Cloud-based video behavior monitoring system

Abstract

The invention relates to the technical field of video monitoring, in particular to a cloud-based video behavior monitoring system, which comprises a video bandwidth control module, a behavior calculation scheduling module, a monitoring area adjustment module, an environment task optimization module and a monitoring resource allocation module. According to the invention, through real-time bandwidth state monitoring and data analysis, dynamic matching of video data volume, improvement of stability and quality of video transmission, optimization of task scheduling accuracy based on pixel motion analysis and target activeness evaluation, reasonable distribution of monitoring resources, statistics of behavior density and judgment of trend are realized, and priority uploading of high-density area data is ensured. According to the method and the system, monitoring data imbalance is avoided, network flow detection and bandwidth recovery identification are combined, intelligent task sorting is realized, the resource utilization rate is improved, low-priority task allocation is optimized through calculation load detection and resource analysis, so that monitoring is stably operated in a high-load environment, and the real-time performance, the intelligent level and the accuracy of abnormal behavior detection are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of video surveillance, and in particular, to a cloud-based video behavior monitoring system. Background Art

[0002] The technical field of video surveillance includes technologies for real-time monitoring, analysis, recognition, and storage of targets through electronic devices, and is widely applied in multiple fields such as security prevention, public security, traffic management, and mall surveillance. The core contents of this field include video acquisition, image processing, data storage, real-time monitoring, anomaly detection, and behavior recognition. With the continuous improvement of network technology, storage technology, and computing power, video surveillance is gradually developing towards intelligence and cloudification, emphasizing the efficient analysis and processing of surveillance data to improve the surveillance effect and response speed.

[0003] Among them, a video behavior monitoring system refers to a system that uses real-time video data collected by video surveillance devices to perform behavior analysis on the monitored content through a cloud platform. This system mainly focuses on how to identify the behavior patterns of monitored targets through video stream analysis, improve the capabilities of traditional video surveillance in behavior recognition and abnormal event handling, use the cloud platform for data storage and computing, implement behavior detection and recognition in the video stream based on image analysis technology, and return the processing results to the terminal device. By monitoring the behavior of people in a specific area, abnormal behaviors can be identified and alarms can be issued in a timely manner.

[0004] Existing technologies have deficiencies in multiple aspects in the field of video surveillance. Traditional systems use fixed coding parameters in bandwidth management and cannot dynamically adjust according to changes in network status, resulting in a decline in video quality or even data loss when the network condition is poor. In terms of task scheduling, conventional methods are mostly based on fixed frame rates and frame update strategies and fail to flexibly adjust according to target activity and pixel changes, resulting in resources in key monitoring areas being occupied by low-priority tasks. The adjustment of the monitoring area depends on manual settings or static parameters, lacking dynamic recognition of the target behavior density and easily causing low data transmission efficiency. In the case of network traffic fluctuations, existing methods are difficult to accurately predict the bandwidth recovery trend, resulting in a lack of optimization in the execution order of video tasks and affecting the system's response speed to emergencies. The server resource management method is relatively extensive and fails to fully optimize the task execution order in combination with the computing load situation, causing resource waste or task congestion, thereby reducing the system's stability and the integrity of surveillance data, affecting the adaptability of video surveillance in complex environments, and making it difficult to provide high-quality surveillance services under high load or network instability. Summary of the Invention

[0005] The purpose of the present invention is to solve the deficiencies existing in the prior art and to propose a cloud-based video behavior monitoring system.

[0006] To achieve the above object, the present invention adopts the following technical solutions: A cloud-based video behavior monitoring system includes: The video bandwidth control module captures bandwidth data based on the bandwidth status of the video transmission channel, analyzes the time interval and change amplitude of the bandwidth change, adjusts the video frame compression ratio and data volume, monitors the adjusted video transmission status, and obtains the bandwidth-matched video data volume; The behavior calculation scheduling module detects target objects in the video monitoring screen based on the bandwidth-matched video data volume, analyzes the pixel movement amplitude of the target objects, judges the activity of the target objects, adjusts the task priority of the target area, and obtains the video resource volume of the target area; The monitoring area adjustment module calls the video resource volume of the target area, identifies the target behavior density of each area, judges the change trend of the target behavior density, adjusts the video data upload rate of the low-density area, and obtains the area monitoring video stream; The environment task optimization module detects the network traffic bandwidth status according to the area monitoring video stream, identifies the bandwidth recovery speed, judges the stability of the bandwidth recovery, adjusts the task execution order, and obtains the video task execution sorting.

[0007] As a further solution of the present invention, the bandwidth-matched video data volume includes compression ratio parameters, data stream transmission rate, and video quality evaluation index. The video resource volume of the target area includes activity weight, priority coefficient, and pixel change amplitude. The area monitoring video stream includes behavior density parameters, upload rate threshold, and monitoring coverage. The video task execution sorting includes recovery speed index, bandwidth stability coefficient, and task priority sequence.

[0008] As a further solution of the present invention, the video bandwidth control module includes: The bandwidth status capture sub-module captures bandwidth data during different time periods based on the bandwidth status of the video transmission channel, detects the bandwidth utilization rate and data transmission rate of each time period, analyzes the bandwidth change rate between adjacent time periods, and records the bandwidth fluctuation amplitude to obtain a bandwidth fluctuation data set; The bandwidth trend analysis sub-module identifies the bandwidth change interval between adjacent time periods based on the bandwidth fluctuation data set, analyzes the fluctuation frequency of the bandwidth change, and compares the bandwidth stability of different time intervals to obtain bandwidth change trend data; The video data adjustment sub-module calls the bandwidth change trend data, analyzes the video compression ratio matching situation under different bandwidth states, adjusts the video frame compression ratio and transmission data volume, monitors the adjusted video transmission status, and uses the formula: ; Compare the video quality corresponding to the differential encoding parameters, screen the optimal set of encoding parameters, and obtain the video data volume matching the bandwidth; Among them, represents the video data volume matching the bandwidth, represents the bandwidth utilization rate in the i-th time period, represents the video compression ratio in the i-th time period, represents the video transmission time in the i-th time period, represents the video data volume in the i-th time period, represents the total number of time periods.

[0009] As a further solution of the present invention, the behavior calculation and scheduling module includes: The target object detection sub-module extracts the video surveillance screen data based on the video data volume matching the bandwidth, detects multiple target objects in the surveillance screen, distinguishes the background pixels from the target object pixels, extracts the pixel point data of the target object, and obtains the target object pixel distribution data; The pixel motion amplitude analysis sub-module identifies the pixel offset of the target object in the consecutive frame images based on the target object pixel distribution data, statistically analyzes the displacement information, analyzes the pixel change amplitude, and uses the formula: ; Obtain the target object pixel motion trend value; Among them, represents the target object pixel motion trend value, represents the position coordinates of the target object in the -th frame, represents the position coordinates of the target object in the -th frame, represents the total number of frames within the calculation range; The target area task adjustment sub-module compares the pixel change rate of the target object based on the target object pixel motion trend value, screens the active target areas, adjusts the video resource priority, and obtains the video resource volume of the target area.

[0010] As a further solution of the present invention, the monitoring area adjustment module includes: The video resource statistics sub-module calls the video resource volume of the target area, identifies the number of video frames, storage occupancy, and effective monitoring duration, screens the areas below the set threshold, and obtains the statistical value of the video resource volume of the area; The behavior density identification sub-module calls the statistical value of the video resource volume of the area, counts the number of occurrences of the target behavior, identifies the behavior frequency per unit time, the coverage rate per unit area of the area, and the behavior duration, and uses the formula: ; Calculate the target behavior density, screen the areas where the behavior density is lower than the threshold, and obtain the distribution of the target behavior density; Among them, represents the target behavior density, represents the frequency of the target behavior per unit time, represents the unit area coverage rate of the target behavior in the monitored area, represents the average duration of the behavior in the area, represents the statistical value of the regional video resource volume, represents the average value of the video resource volume in the entire region; The monitoring video data upload sub-module calls the distribution of the target behavior density, judges the change trend of the target behavior density, adjusts the video upload rate of the low-density area, reduces the data transmission volume, and obtains the regional monitoring video stream.

[0011] As a further solution of the present invention, the environmental task optimization module includes: The bandwidth status monitoring sub-module extracts the network traffic bandwidth information according to the regional monitoring video stream, monitors the change trend of the bandwidth occupancy, analyzes the short-term fluctuation range, judges the current bandwidth status, and obtains the bandwidth occupancy fluctuation amplitude; The recovery stability judgment sub-module extracts the recovery rates of multiple time periods during the bandwidth recovery process based on the bandwidth occupancy fluctuation amplitude, and uses the formula: ; Calculate the recovery rate of the time period, judge whether the recovery rate remains stable, and obtain the bandwidth recovery stability index; Among them, represents the recovery rate of the time period, represents the time point of the bandwidth occupancy value, represents the time point of the bandwidth occupancy value, represents the time point of the bandwidth recovery duration, represents the time point of the bandwidth recovery duration; The task priority adjustment sub-module screens the video tasks in the active area based on the bandwidth recovery stability index, adjusts the task priority according to the recovery stability and the urgency of the task, and obtains the execution order of the video tasks.

[0012] As a further solution of the present invention, the system further includes a monitoring resource allocation module: The monitoring resource allocation module calls the execution order of the video tasks, detects the computing load situation of the server, identifies the available amount of resources, judges the load upper limit of the resources, screens the low-priority tasks, optimizes the allocation of resources, and obtains the resource scheduling volume of the video behavior monitoring tasks; The resource scheduling volume of the video behavior monitoring task includes a computing load threshold, an available resource quota, and a load upper limit parameter.

[0013] As a further solution of the present invention, the monitoring resource allocation module includes: The computing load detection sub-module calls the video task execution sorting, detects the CPU occupancy rate, memory usage rate, and the number of task execution threads, identifies the task occupancy ratio, judges the load balance situation, and obtains the server computing load ratio; The available resource amount identification sub-module extracts the available computing resource amount based on the server computing load ratio, filters the allocable computing resources, identifies the task load occupancy threshold, and uses the formula: ; Compares the available resources with the task requirements to obtain the available computing resource allocation amount; Wherein, represents the available computing resource allocation amount, represents the total computing resources of the server, represents the computing resource occupancy of the jth task, represents the execution time of the jth task, represents the average value of the task execution time, represents the total number of tasks currently being executed; The video task scheduling optimization sub-module judges the task priority based on the available computing resource allocation amount, filters the low-priority tasks for postponement and merging, optimizes the resource allocation structure, and obtains the resource scheduling volume of the video behavior monitoring task.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, through the real-time monitoring and data analysis of the bandwidth status, the dynamic matching of the video data volume is realized, the stability and quality of video transmission are improved. Based on the pixel motion analysis of the video picture and the evaluation of the target activity, the accuracy of task scheduling is optimized, so that the monitoring resources are more effectively allocated. The behavior density of the monitoring area is statistically analyzed and the trend is judged to ensure that the video data in the high-density area is uploaded first, avoiding the uneven distribution of monitoring data. Combining the real-time detection of network traffic and the dynamic identification of bandwidth recovery, the intelligent sorting of video tasks is realized, and the resource utilization efficiency is improved. In the management of server computing resources, through computing load detection and available resource amount analysis, the allocation method of low-priority tasks is optimized, so that the monitoring under high load can still run stably. Through optimization measures, the real-time and intelligent level of video monitoring is improved. In the environment of network fluctuations and resource constraints, the integrity and clarity of the monitoring video are ensured, and at the same time, the accuracy and response speed of abnormal behavior detection are improved. Description of the Drawings

[0015] Figure 1 is the system flow chart of the present invention; Figure 2 is the flow chart of the video bandwidth control module in the present invention; Figure 3 is the flow chart of the behavior calculation scheduling module in the present invention; Figure 4 is the flow chart of the monitoring area adjustment module in the present invention; Figure 5 is the flow chart of the environment task optimization module in the present invention; Figure 6 is the flow chart of the monitoring resource allocation module in the present invention. Detailed implementation manners

[0016] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to 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 used to limit the present invention.

[0017] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is 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 should not be construed as a limitation of the present invention. In addition, in the description of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0018] Please refer to Figure 1 , a cloud-based video behavior monitoring system includes: The video bandwidth control module captures bandwidth data in different differential time periods based on the bandwidth status of the video transmission channel, analyzes the time interval and change amplitude of the bandwidth change, records the bandwidth change trend value, adjusts the video frame compression ratio and data volume, monitors the adjusted video transmission status, analyzes the impact of the adjustment on the video quality, screens the optimal coding parameters, and obtains the bandwidth-matched video data volume; The behavior calculation scheduling module detects the target objects in the video monitoring screen based on the bandwidth-matched video data volume, analyzes the pixel movement amplitude of the target objects, compares the pixel change trends, judges the activity of the target objects, adjusts the task priority of the target area, and obtains the video resource volume of the target area; The monitoring area adjustment module calls the video resource volume of the target area, counts the occurrence times of the target behavior in the monitoring screen, identifies the target behavior density of each area, judges the change trend of the target behavior density, adjusts the video data upload rate of the low-density area, and obtains the area monitoring video stream; The environmental task optimization module detects the network traffic bandwidth status according to the area monitoring video stream, identifies the bandwidth recovery speed, judges the stability of the bandwidth recovery, screens the video processing tasks in the active area, adjusts the task execution order, and obtains the video task execution sorting; The monitoring resource allocation module calls the video task execution sorting, detects the server computing load situation, identifies the available amount of resources, judges the load upper limit of the resources, screens the low-priority tasks, optimizes the resource allocation, and obtains the resource scheduling volume of the video behavior monitoring task.

[0019] The bandwidth matching video data volume includes compression ratio parameters, data stream transmission rate, video quality evaluation index. The video resource volume of the target area includes activity weight, priority coefficient, pixel change amplitude. The area monitoring video stream includes behavior density parameters, upload rate threshold, monitoring coverage. The video task execution sorting includes recovery speed index, bandwidth stability coefficient, task priority sequence. The resource scheduling volume of the video behavior monitoring task includes computing load threshold, available resource quota, load upper limit parameter.

[0020] Please refer to Figure 2 , the video bandwidth control module includes: The bandwidth status capture sub-module captures the bandwidth data in different time periods based on the bandwidth status of the video transmission channel, detects the bandwidth utilization rate and data transmission rate in each time period, analyzes the bandwidth change rate in adjacent time periods, and records the bandwidth fluctuation amplitude to obtain the bandwidth fluctuation data set; In network video surveillance, bandwidth usage data is periodically obtained from each surveillance node. For example, the bandwidth usage is recorded every 5 minutes. The data includes real-time data transmission rate and bandwidth utilization rate. For example, if the bandwidth utilization rate of a certain surveillance point continuously exceeds 80% during the working period (such as from 9 am to 11 am), it is marked as a high bandwidth usage period. Detect the bandwidth utilization rate and data transmission rate for each time period. This not only involves data recording but also requires analyzing the data fluctuations. For example, the stability of the bandwidth is evaluated by calculating the average value and standard deviation of the bandwidth usage for each time period. Calculate the bandwidth change rate between adjacent time periods using a simple difference method, that is, subtracting the bandwidth utilization rate of the previous time period from that of the next time period to obtain the increasing or decreasing trend of the bandwidth. Record the bandwidth fluctuation amplitude for each time period, and judge the fluctuation amplitude by the difference between the maximum and minimum values of the bandwidth utilization rate. Finally, obtain the bandwidth fluctuation data set and integrate the data to form a comprehensive description of the network state. The data can be used for further network optimization and capacity planning, thereby optimizing the quality and efficiency of video transmission.

[0021] Based on the bandwidth fluctuation data set, the bandwidth trend analysis sub-module identifies the bandwidth change intervals between adjacent time periods, analyzes the fluctuation frequency of the bandwidth change, compares the bandwidth stability in different differential time intervals, and obtains the bandwidth change trend data. By marking the time points of each bandwidth change and calculating the time difference between the time points to determine the frequency of the bandwidth change, analyze the fluctuation frequency of the bandwidth change, which can be achieved through spectrum analysis methods. For example, perform Fourier transform on the time series data to obtain the intensity of different frequency components, which helps to identify the main periodic characteristics of the bandwidth change. Calculate the bandwidth change trend value. The moving average or exponential smoothing method can be used to smooth the bandwidth data to reduce the influence of short-term fluctuations and obtain a more stable bandwidth trend line. Compare the bandwidth stability in different time intervals by calculating the variance or standard deviation of the bandwidth stability in different time intervals to evaluate the fluctuation degree of the bandwidth. Screen the set of bandwidth change intervals by setting a threshold, such as setting a certain percentage of the standard deviation as the threshold, and screen the time intervals with bandwidth stability lower than this threshold as the key intervals of bandwidth fluctuation. Finally, obtain the bandwidth change trend data and integrate the data into the final bandwidth management report to provide decision support for network administrators.

[0022] The video data adjustment sub-module calls the bandwidth change trend data, analyzes the video compression ratio matching situation under different bandwidth states, adjusts the video frame compression ratio and the amount of transmitted data, and monitors the adjusted video transmission state. Use the formula: ; Compare the video quality corresponding to different differential coding parameters, screen the optimal set of coding parameters, and obtain the video data volume matching the bandwidth. Among them, represents the amount of video data matching the bandwidth, represents the bandwidth utilization rate in the i-th time period, represents the video compression ratio in the i-th time period, represents the video transmission time in the i-th time period, represents the amount of video data in the i-th time period, represents the total number of time periods; Analyze the adaptation of the video compression ratio under different bandwidth states, which requires adjusting the video compression ratio according to the real-time monitoring results of bandwidth data to adapt to the network transmission capacity. Assume that the average utilization rate of the bandwidth of a certain monitoring system is 75% during the peak period (such as 8:00 - 10:00 in the morning), and the average utilization rate is 30% during the low period (such as 2:00 - 4:00 in the early morning). During the peak period, in order to ensure video smoothness, the compression ratio needs to be increased, such as from 5:1 to 8:1, and during the low period, the compression ratio can be decreased, such as from 5:1 to 3:1. Calculate the amount of video data matching the bandwidth. Assume that the bandwidth utilization rate in a certain time period is and the video frame compression ratio is and the video transmission time is seconds, and the amount of single-frame video data is MB; Substitute the values for calculation: ; Operate to obtain the video data matching degree. By comparing the adaptation of different compression ratios and bandwidth data, the best matching parameters can be obtained. Compare the video quality corresponding to different coding parameters. For example, in case, the video frame rate decreases, but the network transmission is more stable, and in case, the picture quality is better but there will be network fluctuations. Screen the optimal coding parameter set. Assume that finally is selected for use during the peak period, and is selected for use during the low period as the best parameters, and finally the amount of video data matching the bandwidth is obtained, that is, the amount of video data most suitable for transmission under this bandwidth condition; By combining the bandwidth utilization rate, video compression ratio, video data volume, and transmission time, calculate the most suitable video data matching degree in different network environments, so as to dynamically adjust video parameters to ensure that a reasonable video quality can still be maintained under bandwidth constraints, and provide a higher-definition video stream when the bandwidth is sufficient. This result shows that when the bandwidth utilization rate is high, appropriately increasing the compression ratio can ensure the stability of transmission, and during periods of low bandwidth, compression can be reduced to provide higher-quality video, thereby optimizing the use efficiency of video bandwidth.

[0023] Please refer to Figure 3 The behavior calculation scheduling module includes: The target object detection sub-module extracts video surveillance footage data based on the bandwidth-matched video data volume, detects multiple target objects in the surveillance footage, differentiates background pixels from target object pixels, extracts the pixel point data of the target objects, and obtains the target object pixel distribution data; Through the acquisition of video data, the preliminary analysis of the surveillance footage is initiated. This involves receiving data from multiple camera sources and differentiating the dynamic and static parts in the footage in real time. For example, in a parking lot surveillance, vehicles can be differentiated from the background such as buildings and roads, and the vehicles are regarded as dynamic target objects. This process requires the use of image processing techniques to identify different pixel point data, and then continue to record the position and size of each identified target object. The data will be used for subsequent behavior analysis and alarm of security incidents. This classification process not only improves the efficiency of surveillance but also reduces the frequency of false alarms. In this way, more accurate security surveillance services can be provided for shopping malls, parking lots, etc., and the target object pixel distribution data is obtained.

[0024] Based on the target object pixel distribution data, the pixel motion amplitude analysis sub-module identifies the pixel offset of the target object in consecutive frame images, statistics the displacement information, analyzes the pixel change amplitude, and uses the formula: ; Obtain the target object pixel motion trend value; Wherein, represents the target object pixel motion trend value, represents the position coordinates of the target object in the th frame, represents the position coordinates of the target object in the th frame, represents the total number of frames within the calculation range; It plays a crucial role in security surveillance. By analyzing the target object pixel distribution data, the moving trajectory of the target object in consecutive video frames is calculated. For example, in a supermarket surveillance, the surveillance camera can track the walking path of customers to analyze the customers' attention to goods or detect abnormal behaviors such as staying in a certain area for a long time or moving abnormally fast. Extract the target object in the video and obtain its position coordinates in different frames. For example, the coordinates of a certain customer at frame are , and the coordinates at frame are . To analyze its motion amplitude, calculate the pixel displacement between adjacent frames and perform cumulative calculation; Suppose the supermarket surveillance system detects that the coordinates of a certain customer in four consecutive frames are: Frame 1: , Frame 2: , Frame 3: , Frame 4: ; Substitute the data into the formula for calculation: ; Finally, the calculated target object pixel motion trend value for this customer is 21.89 pixels. Based on this data, further analyze the motion trend of the target object. For example, if the pixel motion amplitude of multiple target objects in a certain area remains within a large numerical range (such as 20 - 30 pixels) for a long time, it means that this area is a highly active area, such as a checkout counter, an entrance passage, etc. And if the motion amplitude of a certain target object suddenly increases within a short period of time (such as from 5 pixels to 30 pixels), it indicates an abnormal situation, such as a customer running or arguing. Based on this analysis, the camera angle can be adjusted or an alarm can be sent to ensure safety.

[0025] The target area task adjustment sub - module, based on the target object pixel motion trend value, compares the pixel change rate of the target object, filters the target areas with high activity, adjusts the video resource priority, and obtains the video resource volume of the target area; By comparing the pixel change rates in different time periods, it can be determined which areas have a high activity of target objects. This is particularly important in monitoring in places such as airports or train stations. By identifying highly active areas, the monitoring can adjust resources, optimize the focus and recording quality of cameras, ensure that key areas receive sufficient attention and recording. This adjustment relies on real - time data analysis and priority setting, automatically adjusting the storage resources and playback quality according to the activity results to ensure that high - quality videos can be quickly reviewed when needed. This process finally obtains the video resource volume of the target area.

[0026] Please refer to Figure 4 , the monitoring area adjustment module includes: The video resource statistics sub - module calls the video resource volume of the target area, identifies the number of video frames, storage occupancy, and effective monitoring duration, filters the areas below the set threshold, and obtains the statistical value of the video resource volume of the area; Monitor the statistics of the number of video frames, the storage occupancy of video data, and the effective monitoring duration for each area. For example, in the monitoring of a shopping mall, video is taken in areas such as the entrance, exit, and inside the mall. By counting the number of video frames per hour in an area, the working efficiency of the monitoring device can be evaluated. The calculation of the storage occupancy helps determine the required storage space size, and the statistics of the effective monitoring duration are used to evaluate the monitoring effect and adjust the monitoring strategy. The statistical results directly affect the resource allocation and optimization strategy of the monitoring. During the execution process, the setting of the threshold needs to be obtained based on the analysis of past data. For example, summarize the number of video frames and storage occupancy in each time period in the past week and calculate their average value as the threshold. Areas below this threshold will be marked as inefficient areas and the monitoring device needs to be adjusted or upgraded. Such a process is completed through specific data monitoring and calculation to obtain the statistical value of the regional video resource volume.

[0027] The behavior density recognition sub-module calls the statistical value of the regional video resource volume, counts the number of occurrences of the target behavior, and identifies the behavior frequency per unit time, the coverage rate per unit area of the region, and the behavior duration. The formula is used: ; Calculate the target behavior density, filter out the areas where the behavior density is lower than the threshold, and obtain the distribution of the target behavior density; Among them, represents the target behavior density, represents the target behavior frequency per unit time, represents the coverage rate per unit area of the target behavior in the monitoring area, represents the average duration of the behavior in the area, represents the statistical value of the regional video resource volume, represents the average value of the video resource volume of the entire region; By counting the number of occurrences of the target behavior in the monitoring screen, calculate the behavior frequency per unit time, the behavior coverage rate per unit area of the region, and the behavior duration. For example, in the monitoring of a large shopping mall, the camera at the entrance records a higher flow of people during peak hours (such as from 5 pm to 8 pm), while the flow of people is lower during the early morning hours. Suppose the total number of times the behavior of customers entering is recorded in a certain area per unit time is F = 300 times, the area of this region is S = 100 square meters, the average time customers stay in this region is L = 30 seconds, and the statistical value of the video resource volume of this region is E = 5000 MB, and the average value of the video resource volume of the entire region A = 5200 MB; The denominator part for calculation is: ; Then calculate the numerator part: ; Finally, calculate the target behavior density: ; This value indicates a relatively high behavior density in this area during this time period, suggesting that this area is a relatively active area within the mall. If the D value of a certain area is below 50, it can be determined that this area is a low-density area and the monitoring strategy needs to be adjusted. Areas with behavior density below the set threshold are screened out, and finally the target behavior density distribution is obtained.

[0028] The monitoring video data upload sub-module calls the target behavior density distribution, judges the change trend of the target behavior density, adjusts the video upload rate of the low-density area, reduces the data transmission volume, and obtains the regional monitoring video stream; Adjust the data upload rate for areas with low target behavior density. For example, in the monitoring of a warehouse area with low passenger flow, high-frequency video data upload is not required. Adjusting the data upload rate can reduce the burden of data storage and processing. During the execution process, by analyzing the target behavior density distribution, judge the change trend of the target behavior density in each area, adjust the video data upload rate for areas with low target behavior density, reduce the data transmission volume in the low-density area, and the adjustment strategy is based on the specific value of the density. For example, if the behavior density of a certain area is lower than 30% of the average value of the whole area, then reduce its video data upload rate by 50%. Such adjustments are implemented through specific numerical analysis and threshold judgment, and finally the regional monitoring video stream is obtained.

[0029] Please refer to Figure 5 , the environmental task optimization module includes: The bandwidth status monitoring sub-module extracts network traffic bandwidth information based on the regional monitoring video stream, monitors the change trend of bandwidth occupancy, analyzes the short-term fluctuation range, judges the current bandwidth status, and obtains the bandwidth occupancy fluctuation amplitude; This is done by observing the Internet usage of each company in a commercial building and analyzing the data to determine when the data traffic reaches its peak. This monitoring helps network administrators adjust network resource allocation according to peak bandwidth usage to ensure the continuity of network services. During the detailed execution process, the bandwidth data collected every minute is recorded and compared with the data of the previous minute. In this way, administrators can observe in real time whether the bandwidth occupancy status exceeds the normal fluctuation range. If an abnormality is found, immediate measures are taken, such as increasing the bandwidth or optimizing the traffic. Through example simulation, assume that the bandwidth occupancy suddenly increases from 40% to 80% within one minute, and this is regarded as an abnormal fluctuation. At this time, further technical intervention is needed to adjust the bandwidth allocation. Through this continuous monitoring and immediate response, it helps to maintain the smooth operation of network traffic and obtain the bandwidth occupancy fluctuation amplitude.

[0030] The recovery stability judgment sub-module extracts the recovery rates of multiple time periods during the bandwidth recovery process based on the bandwidth occupancy fluctuation amplitude, using the formula: ; Calculate the recovery rate of the time period, determine whether the recovery rate remains stable, and obtain the bandwidth recovery stability index; Among them, represents the recovery rate of the time period, represents the time point bandwidth occupancy value of, represents the time point bandwidth occupancy value of, represents the time point bandwidth recovery duration of, represents the time point bandwidth recovery duration of; Calculate the recovery rate in combination with the time interval. This calculation method is widely applicable to scenarios such as data center management, enterprise network optimization, and video streaming server load adjustment. For example, when the bandwidth of a server in a certain area drops due to high load, the administrator needs to evaluate the recovery ability of the server in order to optimize resource allocation. The key to the calculation is to obtain the bandwidth occupancy values at different time points and calculate the rate of change of the bandwidth change. In the case of a short network fluctuation of the server, its recovery speed can be analyzed by recording its bandwidth change process. Assume that at = 10s, the bandwidth occupancy value is = 30Mbps, and at = 20s, the bandwidth occupancy value recovers to = 80Mbps; Substitute specific values for calculation: ; This result shows that within a 10-second time interval, the recovery rate of the time period is 15.8. This value can be used to evaluate the network's recovery ability. In different scenarios, the change in the recovery rate can help network administrators optimize the bandwidth management strategy. For example, if the recovery rate is lower than a certain benchmark value, it means that the network recovery is slow and further resource allocation optimization or bandwidth expansion is required. If the recovery rate is too high, it means that the bandwidth resource configuration is unbalanced and the task traffic needs to be reallocated. By calculating the recovery rates of different time periods, determine whether the recovery rate remains stable, and obtain the bandwidth recovery stability index.

[0031] The task priority adjustment sub-module filters the active area video tasks based on the bandwidth recovery stability index, adjusts the task priorities according to the recovery stability and task urgency, and obtains the video task execution order; Analyze the network activity and bandwidth requirements of each video task, and use the data to optimize the task execution order. An actual example is on a large online education platform where the administrator needs to prioritize ensuring the smooth playback of teaching videos and adjust the transmission priorities of different video sources according to the current bandwidth recovery stability index. During the refinement of the execution process, evaluate the urgency and bandwidth requirements of each task, and then sort the tasks according to priority. The task with the highest priority will obtain the required bandwidth resources. Through instance simulation, if the live video of a class requires stable bandwidth while the recorded video of another class can be processed later, the administrator will perform dynamic bandwidth allocation according to the requirements to ensure the smooth execution of critical tasks and obtain the video task execution sorting.

[0032] Please refer to Figure 6 , the monitoring resource allocation module includes: The computing load detection sub-module calls the video task execution sorting, detects the CPU occupancy rate, memory usage rate, and the number of task execution threads, identifies the task occupancy ratio, judges the load balancing situation, and obtains the server computing load ratio; Obtain the current CPU occupancy rate and memory usage rate of the server through a real-time monitoring tool. For example, in a certain data center, the server records the CPU and memory usage of each task when processing multiple parallel tasks. Based on the obtained data, calculate the resource occupancy ratio of each task through a certain mathematical model. This model takes into account the ratio of the execution time of the task to the amount of resources occupied to determine which tasks are less efficient in resource usage. For example, if a task A has a CPU occupancy rate of 70% and a memory usage rate of 30%, and its execution time is much longer than that of other tasks, it is judged to be inefficient. Through this method, the resource allocation can be dynamically adjusted to ensure more efficient resource utilization, and finally obtain the server computing load ratio. This ratio helps the administrator judge the overall load situation for making corresponding adjustment strategies.

[0033] The available resource identification sub-module extracts the available computing resources based on the server computing load ratio, screens the allocable computing resources, identifies the task load occupancy threshold, and uses the formula: ; Compare the available resources with the task requirements to obtain the available computing resource allocation amount; Among them, represents the available computing resource allocation amount, represents the total server computing resources, represents the computing resource occupancy of the jth task, represents the execution time of the jth task, represents the average value of the task execution time, represents the total number of tasks currently being executed; Estimate the remaining available computing resources to ensure the rationality of task scheduling. The total computing resources of the server are set to 1000 units, which can be determined by monitoring the hardware specifications of the server, such as the number of CPU cores, the amount of available memory, etc. Analyze the currently running tasks. Assume that three tasks are being executed simultaneously, and the computing resource consumption of each task is Task 1: 200 units, Task 2: 300 units, Task 3: 400 units. It is necessary to obtain the execution time of each task. Assume that the execution time of Task 1 is 1 hour, Task 2 is 2 hours, and Task 3 is 1.5 hours. Based on the data, calculate the mean value of the execution times of all tasks. The calculation method is as follows: ; Next, calculate the sum of the absolute deviations of the execution times of each task from the mean value: ; Substitute the above values into the formula for calculating the remaining computing resources: ; From this, it can be obtained that the currently remaining available computing resources are 100 units. Under the current server load status, there are still 100 units of computing resources that can be used for the scheduling of new tasks or the expansion of existing tasks. This value can be used as a reference basis for the subsequent optimization of video task scheduling to determine whether additional task loads can be added or whether the task execution order needs to be adjusted to make full use of server resources and improve the overall operation efficiency.

[0034] The video task scheduling optimization sub-module determines task priorities based on the available computing resource allocation, filters out low-priority tasks for deferral and merging, optimizes the resource allocation structure, and obtains the resource scheduling volume for video behavior monitoring tasks; Evaluate the priorities of each task, perform priority sorting according to the urgency and resource requirements of the tasks. For example, critical video analysis tasks are given higher priorities due to their high requirements for real-time performance. According to the known resource allocation volume and the priorities of each task, use an algorithm to determine which low-priority tasks can be deferred or merged with other tasks for execution, thereby optimizing the overall utilization efficiency of resources. Through such resource adjustments, finally obtain the resource scheduling volume for video behavior monitoring tasks, indicating how to reasonably allocate the remaining resources to support the smooth progress of tasks while ensuring the priority execution of critical tasks.

[0035] The above are only the preferred embodiments of the present invention and do not limit the present invention in other forms. Any person skilled in the relevant art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A cloud-based video behavior monitoring system, characterized in that: The system comprises: The video bandwidth control module captures bandwidth data based on the bandwidth status of the video transmission channel, analyzes the time interval and change amplitude of bandwidth changes, adjusts the video frame compression ratio and data volume, monitors the adjusted video transmission status, and obtains the bandwidth matching video data volume; The behavior calculation scheduling module matches the video data volume based on the bandwidth, detects the target object in the video surveillance screen, analyzes the pixel motion amplitude of the target object, determines the activity of the target object, adjusts the task priority of the target area, and obtains the video resource volume of the target area; The monitoring area adjustment module calls the target area video resource amount, identifies the target behavior density of each area, determines the change trend of the target behavior density, adjusts the video data upload rate of the low-density area, and obtains the regional monitoring video stream; The environmental task optimization module monitors the video stream according to the area, detects the network traffic bandwidth status, identifies the bandwidth recovery speed, determines the stability of bandwidth recovery, adjusts the task execution order, and obtains the video task execution ranking.

2. The cloud-based video behavior monitoring system according to claim 1, characterized in that: The bandwidth matching video data volume includes compression ratio parameters, data stream transmission rate, and video quality assessment indicators; the target area video resource volume includes activity weight, priority coefficient, and pixel change amplitude; the area monitoring video stream includes behavior density parameters, upload rate threshold, and monitoring coverage; the video task execution sorting includes recovery speed index, bandwidth stability coefficient, and task priority sequence.

3. The cloud-based video behavior monitoring system according to claim 1, characterized in that: The video bandwidth control module comprises: The bandwidth status capture submodule captures bandwidth data in differentiated time periods based on the bandwidth status of the video transmission channel, detects the bandwidth utilization and data transmission rate of each time period, analyzes the bandwidth change rate of adjacent time periods, and records the bandwidth fluctuation amplitude to obtain a bandwidth fluctuation data set; The bandwidth trend analysis submodule identifies the bandwidth change intervals between adjacent time periods based on the bandwidth fluctuation data set, analyzes the bandwidth fluctuation frequency, compares the bandwidth stability of differentiated time intervals, and obtains bandwidth change trend data; The video data adjustment submodule calls the bandwidth change trend data, analyzes the video compression ratio matching under the differentiated bandwidth state, adjusts the video frame compression ratio and the amount of transmitted data, monitors the adjusted video transmission state, and uses the formula: ； Compare the video quality corresponding to the differentiated encoding parameters, select the optimal encoding parameter set, and obtain the bandwidth matching video data volume; in, Represents the amount of video data that matches the bandwidth. represents the bandwidth utilization rate in the i-th time period, represents the video compression ratio of the i-th time period, represents the video transmission time of the i-th time period, represents the amount of video data in the i-th time period, Represents the total number of time periods.

4. The cloud-based video behavior monitoring system according to claim 3, characterized in that: The behavior calculation scheduling module includes: The target object detection submodule extracts the video surveillance screen data based on the bandwidth matching video data volume, detects multiple target objects in the surveillance screen, distinguishes background pixels from target object pixels, extracts pixel point data of the target object, and obtains pixel distribution data of the target object; The pixel motion amplitude analysis submodule identifies the pixel offset of the target object in the continuous frame images based on the pixel distribution data of the target object, counts the displacement information, and analyzes the pixel change amplitude using the formula: ； Get the pixel motion trend value of the target object; in, Represents the pixel motion trend value of the target object, Represents the target object in The position coordinates of the frame, Represents the target object in The position coordinates of the frame, Represents the total number of frames within the calculation range; The target area task adjustment submodule compares the pixel change rate of the target object based on the pixel motion trend value of the target object, screens the target area of ​​activity, adjusts the video resource priority, and obtains the video resource amount of the target area.

5. The cloud-based video behavior monitoring system according to claim 4, characterized in that: The monitoring area adjustment module includes: The video resource statistics submodule calls the target area video resource amount, identifies the number of video frames, storage occupancy and effective monitoring time, selects the area below the set threshold, and obtains the regional video resource amount statistics; The behavior density identification submodule calls the statistical value of the video resource volume in the region, counts the number of occurrences of the target behavior, identifies the frequency of the behavior per unit time, the coverage rate per unit area of ​​the region, and the duration of the behavior, using the formula: ； Calculate the target behavior density, filter out areas where the behavior density is lower than the threshold, and obtain the target behavior density distribution; in, represents the target behavior density, Represents the frequency of target behavior per unit time, Represents the unit area coverage of the target behavior in the monitoring area, represents the average duration of the behavior in the area, Represents the statistical value of regional video resources. Represents the average amount of video resources in the entire region; The monitoring video data uploading submodule calls the target behavior density distribution, determines the target behavior density change trend, adjusts the video upload rate of the low-density area, reduces the data transmission volume, and obtains the regional monitoring video stream.

6. The cloud-based video behavior monitoring system according to claim 5, characterized in that: The environmental task optimization module includes: The bandwidth status monitoring submodule monitors the video stream according to the area, extracts the network traffic bandwidth information, monitors the bandwidth occupancy change trend, analyzes the short-term fluctuation range, determines the current bandwidth status, and obtains the bandwidth occupancy fluctuation range; The recovery stability judgment submodule extracts the recovery rate of multiple time periods during the bandwidth recovery process based on the bandwidth occupancy fluctuation amplitude, using the formula: ； Calculate the recovery rate of the time period, determine whether the recovery rate remains stable, and obtain the bandwidth recovery stability index; in, represents the recovery rate of the time period, Representing time point The bandwidth usage value of Representing time point The bandwidth usage value of Representing time point The bandwidth recovery time is Representing time point The bandwidth recovery time; The task priority adjustment submodule screens the video tasks in the active area based on the bandwidth recovery stability index, adjusts the task priority according to the recovery stability and the task urgency, and obtains the video task execution ranking.

7. The cloud-based video behavior monitoring system according to claim 1, characterized in that: The system also includes a monitoring resource allocation module: The monitoring resource allocation module calls the video task execution sorting, detects the server computing load, identifies the available amount of resources, determines the upper limit of resource load, screens low-priority tasks, optimizes resource allocation, and obtains the resource scheduling amount of the video behavior monitoring task; The resource scheduling amount of the video behavior monitoring task includes calculation load threshold, available resource quota, and load upper limit parameters.

8. The cloud-based video behavior monitoring system according to claim 7, characterized in that: The monitoring resource allocation module includes: The computing load detection submodule calls the video task execution sorting, detects the CPU occupancy rate, memory usage rate and the number of task execution threads, identifies the task occupancy ratio, determines the load balancing situation, and obtains the server computing load ratio; The resource availability identification submodule extracts the available computing resources based on the server computing load ratio, screens the allocatable computing resources, and identifies the task load occupancy threshold using the formula: ； Compare the available resources with the task requirements to obtain the available computing resource allocation; in, Represents the amount of available computing resources allocated, Represents the total computing resources of the server, represents the computing resource usage of the jth task, represents the execution time of the jth task, represents the mean task execution time, Represents the total number of tasks currently being executed; The video task scheduling optimization submodule determines the task priority based on the available computing resource allocation, screens low-priority tasks for postponement and merging, optimizes the resource allocation structure, and obtains the resource scheduling amount of the video behavior monitoring task.

Images