritized in high-risk areas.

2. The video surveillance security control method based on the Internet of Things according to claim 1 is characterized in that: Collect multi-source sensor data, transmit the data through the IoT sensor network, synchronize the data using the clock synchronization protocol, and pre-process the collected data. The specific process is as follows: Sensors include environmental sensors, crowd detectors, noise sensors, infrared sensors, smoke alarms, and cameras; Environmental sensors collect environmental data, crowd detectors, noise sensors, infrared sensors, and smoke alarms collect data as sensor data, and cameras collect data as video data; Use Kalman filtering to denoise environmental data and sensor data, use the isolation forest algorithm to detect outliers in the data, and use the non-local mean algorithm to denoise video data; Use Z-score standardization to standardize environmental data and sensor data; Use dynamic time warping algorithm to time align environmental data, sensor data, and video data.

3. The video surveillance security control method based on the Internet of Things according to claim 2 is characterized in that: The association analysis is performed on the preprocessed data. The specific process is as follows: Integrate various types of collected data into a feature matrix, where columns represent data or feature dimensions; Use Bayesian networks to perform conditional probability calculations to infer the probability of occurrence of specific features; Use PCA to reduce the dimension of the feature matrix, identify the principal components, and calculate the correlation coefficient or mutual information between different data features to quantify the correlation between each data; Based on the results of association analysis, and combined with Bayesian network inference and principal components extracted by PCA, a multi-dimensional scenario model is created.

4. The video surveillance security control method based on the Internet of Things according to claim 3 is characterized in that: Extract historical information and scenario patterns from the collected data, determine the risk level of each area, determine the priority of regional resource scheduling based on the risk level of each area, and use the Bayesian network to simulate scenario changes in different areas. The specific process is as follows: The construction of historical information datasets involves extracting information of different dimensions from multiple data sources and constructing a joint dataset; On the joint dataset, association rule mining algorithms are used to identify frequently occurring scene patterns; Calculate the support and confidence of each event combination, which is defined as the frequency of occurrence of a certain event combination; Use time series pattern mining to identify recurring patterns of events over time; After identifying historical high-risk areas and scenario patterns, causal reasoning and modeling of different risk factors are performed through Bayesian networks, and different risk factors are constructed into a causal relationship graph to simulate scenario changes under sensor values ​​in different areas.

5. The video surveillance security control method based on the Internet of Things according to claim 4 is characterized in that: Use time series pattern mining to identify repeated patterns of events in the time dimension. The specific process is as follows: Perform sliding window construction to divide the historical time series data into multiple subsequences in a sliding window manner; In each subsequence, an autoregressive model is used to model the temporal characteristics of event occurrence; Cluster analysis was used to divide historical data into regions, and the K-Means clustering algorithm was used to aggregate data from different regions and identify high-risk areas as cluster centers; Select the cluster center, and the Euclidean distance measures the distance from each data point to the cluster center; According to the distance between the data point and the cluster center, the data point is assigned to the nearest cluster center, and the cluster center is recalculated until the clustering result converges; Cluster analysis is used to identify areas with similar risk patterns, and high-risk areas are marked as priority areas for resource scheduling.

6. The video surveillance security control method based on the Internet of Things according to claim 5 is characterized in that: Based on the results of real-time regional risk assessment and priority allocation, a collaborative optimization algorithm is used to dynamically schedule camera resources for different areas, and monitoring resources are allocated preferentially to high-risk areas. The specific process is as follows: Conduct task priority analysis for the monitoring area. The task priority of the monitoring area is determined by risk level, event urgency, and camera status; The risk level indicates the current security risk of the area, the event urgency is the urgency assessment of the current task, and the camera status is the current status of the monitoring device, including available resources, bandwidth, and power consumption; The event urgency is calculated based on motion detection and behavior anomaly detection to obtain the matching degree between the current event and similar historical events; Determine the current status of each camera based on task priority, camera resource status, and environmental restrictions, and determine whether the working mode needs to be adjusted. The task priority is the urgency of the event required for monitoring in the current area. The camera resource status includes the power consumption, bandwidth, and network transmission status of the current camera. Environmental restrictions include network bandwidth and energy consumption restrictions. Coordinate and manage cameras in the area based on determined task priorities and camera working modes.

7. The video surveillance security control method based on the Internet of Things according to claim 6 is characterized in that: According to the determined task priority and camera working mode, coordinate and manage the cameras in the area. The specific steps are as follows: Obtain the camera overlap in the coverage area. When the camera overlap is higher than the set threshold, collaborative scheduling is performed. After the cameras to be co-scheduled are determined, the synergy between the cameras is calculated. The synergy calculation is quantified based on the spatial position, task priority, and current working mode of the cameras. Calculate the synergy effect by the spatial distance and coverage radius between cameras; When there is a synergy between cameras, resource scheduling and control scheduling of the angle, focal length and rotation speed between cameras are performed through collaborative resource allocation.

8. A video surveillance security control system based on the Internet of Things, used to implement the video surveillance security control method based on the Internet of Things according to any one of claims 1 to 7, characterized in that: include: The data processing module is used to collect multi-source sensor data, transmit the data through the IoT sensor network, synchronize the data using the clock synchronization protocol, and pre-process the collected data; The scenario demand analysis module is used to perform correlation analysis on the preprocessed data, extract historical information and scenario patterns from the collected data, determine the risk level of each area, determine the priority of regional resource scheduling based on the risk level of each area, and use the Bayesian network to simulate scenario changes in different areas; The monitoring control module is used to dynamically schedule camera resources for different areas using a collaborative optimization algorithm based on the results of real-time regional risk assessment and priority allocation, and to prioritize monitoring resources in high-risk areas.

Images