Terminal equipment collaborative video monitoring method and system

Abstract

The invention relates to the technical field of network video monitoring, in particular to a terminal equipment collaborative video monitoring method and system, which comprises the following steps of: monitoring video data based on terminal equipment, extracting a motion vector, a pixel intensity value and a color histogram of each frame, carrying out space-time alignment, calculating an average value and a variance of each parameter, and calculating the average value and the variance of each parameter; time sequence characteristics are analyzed, parameter changes of key frames are extracted, and an initial behavior characteristic set is established; according to the method, the motion vector, the pixel intensity and the color histogram of each frame of video are analyzed, and the data are subjected to space-time alignment and statistical analysis, so that the understanding of a behavior mode is improved, and clustering analysis helps a system to efficiently identify abnormities, monitor in real time and calculate parameter deviation to ensure that abnormal behaviors are found in time; the adaptability and efficiency of the system to the monitoring environment are further improved through data exchange and collaborative analysis between the terminal devices, and the coverage range and the response speed of the monitoring system are remarkably enhanced through the highly collaborative processing mode.

Description

Technical Field

[0001] The present invention relates to the technical field of network video monitoring, and in particular to a terminal device collaborative video monitoring method and system. Background Art

[0002] The field of network video surveillance technology involves technologies and methods for transmitting and managing video data using computer networks. This includes acquiring video data from video capture devices such as cameras, sending this data through wired or wireless networks, and processing and storing video content on servers or cloud platforms. Key components of this technology area include video compression and transmission technology, real-time or non-real-time video surveillance software, and data security and privacy protection measures. Network video surveillance is widely used in home security, public safety, industrial monitoring, and commercial environments to achieve functions such as remote monitoring, event detection, and subsequent video analysis.

[0003] Among them, the terminal device collaborative video surveillance method refers to multiple terminal devices such as smart phones, tablets or professional monitoring equipment working together through the network to achieve more efficient video data processing and monitoring management. This method allows data exchange and sharing of processing resources between terminal devices, thereby optimizing the response speed and data processing capabilities of the monitoring system. The uses of terminal device collaborative video surveillance mainly include improving monitoring coverage, enhancing the flexibility and scalability of video surveillance systems, and improving monitoring accuracy and event response efficiency by integrating multiple sensor information. This technology is particularly important in commercial security, traffic monitoring and smart home systems, and can provide more comprehensive and accurate monitoring solutions.

[0004] Existing technologies mainly rely on centralized processing, which is prone to delays in the face of large-scale data and lacks real-time performance. Lack of multi-device collaboration makes the efficiency of handling emergencies low. Existing methods are not deep enough in video analysis, often resulting in false alarms or missed alarms, affecting the accuracy of monitoring. Insufficient data sharing and integration, limited monitoring scope, and reduced security. In addition, the scalability and data security of existing systems are insufficient, and they cannot effectively respond to complex security threats, reducing the practicality and reliability of the system. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a terminal device collaborative video monitoring method and system.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a terminal device cooperative video monitoring method, comprising the following steps: S1: Based on the terminal device monitoring video data, extract the motion vector, pixel intensity value and color histogram of each frame, perform spatiotemporal alignment, calculate the mean and variance of each parameter, analyze the time series characteristics, extract the parameter changes of key frames, and establish the initial behavior feature set; S2: Based on the initial behavior feature set, compare the motion vectors, pixel intensity values ​​and color histograms of consecutive frames, calculate the differences, correlations and covariances, group the parameters of the same distribution, perform cluster analysis, analyze the central value, variance and range of each cluster, identify abnormal points, verify the clustering results, adjust the clustering parameters, and establish a normal behavior pattern model; S3: Based on the normal behavior pattern model, monitor the real-time motion vector, pixel intensity value and color histogram of the terminal device, calculate the deviation from the model, determine the fluctuation value of the deviation, identify the parameters exceeding the threshold, mark them as abnormal, record and classify the abnormal parameters, analyze the distribution characteristics, identify the abnormal behavior type, and establish the abnormal detection result; S4: Based on the anomaly detection result, exchange anomaly parameters and detection status between terminal devices, filter parameters exceeding the threshold, merge data of multiple terminal devices, update the normal behavior pattern model, adjust the detection threshold, re-evaluate the anomaly detection conditions, and build a collaborative anomaly detection model.

[0007] As a further solution of the present invention, the step of acquiring the initial behavior feature set is specifically: S111: extracting data of each frame from the surveillance video, including motion vector, pixel intensity and color histogram, calculating features of multiple frames of data, and generating a frame data feature set; S112: performing spatiotemporal alignment on the frame data feature set, using a calculation method of timestamp difference and spatial position change to ensure the temporal and spatial continuity of the data, and obtaining a spatiotemporal alignment feature set; S113: Statistical analysis is performed on the spatiotemporal alignment feature set using the formula: ; and ; Calculate the mean and variance of each parameter to obtain the parameter statistical feature set; in, represents the average value, represents the variance, represents a data item extracted from the spatiotemporal aligned feature set, representing motion vectors, pixel intensity or color histogram data, is the number of data items, indicating the total number of data involved in the calculation of the mean and variance; S114: Analyze the time series characteristics in the parameter statistical feature set, identify and extract significant changes in parameters in key frames, and establish an initial behavior feature set.

[0008] As a further solution of the present invention, the step of acquiring the normal behavior pattern model is specifically as follows: S211: Based on the initial behavior feature set, compare the motion vectors, pixel intensity values ​​and color histograms of consecutive frames, calculate the difference between each frame, obtain the difference amount, and generate a difference feature set; S212: For the data in the difference feature set, the formula is used: ; Calculate the correlation and covariance to obtain the correlation feature set; in, represent and The correlation coefficient between them is used to evaluate the relationship between the differentiation parameters. Representative A target parameter difference of the frame, Represents another parameter difference of the same frame, and Represents all frames and The average value of S213: Based on the correlation feature set, apply a clustering algorithm to group the parameters, analyze the central value, variance and range of each cluster, perform clustering, and obtain a clustering result set; S214: Identify and verify outliers on the clustering result set, optimize the model by adjusting clustering parameters, and establish a normal behavior pattern model.

[0009] As a further solution of the present invention, the step of obtaining the abnormality detection result is specifically: S311: monitoring the real-time motion vector, pixel intensity value and color histogram of the terminal device, calculating the deviation between the current frame and the model based on the normal behavior mode model, and generating a deviation feature set; S312: For the data in the deviation feature set, a statistical method formula is used: ; Calculate the fluctuation value of the deviation and generate a fluctuation value result set; in, represents a single deviation value from the deviation feature set, used to calculate the difference from the mean deviation, Represents the average of the deviation values, which is used to evaluate the central position of the overall deviation. is the number of samples, indicating the total number of elements in the deviation feature set, used for normalized fluctuation value calculation, The fluctuation value representing the deviation is used to quantify the stability of the deviation and assist in identifying abnormal fluctuations; S313: Identify parameters exceeding a preset threshold from the fluctuation value result set, mark the parameters as abnormal, and generate an abnormal marking set by combining the fluctuation value and the threshold judgment; S314: Based on the abnormal tag set, record and classify abnormal parameters, identify abnormal behavior types by analyzing the distribution characteristics of the parameters, and establish abnormal detection results.

[0010] As a further solution of the present invention, the steps of acquiring the collaborative anomaly detection model are specifically as follows: S411: Based on the abnormality detection result, the terminal devices exchange key abnormality parameters and the current detection status, implement data synchronization through a defined communication protocol, and generate a data status after the exchange; S412: from the exchanged data state, all abnormal parameters exceeding a preset threshold are screened by a comparison algorithm to ensure that only key data affecting system performance is processed, and a screened abnormal parameter set is generated; S413: merging the screened abnormal parameter set with data from multiple terminals, applying data fusion technology to integrate information from differentiated sources, updating the normal behavior pattern model, and adjusting the detection threshold inside the model to match the current data behavior, thereby generating an updated model; S414: Using the updated model, re-evaluate and optimize the abnormality detection conditions, using the statistical analysis formula: ; Building collaborative anomaly detection models; in, Represents the value of each monitored abnormal parameter, which is used to determine the degree of deviation of the data point. Represents the average value of the abnormal parameter, which is used to measure the central position of a set of data. Represents the standard deviation of the parameter, which is used to describe the dispersion of the data. Adjust the significance of deviations from the mean, Adjust the influence of standard deviation to control the contribution of deviation from the mean and data fluctuation to anomaly detection. Represents the total number of data points involved in the calculation, used to obtain the average of the weighted sum and balance the impact of differentiated terminal device data points. Represents collaborative anomaly detection metrics.

[0011] A terminal device collaborative video monitoring system, the terminal device collaborative video monitoring system is used to execute the terminal device collaborative video monitoring method, the system comprises: The data acquisition module extracts the motion vector, pixel intensity value and color histogram of each frame based on the terminal device monitoring video data, performs spatiotemporal alignment, calculates the mean and variance of the parameters, extracts the parameter changes of the key frames, and establishes the initial behavior feature set; The feature analysis module compares the associated parameters of consecutive frames based on the initial behavior feature set, calculates differences and covariances, performs parameter grouping and clustering, analyzes the central value and range of the clusters, identifies and excludes abnormal points, verifies the results and adjusts the clustering parameters, and establishes a normal behavior pattern model; The behavior modeling module monitors real-time parameters based on the normal behavior pattern model, calculates deviations from the model, determines the fluctuation value of the deviation, identifies parameters exceeding the threshold, marks anomalies, records and classifies abnormal parameters, analyzes distribution characteristics, and establishes anomaly detection results; The anomaly detection module exchanges anomaly parameters and detection status between terminal devices based on the anomaly detection results, filters and synchronizes parameters exceeding thresholds, merges data, updates the normal behavior pattern model, adjusts thresholds, re-evaluates detection conditions, and establishes a collaborative anomaly detection model; The collaborative update module integrates multiple terminal data based on the collaborative anomaly detection model, synchronously updates the model and detection status, adjusts the threshold, and records the system update status.

[0012] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, the understanding of behavior patterns is improved by analyzing the motion vector, pixel intensity and color histogram of each frame of video, and performing spatiotemporal alignment and statistical analysis on these data. Cluster analysis helps the system to efficiently identify anomalies, and real-time monitoring and parameter deviation calculation ensure timely detection of abnormal behaviors. Data exchange and collaborative analysis between terminal devices further improve the system's adaptability and efficiency to the monitoring environment. This highly collaborative processing method significantly enhances the coverage and response speed of the monitoring system, providing a more comprehensive and reliable security solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a schematic diagram of the workflow of the present invention; Figure 2 Flow chart of the steps for obtaining the initial behavior feature set of the present invention; Figure 3 A flow chart of the steps for obtaining the normal behavior pattern model of the present invention; Figure 4 A flowchart of the steps for obtaining abnormality detection results of the present invention; Figure 5 The figure is a flow chart of the steps for obtaining the collaborative anomaly detection model of the present invention. DETAILED DESCRIPTION

[0014] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0015] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0016] Example: See also Figure 1 The present invention provides a technical solution: a terminal device cooperative video monitoring method, comprising the following steps: S1: Based on the terminal device monitoring video data, extract the motion vector, pixel intensity value and color histogram of each frame, perform spatiotemporal alignment, calculate the mean and variance of each parameter, analyze the time series characteristics, extract the parameter changes of key frames, and establish the initial behavior feature set; S2: Based on the initial behavior feature set, compare the motion vectors, pixel intensity values, and color histograms of consecutive frames, calculate the differences, correlations, and covariances, group the parameters of the same distribution, perform cluster analysis, analyze the center value, variance, and range of each cluster, identify outliers, verify the clustering results, adjust the clustering parameters, and establish a normal behavior pattern model; S3: Based on the normal behavior pattern model, monitor the real-time motion vector, pixel intensity value and color histogram of the terminal device, calculate the deviation from the model, determine the fluctuation value of the deviation, identify the parameters exceeding the threshold, mark them as abnormal, record and classify the abnormal parameters, analyze the distribution characteristics, identify the abnormal behavior type, and establish the abnormal detection results; S4: Based on the anomaly detection results, exchange anomaly parameters and detection status between terminal devices, filter parameters that exceed the threshold, merge data from multiple terminal devices, update the normal behavior pattern model, adjust the detection threshold, re-evaluate the anomaly detection conditions, and build a collaborative anomaly detection model.

[0017] The initial behavior feature set specifically includes parameter mean, parameter variance, and key frame changes. The normal behavior pattern model includes cluster center value, cluster parameter variance, and cluster range. The anomaly detection results specifically include anomaly labels, anomaly categories, and anomaly distributions. The collaborative anomaly detection model includes parameter synchronization status, model update records, and threshold adjustment records.

[0018] See also Figure 2 , the specific steps for obtaining the initial behavior feature set are: S111: extracting data of each frame from the surveillance video, including motion vector, pixel intensity and color histogram, calculating features of multiple frames of data, and generating a frame data feature set; Extract the data of each frame from the surveillance video. This step involves extracting the motion vector, pixel intensity and color histogram in the video frame. The acquisition of data relies on image processing technology, including edge detection, color analysis and capture of dynamic changes. Each frame is analyzed to extract detailed information about its content. The motion vector is obtained by tracking the movement between pixels, and the pixel intensity is calculated by calculating the brightness or grayscale level of each pixel. The color histogram reflects the distribution of colors in the frame. The extracted data will form a frame data feature set, which provides basic data for analysis, thereby performing spatiotemporal alignment.

[0019] S112: performing spatiotemporal alignment on the frame data feature set, using a calculation method of timestamp difference and spatial position change to ensure the temporal and spatial continuity of the data, and obtaining a spatiotemporal alignment feature set; The frame data feature set is temporally and spatially aligned. This process ensures that video frames from different time points and spatial locations can be compared and analyzed under the same reference frame. This requires synchronization of time stamps and calibration of spatial positions of video frames. Temporal and spatial alignment involves calculating the difference between timestamps and transforming spatial coordinates to ensure data continuity. By using timestamp and geo-tagging information, feature data extracted from multiple frames can be sorted and aligned in time order and spatial position to generate a temporally and spatially aligned feature set, which provides a data basis for analyzing dynamic changes between frames.

[0020] S113: Statistical analysis is performed on the spatiotemporal alignment feature set using the formula: ; and ; Calculate the mean and variance of each parameter to obtain the parameter statistical feature set; in, represents the average value, represents the variance, represents a data item extracted from the spatiotemporal aligned feature set, representing motion vectors, pixel intensity or color histogram data, is the number of data items, indicating the total number of data involved in the calculation of the mean and variance; formula: ; and ; The benefit of the formula is that it provides basic statistical information for the analysis of time series data by calculating the mean and variance, reflecting the central tendency and dispersion of the data.

[0021] Detailed explanation of the formula and the process of formula calculation and derivation: set up Represent the element values ​​in the spatiotemporal alignment feature set, where , sample data They are (sample data, set with 100 elements), Calculate the mean : ; Calculating variance : ; The results show that the mean of the data is 17.5 and the variance is 4.5, which reflects the dispersion of the data points around the mean and the stability of the data.

[0022] S114: Analyze the time series characteristics in the parameter statistical feature set, identify and extract significant changes in the parameters in the key frames, and establish an initial behavior feature set.

[0023] Through time series analysis of parameter statistical feature sets, this process involves identifying key frames and extracting significant parameter changes in these frames to establish behavioral feature sets. By calculating the statistical indicators of each parameter, including trends, periodicity and seasonal factors, its performance in the time series is analyzed to identify patterns of key events or behaviors. This requires analysis of time series data, including moving average, exponential smoothing or time series analysis methods, to establish an initial behavioral feature set.

[0024] See also Figure 3 , the specific steps for obtaining the normal behavior pattern model are: S211: Based on the initial behavior feature set, compare the motion vectors, pixel intensity values, and color histograms of consecutive frames, calculate the difference between each frame, obtain the difference amount, and generate a difference feature set; Based on the comparative analysis of the initial behavioral feature set, the differences in motion vectors, pixel intensity values ​​and color histograms of consecutive frames are calculated. The differences are obtained by comparing the data changes of adjacent frames frame by frame. This process involves image processing technology. The image recognition algorithm is used to measure the degree of change of the target pixel in each frame. The calculation of the motion vector depends on the position offset between frames. This offset is determined by the tracking algorithm to ensure data acquisition, and the difference in color histograms is obtained through statistical analysis of the color space. This series of operations requires not only algorithm support, but also high-speed processing capabilities of hardware to ensure the processing of each frame of data and the real-time generation of difference data. The core of this process is to quickly and accurately capture the changes in motion and color, which is crucial for subsequent behavioral analysis. The generated difference feature set will provide a basis for the next step of data analysis.

[0025] S212: For the data in the difference feature set, the formula is used: ; Calculate the correlation and covariance to obtain the correlation feature set; in, represent and The correlation coefficient between them is used to evaluate the relationship between the differentiation parameters. Representative A target parameter difference of the frame, Represents another parameter difference of the same frame, and Represents all frames and The average value of formula: ; The benefit of the formula is that it enhances the analysis of the differences and correlations of differentiated data dimensions through the method of standardized covariance.

[0026] Detailed explanation of the formula and the process of formula calculation and derivation: There are three sets of data, corresponding to the motion vector differences: , the pixel intensity difference is , for the correlation calculation of motion vector and pixel intensity: Calculate the average , ; calculate and : ; Calculate the numerator: ; Calculate the denominator: ; calculate ; The results show that there is a very high correlation between motion vectors and pixel intensities, which is very useful when analyzing dynamically changing features in video data, because high correlation indicates the existence of common behavior patterns.

[0027] S213: Based on the correlation feature set, a clustering algorithm is applied to group the parameters, the central value, variance and range of each cluster are analyzed, clustering is performed, and a clustering result set is obtained; By analyzing the clustering result set, especially the statistics of the center value, variance and range of each cluster, the clustering algorithm not only identifies the inherent pattern of the data, but also determines the probability boundary of the pattern. This statistical analysis involves mathematical calculations, including variance analysis and range estimation. The center value is determined by finding the mean of the data points in multiple classes. The variance gives the degree to which the data points deviate from the center value, and the range defines the maximum fluctuation limit of the data points in each cluster. The statistical parameters provide a scientific basis for analyzing the stability and reliability of clustering. Through analysis, the clustering parameters are adjusted, the performance of the model is optimized, and it is ensured that the model can reflect the true pattern of the behavioral data. The generated clustering result set becomes the basis for subsequent abnormal behavior detection and behavior prediction.

[0028] S214: Identify and verify outliers in the clustering result set, optimize the model by adjusting clustering parameters, and establish a normal behavior pattern model.

[0029] The identification of outliers and the adjustment of clustering parameters is a key process. Through the analysis of data outliers, noise in the data or unexpected behavior patterns can be identified. This process includes comparing the distance between each data point and its cluster center. If the distance exceeds the set threshold, the data point is marked as an anomaly. The adjustment of clustering parameters is based on the distribution of outliers, aiming to improve the sensitivity and accuracy of the clustering algorithm. Through these operations, the established normal behavior pattern model can better match the actual distribution of the data, ensure the effectiveness and applicability of the model, and provide a solid foundation for behavior prediction.

[0030] See also Figure 4 , the specific steps for obtaining anomaly detection results are: S311: monitor the real-time motion vector, pixel intensity value and color histogram of the terminal device, calculate the deviation between the current frame and the model based on the normal behavior pattern model, and generate a deviation feature set; Monitor real-time data to evaluate the difference between device behavior and the established normal behavior model. This process first involves the acquisition and transmission of real-time data, including the motion vector, pixel intensity value and color histogram of the terminal device. The data is collected in real time by a variety of sensors and cameras and sent to the processing center through the communication network. The processing center analyzes the data stream in real time and compares the difference between each frame of data and the normal behavior model. In order to ensure the real-time and accuracy of data processing, a high-speed data processor and real-time operating system are used to reduce latency, and optimized algorithms are used to improve data processing efficiency. In this way, any behavior that deviates from the normal mode can be immediately discovered to provide a basis for subsequent abnormal judgment.

[0031] S312: For the data in the deviation feature set, the statistical method formula is used: ;

[0032] Calculate the fluctuation value of the deviation and generate a fluctuation value result set; in, represents a single deviation value from the deviation feature set, used to calculate the difference from the mean deviation, Represents the average of the deviation values, which is used to evaluate the central position of the overall deviation. is the number of samples, indicating the total number of elements in the deviation feature set, used for normalized fluctuation value calculation, The fluctuation value representing the deviation is used to quantify the stability of the deviation and assist in identifying abnormal fluctuations; formula: ;

[0033] The benefit of the formula is that by calculating the standard deviation of the deviation value, the stability of each behavioral deviation can be quantified, which helps determine whether the deviation is random or systematic, and enhances the sensitivity to abnormal behavior. Detailed explanation of the formula and the process of formula calculation and derivation: The following deviation data was collected within one hour. , the mean deviation Calculated as: ;

[0034] Substitute into the formula: ; ; ; ; ;

[0035] The results show that the current data fluctuation degree is 2.628, indicating that the data fluctuates greatly, which indicates abnormal behavior or equipment failure, and needs to be checked and verified to determine the type of abnormality.

[0036] S313: Identify parameters that exceed a preset threshold from the fluctuation value result set, mark the parameters as abnormal, and generate an abnormal marking set by combining the fluctuation value and the threshold judgment; Identifying and marking abnormal parameters is a task that requires real-time response. First, the system will compare the fluctuation values ​​calculated previously with the set thresholds. These thresholds are set based on data analysis and expert knowledge to ensure that real abnormalities rather than occasional data fluctuations can be effectively identified. Once a parameter exceeding the threshold is detected, the system will immediately trigger an alarm and send the data to the monitoring center through an interface. The operator at the monitoring center will conduct a preliminary analysis of the data to determine whether technical intervention or adjustment of equipment settings is required. The data recording and classification system used in this process can assist technicians in identifying the type and frequency of problems and provide data support for equipment maintenance and optimization.

[0037] S314: Based on the abnormal tag set, record and classify abnormal parameters, identify abnormal behavior types by analyzing the distribution characteristics of the parameters, and establish abnormal detection results.

[0038] In the stage of identifying abnormal behavior types and establishing abnormal detection results, the system utilizes the statistical analysis results of abnormal label sets and multiple parameters. Analyzing the distribution characteristics involves sorting, classifying and frequency analyzing abnormal data to determine the type and severity of abnormal behavior. The algorithm can learn and extract information from the data. The output anomaly detection results not only mark the anomaly type, but also predict development trends, which is critical for taking preventive measures and reducing potential negative impacts. In addition, the system will update the behavior model based on these analyses to enhance its ability to predict future abnormal behaviors and provide a basis for operation and decision-making for the maintenance team.

[0039] See also Figure 5 ,The specific steps for obtaining the collaborative anomaly detection model are: S411: Based on the abnormality detection result, the terminal devices exchange key abnormality parameters and the current detection status, implement data synchronization through the defined communication protocol, and generate the data status after the exchange; Based on the anomaly detection results, data exchange is carried out to ensure the consistency and timeliness of multi-terminal information. First, the terminal devices synchronize the anomaly parameters and detection status through the defined network communication protocol. This process involves the assembly, encryption and sending of data packets. After receiving the data, each terminal decrypts and verifies the data to verify the integrity and authenticity. At this stage, the key is to use data integrity verification algorithms and timestamp verification to ensure the security and accuracy of data synchronization. Then, the received data is logged, including the source of the data, timestamp, and the value of the anomaly parameters, for subsequent analysis and auditing. Through the exchange and recording process, the system can not only respond when anomalies are mined, but also track the time and location of the anomaly, enhancing the overall reliability and efficiency of the network security monitoring system.

[0040] S412: from the exchanged data state, all abnormal parameters exceeding a preset threshold are screened by a comparison algorithm to ensure that only key data affecting system performance is processed, and a screened abnormal parameter set is generated; The process of screening abnormal parameters from the data status involves the application of multiple data processing technologies. To screen abnormal parameters that exceed the preset threshold, we first define the threshold and perform preliminary screening on the data, including comparing the current values ​​of multiple parameters with the threshold, using conditional judgment statements for initial screening, and then performing secondary verification on the screened data, including calculating the moving average or standard deviation of the parameters to determine whether they really show an abnormal trend. This step is to reduce false alarms and improve the accuracy of the system. Then, the verified abnormal parameters are sorted and classified, which involves sorting algorithms and classification algorithms. The data is marked as key anomalies to provide a basis for subsequent processing. This processing flow shows the entire process from data collection to anomaly verification, ensuring the rigor of processing and the accuracy of data.

[0041] S413: merging the screened abnormal parameter set with the data of multiple terminals, applying data fusion technology to integrate information from differentiated sources, updating the normal behavior pattern model, and adjusting the detection threshold inside the model to match the current data behavior, thereby generating an updated model; In the process of merging data from multiple terminal devices and updating the normal behavior pattern model, first collect abnormal parameter data from multiple terminals, including data integration and preprocessing, such as data formatting and cleaning, to ensure data consistency and availability, and then use weighted average or multi-layer clustering analysis to integrate data from differentiated sources to enhance the representativeness and accuracy of the data. Subsequently, update the behavior pattern model based on the merged data set, including learning rate, error threshold, and use support vector machine or neural network to retrain the model to match the current data behavior. Finally, test and verify the updated model to ensure the effectiveness and adaptability of the model. The entire process illustrates each step of the model update to ensure the accuracy and effectiveness of the model update.

[0042] S414: Using the updated model, re-evaluate and optimize the anomaly detection conditions, using the statistical analysis formula: ;

[0043] Building collaborative anomaly detection models; in, Represents the value of each monitored abnormal parameter, which is used to determine the degree of deviation of the data point. Represents the average value of the abnormal parameter, which is used to measure the central position of a set of data. Represents the standard deviation of the parameter, which is used to describe the dispersion of the data. Adjust the significance of deviations from the mean, Adjust the influence of standard deviation to control the contribution of deviation from the mean and data fluctuation to anomaly detection. Represents the total number of data points involved in the calculation, used to obtain the average of the weighted sum and balance the impact of differentiated terminal device data points. Represents collaborative anomaly detection metrics.

[0044] formula: ;

[0045] The benefit of the formula is that by incorporating the deviation and fluctuation of the anomaly parameters, the adjusted model is more sensitive and accurate, can match the cyclically changing data behavior, and improve the efficiency and accuracy of anomaly detection.

[0046] Detailed explanation of the formula and the process of formula calculation and derivation: Set up with 5 data points: 10, 12, 9, 11, 8, average , standard deviation ,set up , , The calculation formula is as follows: ; ; ; ;

[0047] The results show that according to the given data and parameter settings, the collaborative anomaly detection index calculated by the model is 0.843. The result reflects the sensitivity and responsiveness of the current anomaly detection, which helps to adjust the thresholds and parameters in time and improve the early warning capability of the system.

[0048] A terminal device collaborative video monitoring system, the terminal device collaborative video monitoring system is used to execute the terminal device collaborative video monitoring method, the system comprises: The data acquisition module extracts the motion vector, pixel intensity value and color histogram of each frame based on the terminal device monitoring video data, performs spatiotemporal alignment, calculates the mean and variance of the parameters, extracts the parameter changes of the key frames, and establishes the initial behavior feature set; The feature analysis module compares the associated parameters of consecutive frames based on the initial behavior feature set, calculates the difference and covariance, groups and clusters the parameters, analyzes the central value and range of the clusters, identifies and excludes abnormal points, verifies the results and adjusts the clustering parameters, and establishes a normal behavior pattern model; The behavior modeling module is based on the normal behavior pattern model, monitors real-time parameters, calculates deviations from the model, determines the fluctuation value of the deviation, identifies parameters that exceed the threshold, marks anomalies, records and classifies abnormal parameters, analyzes distribution characteristics, and establishes anomaly detection results; Based on the anomaly detection results, the anomaly detection module exchanges anomaly parameters and detection status between terminal devices, filters and synchronizes parameters that exceed the threshold, merges data, updates the normal behavior pattern model, adjusts the threshold, re-evaluates the detection conditions, and establishes a collaborative anomaly detection model; The collaborative update module is based on the collaborative anomaly detection model, integrates multiple terminal data, synchronously updates the model and detection status, adjusts the threshold, and records the system update status.

[0049] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A terminal device collaborative video monitoring method, characterized in that: The following steps are involved: Based on the terminal device monitoring video data, the motion vector, pixel intensity value and color histogram of each frame are extracted, time-space alignment is performed, the mean and variance of each parameter are calculated, the time series characteristics are analyzed, the parameter changes of the key frames are extracted, and the initial behavior feature set is established; Based on the initial behavior feature set, the motion vectors, pixel intensity values ​​and color histograms of consecutive frames are compared, the differences, correlations and covariances are calculated, the parameters of the same distribution are grouped, cluster analysis is performed, the central value, variance and range of each cluster are analyzed, outliers are identified, clustering results are verified, clustering parameters are adjusted, and a normal behavior pattern model is established; Based on the normal behavior pattern model, monitor the real-time motion vector, pixel intensity value and color histogram of the terminal device, calculate the deviation from the model, determine the fluctuation value of the deviation, identify the parameters exceeding the threshold, mark them as abnormal, record and classify the abnormal parameters, analyze the distribution characteristics, identify the abnormal behavior type, and establish the abnormal detection result; Based on the anomaly detection results, abnormal parameters and detection status are exchanged between terminal devices, parameters exceeding the threshold are screened, data of multiple terminal devices are merged, the normal behavior pattern model is updated, the detection threshold is adjusted, the anomaly detection conditions are re-evaluated, and a collaborative anomaly detection model is constructed.

2. The terminal device cooperative video monitoring method according to claim 1, characterized in that: The steps for obtaining the initial behavior feature set are specifically as follows: Extract the data of each frame from the surveillance video, including motion vector, pixel intensity and color histogram, calculate the features of multiple frames of data, and generate a frame data feature set; Performing spatiotemporal alignment on the frame data feature set, using a calculation method of timestamp difference and spatial position change to ensure the temporal and spatial continuity of the data, and obtaining a spatiotemporal alignment feature set; The spatiotemporal alignment feature set is statistically analyzed using the formula: ； and ； Calculate the mean and variance of each parameter to obtain the parameter statistical feature set; in, represents the average value, represents the variance, represents a data item extracted from the spatiotemporal aligned feature set, representing motion vectors, pixel intensity or color histogram data, is the number of data items, indicating the total number of data involved in the calculation of the mean and variance; The time series characteristics in the parameter statistical feature set are analyzed, significant changes in parameters in key frames are identified and extracted, and an initial behavior feature set is established.

3. The terminal device cooperative video monitoring method according to claim 2, characterized in that: The steps for obtaining the normal behavior pattern model are specifically as follows: Based on the initial behavior feature set, the motion vectors, pixel intensity values ​​and color histograms of consecutive frames are compared, and the difference between each frame is calculated to obtain the difference amount, thereby generating a difference feature set; For the data in the difference feature set, the formula is used: ； Calculate the correlation and covariance to obtain the correlation feature set; in, represent and The correlation coefficient between them is used to evaluate the relationship between the differentiation parameters. Representative A target parameter difference of the frame, Represents another parameter difference of the same frame, and Represents all frames and The average value of Based on the correlation feature set, a clustering algorithm is applied to group the parameters, the central value, variance and range of each cluster are analyzed, clustering is performed, and a clustering result set is obtained; The clustering result set is subjected to outlier identification and verification, and the model is optimized by adjusting clustering parameters to establish a normal behavior pattern model.

4. The terminal device cooperative video monitoring method according to claim 3, characterized in that: The steps for obtaining the abnormal detection result are specifically as follows: Monitoring the real-time motion vector, pixel intensity value and color histogram of the terminal device, calculating the deviation between the current frame and the model based on the normal behavior pattern model, and generating a deviation feature set; For the data in the deviation feature set, the statistical method formula is used: ； Calculate the fluctuation value of the deviation and generate a fluctuation value result set; in, represents a single deviation value from the deviation feature set, used to calculate the difference from the mean deviation, Represents the average of the deviation values, used to evaluate the central position of the overall deviation. is the number of samples, indicating the total number of elements in the deviation feature set, used for normalized fluctuation value calculation, The fluctuation value representing the deviation is used to quantify the stability of the deviation and assist in identifying abnormal fluctuations; Identify parameters that exceed a preset threshold from the fluctuation value result set, mark the parameters as abnormal, and generate an abnormal marking set by combining the fluctuation value and the threshold judgment; Based on the abnormal tag set, abnormal parameters are recorded and classified, and the abnormal behavior type is identified by analyzing the distribution characteristics of the parameters to establish an abnormal detection result.

5. The terminal device cooperative video monitoring method according to claim 4, characterized in that: The steps for obtaining the collaborative anomaly detection model are specifically as follows: Based on the abnormality detection result, the terminal devices exchange key abnormality parameters and the current detection status, implement data synchronization through the defined communication protocol, and generate the data status after the exchange; From the exchanged data state, all abnormal parameters exceeding a preset threshold are screened by a comparison algorithm to ensure that only key data affecting system performance is processed, and a screened abnormal parameter set is generated; Merging the screened abnormal parameter set with data from multiple terminals, applying data fusion technology to integrate information from differentiated sources, updating the normal behavior pattern model, and adjusting the detection threshold within the model to match the current data behavior to generate an updated model; Using the updated model, the anomaly detection conditions are re-evaluated and optimized, using the statistical analysis formula: ； Building collaborative anomaly detection models; in, Represents the value of each monitored abnormal parameter, which is used to determine the degree of deviation of the data point. Represents the average value of the abnormal parameter, which is used to measure the central position of a set of data. Represents the standard deviation of the parameter, which is used to describe the dispersion of the data. Adjust the significance of deviations from the mean, Adjust the influence of standard deviation to control the contribution of deviation from the mean and data fluctuation to anomaly detection. Represents the total number of data points involved in the calculation, used to obtain the average of the weighted sum and balance the impact of differentiated terminal device data points. Represents collaborative anomaly detection metrics.

6. A terminal device cooperative video monitoring system, characterized in that: According to the terminal device collaborative video surveillance method according to any one of claims 1 to 5, the system comprises: The data acquisition module extracts the motion vector, pixel intensity value and color histogram of each frame based on the terminal device monitoring video data, performs spatiotemporal alignment, calculates the mean and variance of the parameters, extracts the parameter changes of the key frames, and establishes the initial behavior feature set; The feature analysis module compares the associated parameters of consecutive frames based on the initial behavior feature set, calculates differences and covariances, performs parameter grouping and clustering, analyzes the central value and range of the clusters, identifies and excludes abnormal points, verifies the results and adjusts the clustering parameters, and establishes a normal behavior pattern model; The behavior modeling module monitors real-time parameters based on the normal behavior pattern model, calculates deviations from the model, determines the fluctuation value of the deviation, identifies parameters exceeding the threshold, marks anomalies, records and classifies abnormal parameters, analyzes distribution characteristics, and establishes anomaly detection results; The anomaly detection module exchanges anomaly parameters and detection status between terminal devices based on the anomaly detection results, filters and synchronizes parameters exceeding thresholds, merges data, updates the normal behavior pattern model, adjusts thresholds, re-evaluates detection conditions, and establishes a collaborative anomaly detection model; The collaborative update module integrates multiple terminal data based on the collaborative anomaly detection model, synchronously updates the model and detection status, adjusts the threshold, and records the system update status.

Images