Digital video monitoring processing system based on artificial intelligence

By integrating artificial intelligence technology into the video surveillance system, using deep learning algorithms and geographic information systems for video analysis and abnormal behavior correlation, the problem of insufficient accuracy and intelligence in the existing system is solved, and efficient monitoring and resources are achieved Optimization improves the stability and user experience of the system.

CN118984428AInactive Publication Date: 2024-11-19ANHUI TALIAN INTELLIGENT TECH CO LTD
View PDF 0 Cites 6 Cited by

Patent Information

Application Number
CN202410981090.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-22
Publication Date
2024-11-19
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing video surveillance systems have shortcomings in accuracy, intelligence level and environmental adaptability, and the integration of artificial intelligence technology into the system faces challenges such as complex technology implementation, high computing resource requirements, and difficult system integration.

Method used

A digital video surveillance and processing system based on artificial intelligence was designed, including video acquisition module, video preprocessing module, intelligent analysis module and association module. Deep learning algorithms are used for real-time analysis and abnormal behavior recognition, and combined with multiple geographic information systems Data from surveillance cameras evaluate and correlate abnormal behaviors, providing complete context and event correlation analysis.

Benefits of technology

It significantly improves the intelligence level and response speed of the video surveillance system, reduces dependence on manual monitoring, improves monitoring efficiency and accuracy, optimizes resource configuration, enhances the stability and user experience of the system, and through redundancy and The fault tolerance mechanism solves the difficulty of integrating computing resources and systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118984428A_ABST
    Figure CN118984428A_ABST
Patent Text Reader

Abstract

The invention, which relates to the technical field of monitoring video processing, discloses an artificial intelligence-based digital video monitoring processing system comprising a video acquisition module, a video preprocessing module, an intelligent analysis module, an anomaly detection module and an association module. The video acquisition module is used for acquiring video data from a plurality of monitoring cameras in real time; according to the invention, through integration of an advanced artificial intelligence technology, the intelligent level of the video monitoring system is obviously improved, and compared with a traditional monitoring system, the intelligent level of the video monitoring system is greatly improved. According to the system, target detection, behavior recognition and abnormal behavior analysis can be automatically carried out, dependence on manual monitoring is reduced, the adaptive capacity of the system to complex scenes and the response speed of the system to abnormal events are greatly improved through application of the deep learning algorithm, and therefore monitoring efficiency and accuracy are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of monitoring video processing, and in particular to a digital video monitoring processing system based on artificial intelligence. Background Art

[0002] The rapid development of artificial intelligence technology, especially the breakthrough of deep learning in the field of image recognition and video analysis, has provided new possibilities for the intelligent upgrade of video surveillance systems. AI technology can automatically identify and classify targets and behaviors in videos through training and learning a large amount of data, significantly improving the accuracy and response speed of surveillance systems. However, the effective integration of artificial intelligence technology into video surveillance systems still faces challenges such as complex technical implementation, high computing resource requirements, and difficulty in system integration.

[0003] The rapid development of artificial intelligence technology, especially the breakthrough of deep learning in the field of image recognition and video analysis, has provided new possibilities for the intelligent upgrade of video surveillance systems. AI technology can automatically identify and classify targets and behaviors in videos through training and learning large amounts of data, significantly improving the accuracy and response speed of monitoring systems. However, the effective integration of artificial intelligence technology into video surveillance systems still faces challenges such as complex technical implementation, high computing resource requirements, and difficulty in system integration.

[0004] In order to solve the above defects, a technical solution is now provided. Summary of the invention

[0005] The purpose of the present invention is to solve the deficiencies of the existing video surveillance system in terms of accuracy, intelligence level and environmental adaptability, and to propose a digital video surveillance processing system based on artificial intelligence.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] The digital video surveillance processing system based on artificial intelligence includes:

[0008] Video acquisition module, used to obtain video data from multiple surveillance cameras in real time;

[0009] A video preprocessing module, used for preprocessing the collected video data;

[0010] Intelligent analysis module, which is used to analyze video content in real time using deep learning algorithms;

[0011] Anomaly detection module, used to set detection thresholds according to application scenarios, including motion detection and abnormal behavior recognition;

[0012] The association module is used to determine the surveillance video data associated with the abnormal behavior data based on the detected abnormal behavior video data combined with the geographical location and coverage of all surveillance cameras. The specific steps are as follows:

[0013] Receive abnormal behavior data output by the anomaly detection module, including timestamp, surveillance camera ID and anomaly type information, and integrate with the geographic information system to store and query the geographic location information of the surveillance camera;

[0014] Define the coverage of each surveillance camera through GIS data or the camera's field of view. According to the geographical location where the abnormal behavior occurs, determine which surveillance cameras' coverage is less than the preset association distance from the corresponding surveillance camera, and mark such surveillance cameras as associated surveillance cameras. Again, synchronize the timestamps of all surveillance cameras for accurate time association analysis.

[0015] Retrieve video data of a preset time period before and after the time when the abnormal behavior occurs from the associated surveillance cameras, and fuse the retrieved multi-channel video data; integrate the abnormal behavior data and the data of the associated surveillance cameras to provide a complete context of the abnormal event and track the movement trajectory of the abnormal behavior within the coverage area of ​​different cameras;

[0016] Then, the correlation of the video data of each associated surveillance camera within the preset association distance range is evaluated to evaluate the degree of correlation between the associated surveillance camera and the abnormal behavior.

[0017] Furthermore, the specific steps of evaluating the degree of association between the associated surveillance camera and the abnormal behavior in the association module are as follows:

[0018] Comprehensive analysis is performed by collecting evaluation parameters in the video data, where the evaluation parameters include:

[0019] Time synchronization: The difference between the timestamp of the abnormal behavior and the timestamp of the associated camera video data, which is quantified by the time difference value;

[0020] Spatial distance: The geographical distance between the location where the abnormal behavior occurs and the associated camera, quantified by straight-line distance;

[0021] View coverage: The degree of overlap between the camera’s field of view and the area where the abnormal behavior occurs. The overlap is quantified by the overlap percentage. If there is no overlap, it is recorded as 0.

[0022] Target appearance frequency: The number of times the subject of abnormal behavior, including people or vehicles, appears in the video data of the associated camera, and the number of appearances is used as the standard for quantifying the target appearance frequency;

[0023] Stay time: The time that the abnormal behavior subject stays in the camera's field of view. This stay time is recorded as the stay time value;

[0024] The obtained time difference, straight-line distance, overlap percentage, number of occurrences and retention time are calibrated as cs, zj, cb, cc and ls respectively, and after normalization, they are entered into the following formula: To obtain the key evaluation value GPZ, where ρ1, ρ2, ρ3, ρ4, and ρ5 are the preset weight coefficients of time difference, straight-line distance, overlap percentage, number of occurrences, and retention time, respectively; and the calculated key evaluation value is used as a standard to measure the degree of association between the associated surveillance camera and abnormal behavior.

[0025] Furthermore, the association module is also used for:

[0026] The evaluation values ​​obtained by analyzing all related surveillance cameras are sorted by size, and the related videos are displayed to the user through the user interface in order of the evaluation values; at the same time, the results of the related analysis are sent to alarms and relevant personnel are notified;

[0027] Record abnormal behaviors and related surveillance video data for post-analysis and auditing;

[0028] Provide user configuration options to allow users to adjust specific parameters of association analysis according to specific needs, design redundancy and fault-tolerant mechanisms to ensure that the association module remains operational when some cameras or system components fail, and continuously optimize the algorithm and logic of the association module based on user feedback and performance.

[0029] Furthermore, the specific operation steps of the video acquisition module to obtain video data from multiple surveillance cameras in real time are as follows:

[0030] Determine the signal used by the surveillance camera, including analog or digital signals, and then determine the technical parameters of the surveillance camera, including resolution and frame rate;

[0031] Select the corresponding video capture card or video server according to the number and quality of videos to be captured, install compatible drivers and software on the selected video capture card and video server, and configure the software to identify and control the connected surveillance cameras;

[0032] Then, all surveillance cameras are synchronized and adjusted with the timestamp through the network time protocol service. The synchronization adjustment of the timestamp is carried out by calculating the timestamp synchronization error.

[0033] For IP cameras, add the IP address of the camera through the software, establish a connection and start receiving the video stream. For analog cameras, ensure that the video signal is correctly captured by the capture card;

[0034] Configure the storage path and storage strategy for video data, including loop recording and event-triggered recording, and configure user permissions to ensure that only authorized users can access video data.

[0035] Furthermore, the video preprocessing module preprocesses the collected video data in the following specific operation steps:

[0036] Receive the raw video data transmitted from the video acquisition module and adjust the frame rate and resolution of the video according to processing requirements and system performance;

[0037] Apply denoising algorithms to reduce noise in the video, adjust the contrast and brightness of the video, use edge detection algorithms to enhance edge information in the video, and apply histogram equalization technology to improve the contrast of the video;

[0038] A background subtraction algorithm is used to separate the foreground object and the background, and the format of the video data is converted for subsequent processing and storage, and then the video data is encoded and compressed.

[0039] Furthermore, the specific operation steps of the intelligent analysis module using the deep learning algorithm to perform real-time analysis on the video content are as follows:

[0040] Set the target type and behavior to be identified. The target type includes people or cars, and the behavior includes movement or intrusion. Select the corresponding deep learning framework, including TensorFlow or PyTorch, to develop and train the model.

[0041] Collect existing video data for annotation, including target detection boxes and behavior labels, select the corresponding deep learning model architecture, including convolutional neural network or recurrent neural network, use the annotated data to train the model, and optimize the model parameters;

[0042] Optimize the trained model and deploy it to the video analysis server or edge device, and connect the video data stream output by the video preprocessing module to the intelligent analysis module;

[0043] The model monitors targets in the video, including people and cars, in real time and outputs the location and attributes of the targets.

[0044] Furthermore, the abnormality detection module sets the detection threshold according to the application scenario, and the specific operation steps for motion detection and abnormal behavior identification are as follows:

[0045] Determine the specific needs of the monitoring scenario, including the types of anomalies that need to be detected, including motion, intrusion, and artifacts; configure detection parameters according to the application scenario, including sensitivity and time window;

[0046] Collect video data in normal and abnormal scenarios for algorithm training and testing. Use the collected data to train models to distinguish between normal and abnormal behaviors and set detection thresholds to determine what degree of change or behavior constitutes an abnormality.

[0047] Through the motion detection algorithm based on background subtraction and optical flow technology, the abnormal behavior recognition algorithm is integrated, including the algorithm based on behavior pattern analysis, and the real-time video stream is connected to the anomaly detection module for real-time analysis;

[0048] The preset logic judgment code determines whether an abnormal event occurs based on the model output and threshold. When an abnormal event is detected, an early warning is generated and relevant personnel are notified.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] (1) The present invention significantly improves the intelligence level of the video surveillance system by integrating advanced artificial intelligence technology. Compared with traditional surveillance systems, the present system can automatically perform target detection, behavior recognition and abnormal behavior analysis, reducing the reliance on manual monitoring. The application of deep learning algorithms greatly enhances the system's adaptability to complex scenarios and its response speed to abnormal events, thereby improving monitoring efficiency and accuracy.

[0051] (2) The present invention can more accurately identify and respond to abnormal behaviors in surveillance videos through video acquisition, preprocessing, intelligent analysis and anomaly detection modules, which not only optimizes the configuration of surveillance resources and avoids ineffective monitoring and resource waste, but also improves the security of the surveillance area through fast and accurate anomaly detection and alarm mechanisms. The introduction of the association module enables the system to perform event association analysis across cameras, providing users with a more comprehensive security perspective.

[0052] (3) The present invention takes into account various challenges in practical applications, such as computing resource limitations and system integration difficulties, and adopts redundancy and fault-tolerant mechanisms to ensure the stable operation and high reliability of the system. The provision of user configuration options enables users to adjust system parameters according to specific needs, thereby enhancing the flexibility and applicability of the system. At the same time, the optimized design of the user interface enables users to intuitively obtain and process monitoring information, greatly improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;

[0054] Figure 1 This is the overall system block diagram of the present invention. DETAILED DESCRIPTION

[0055] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] It should be understood that the terms "include" and "comprising" used in the specification and claims of the present disclosure indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0057] It should also be understood that the terms used in this disclosure are only for the purpose of describing specific embodiments and are not intended to limit the disclosure. As used in this disclosure and claims, the singular forms of "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise. It should also be further understood that the term "and / or" used in this disclosure and claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations.

[0058] like Figure 1 As shown, the digital video monitoring processing system based on artificial intelligence includes a video acquisition module, a video preprocessing module, an intelligent analysis module, an anomaly detection module and a correlation module;

[0059] The video acquisition module is used to obtain video data from multiple surveillance cameras in real time;

[0060] Determine the signal used by the surveillance camera, including analog or digital signals, and then determine the technical parameters of the surveillance camera, including resolution and frame rate; select the corresponding video capture card or video server according to the number and quality of video channels to be collected, and install compatible drivers and software on the selected video capture card and video server, and configure the software to identify and control the connected surveillance camera;

[0061] Then, all surveillance cameras are synchronized and adjusted through the network time protocol service. The synchronization adjustment of the timestamp is carried out by calculating the timestamp synchronization error through the following formula: Synchronization error = |t 异常 -t 采集 |, where t 异常 is the timestamp of the abnormal behavior, t 采集 is the timestamp of the video data;

[0062] For IP cameras, add the camera's IP address through the software, establish a connection and start receiving video streams. For analog cameras, ensure that the video signal is correctly captured by the acquisition card. Configure the storage path and storage strategy for video data, including loop recording and event-triggered recording, and configure user permissions to ensure that only authorized users can access video data.

[0063] The video preprocessing module is used to preprocess the collected video data, including denoising, enhancement and format conversion;

[0064] Receive the raw video data transmitted from the video acquisition module, adjust the frame rate and resolution of the video according to the processing requirements and system performance; apply denoising algorithms (such as Gaussian blur, median filtering, etc.) to reduce the noise in the video, adjust the contrast and brightness of the video to enhance the visibility of the image, use edge detection algorithms (such as Sobel operator, Canny operator, etc.) to enhance the edge information in the video, and apply histogram equalization technology to improve the contrast of the video, especially in the case of uneven lighting, specifically through the following formula: I eq (x, y) = λ × I (x, y) + β Specifically, I (x, y) is the intensity value of the original image, I eq (x, y) is the image intensity value after equalization, λ and β are parameters determined according to the cumulative distribution function of the image; the background subtraction algorithm is used to separate the foreground target and the background, and the format of the video data is converted for subsequent processing and storage, and then the video data is encoded and compressed to reduce the storage space and transmission bandwidth requirements.

[0065] The intelligent analysis module is used to perform real-time analysis of video content using deep learning algorithms, including but not limited to target detection, behavior recognition, and abnormal behavior analysis;

[0066] Set the target type and behavior to be identified, including people or cars, and behaviors including movement or intrusion. Select the corresponding deep learning framework, including TensorFlow or PyTorch, to develop and train the model. Collect existing video data for annotation, including target detection boxes and behavior labels, select the corresponding deep learning model architecture, including convolutional neural network (CNN) or recurrent neural network (RNN), use the annotated data to train the model, and optimize the model parameters.

[0067] Optimize the trained model to improve its performance and accuracy in real-time video analysis, deploy the trained model to the video analysis server or edge device, and connect the video data stream output by the video preprocessing module to the intelligent analysis module; use the model to monitor targets in the video in real time, including people and vehicles, and output the location and attributes of the target.

[0068] The anomaly detection module is used to set the detection threshold according to the application scenario, including motion detection and abnormal behavior recognition;

[0069] Determine the specific needs of the monitoring scene, including the types of anomalies that need to be detected, such as motion, intrusion, and relics; configure detection parameters such as sensitivity and time window according to the application scenario; collect video data in normal and abnormal scenes for algorithm training and testing, use the collected data to train the model so that the model can distinguish between normal and abnormal behaviors, and set the detection threshold to determine what degree of change or behavior constitutes an abnormality. The detection threshold is obtained by the formula: motion threshold = average optical flow + k × standard deviation, where the average optical flow is the average value of the optical flow vectors of all pixels in a certain image area or the entire image, which is used to represent the overall motion intensity in the scene, and the standard deviation represents the variability of the size of the optical flow vector, that is, the inconsistency of the motion speed of different pixels in the scene; k is an adjustable constant used to determine the sensitivity of the threshold, and the value of k is set between 0.5-1.72;

[0070] Through the motion detection algorithm based on background subtraction and optical flow technology, the abnormal behavior recognition algorithm is integrated, including the algorithm based on behavior pattern analysis, and the real-time video stream is connected to the anomaly detection module for real-time analysis; the preset logic judgment code determines whether an abnormal event occurs based on the model output and threshold. When an abnormal event is monitored, an early warning is generated and a warning notification is issued to relevant personnel.

[0071] The association module is used to determine the surveillance video data associated with the abnormal behavior data based on the detected abnormal behavior video data combined with the geographical location and coverage of all surveillance cameras;

[0072] Receive abnormal behavior data output by the anomaly detection module, including timestamp, surveillance camera ID and anomaly type information, and integrate with the geographic information system to store and query the geographic location information of the surveillance camera;

[0073] Define the coverage of each surveillance camera through GIS data or the camera's field of view (FOV). According to the geographical location where the abnormal behavior occurs, determine which cameras' coverage is less than the preset association distance from the corresponding surveillance camera, and mark such surveillance cameras as associated surveillance cameras. Synchronize the timestamps of all surveillance cameras again for accurate time association analysis. Retrieve video data of the preset time period before and after the time point when the abnormal behavior occurs from the associated surveillance cameras, and fuse the retrieved multi-channel video data to provide a more comprehensive perspective and information.

[0074] Integrate abnormal behavior data and data from associated surveillance cameras to provide a complete context for abnormal events and track the movement of abnormal behaviors within the coverage of different cameras; then evaluate the relevance of the video data of each associated surveillance camera within the preset association distance range to evaluate the degree of association between the associated surveillance camera and the abnormal behavior. Specifically, a comprehensive analysis is performed by collecting evaluation parameters in the video data, where the evaluation parameters include:

[0075] Time synchronization: the difference between the timestamp of the abnormal behavior and the timestamp of the associated camera video data, and the difference is quantified by the time difference value; spatial distance: the geographical distance between the location where the abnormal behavior occurs and the associated camera, and quantified by the straight-line distance; visual coverage: the overlap between the camera's field of view (FOV) and the area where the abnormal behavior occurs, and the overlap is quantified by the overlap percentage. When there is no overlap, it is recorded as 0; target occurrence frequency: the number of times the abnormal behavior subject, including people or vehicles, appears in the video data of the associated camera, and the number of occurrences is used as the standard for quantifying the target occurrence frequency; residence time: the residence time of the abnormal behavior subject in the camera's field of view, and this residence time is recorded as the residence time value;

[0076] The obtained time difference, straight-line distance, overlap percentage, number of occurrences and retention time are calibrated as cs, zj, cb, cc and ls respectively, and after normalization, they are entered into the following formula: To obtain the key evaluation value GPZ, where ρ1, ρ2, ρ3, ρ4, and ρ5 are the preset weight coefficients of the time difference value, the straight-line distance, the overlap percentage, the number of occurrences, and the retention time value, respectively, and the values ​​are set between 1.12-1.24, 0.92-0.95, 1.11-1.19, 1.23-1.29, and 0.89-0.93; and the calculated key evaluation value is used as a standard to measure the degree of association between the associated surveillance camera and the abnormal behavior;

[0077] The evaluation values ​​obtained from the analysis of all related surveillance cameras are sorted by size, and when the related videos are displayed to the user through the user interface, they are displayed in sequence according to the evaluation values, so that the user can see the video data related to the abnormal behavior at the first time; at the same time, the results of the association analysis are sent to alarm or notify relevant personnel; the abnormal behavior and related surveillance video data are recorded to provide information for post-analysis and audit; user configuration options are provided to allow users to adjust the specific parameters of the association analysis according to specific needs, design redundancy and fault-tolerant mechanisms to ensure that the association module can still work normally when some cameras or system components fail, and continuously optimize the algorithm and logic of the association module based on user feedback and performance.

[0078] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. Digital video surveillance processing system based on artificial intelligence, It is characterized in that include: Video acquisition module, used to obtain video data from multiple surveillance cameras in real time; A video preprocessing module, used for preprocessing the collected video data; Intelligent analysis module, which is used to analyze video content in real time using deep learning algorithms; Anomaly detection module, used to set detection thresholds according to application scenarios, including motion detection and abnormal behavior recognition; The association module is used to determine the surveillance video data associated with the abnormal behavior data based on the detected abnormal behavior video data combined with the geographical location and coverage of all surveillance cameras. The specific steps are as follows: Receive abnormal behavior data output by the anomaly detection module, including timestamp, surveillance camera ID and anomaly type information, and integrate with the geographic information system to store and query the geographic location information of the surveillance camera; Define the coverage of each surveillance camera through GIS data or the field of view of the camera, and determine which surveillance cameras have a coverage distance less than a preset association distance from the location of the corresponding surveillance camera based on the geographical location where the abnormal behavior occurs, and mark such surveillance cameras as associated surveillance cameras; The timestamps of all surveillance cameras are synchronized again for accurate time correlation analysis; Retrieve video data of a preset time period before and after the time when the abnormal behavior occurred from the associated surveillance cameras, and fuse the retrieved multi-channel video data; integrate the abnormal behavior data and the data of the associated surveillance cameras to provide a complete context of the abnormal event and track the movement trajectory of the abnormal behavior within the coverage area of ​​different cameras; Then, the correlation of the video data of each associated surveillance camera within the preset association distance range is evaluated to evaluate the degree of correlation between the associated surveillance camera and the abnormal behavior.

2. The digital video monitoring processing system based on artificial intelligence according to claim 1, It is characterized in that The specific steps of evaluating the degree of association between the associated surveillance camera and the abnormal behavior in the association module are as follows: Comprehensive analysis is performed by collecting evaluation parameters in the video data, where the evaluation parameters include: Time synchronization: The difference between the timestamp of the abnormal behavior and the timestamp of the associated camera video data, which is quantified by the time difference value; Spatial distance: The geographical distance between the location where the abnormal behavior occurs and the associated camera, quantified by straight-line distance; View coverage: The degree of overlap between the camera’s field of view and the area where the abnormal behavior occurs. The overlap is quantified by the overlap percentage. If there is no overlap, it is recorded as 0. Target appearance frequency: The number of times the subject of abnormal behavior, including people or vehicles, appears in the video data of the associated camera, and the number of appearances is used as the standard for quantifying the target appearance frequency; Stay time: The time the abnormal behavior subject stays in the camera's field of view. This stay time is recorded as the stay time value. The obtained time difference, straight-line distance, overlap percentage, number of occurrences and retention time are calibrated as cs, zj, cb, cc and ls respectively, and after normalization, they are entered into the following formula: To get the key evaluation value GPZ, where ρ 1 , 2 , 3 , 4 , 5 The preset weight coefficients are respectively the time difference value, the straight-line distance, the overlap percentage, the number of occurrences and the retention time value; and the calculated evaluation value is used as a standard to measure the degree of association between the associated surveillance camera and the abnormal behavior.

3. The digital video monitoring processing system based on artificial intelligence according to claim 2, It is characterized in that The association module is also used for: The evaluation values ​​obtained by analyzing all related surveillance cameras are sorted by size, and the related videos are displayed to the user through the user interface in order of the evaluation values; at the same time, the results of the related analysis are sent to alarms and relevant personnel are notified; Record abnormal behaviors and related surveillance video data for post-analysis and auditing; Provide user configuration options to allow users to adjust specific parameters of association analysis according to specific needs, design redundancy and fault-tolerant mechanisms to ensure that the association module remains operational when some cameras or system components fail, and continuously optimize the algorithm and logic of the association module based on user feedback and performance.

4. The digital video monitoring processing system based on artificial intelligence according to claim 1, It is characterized in that The specific operation steps of the video acquisition module to obtain video data from multiple surveillance cameras in real time are as follows: Determine the signal used by the surveillance camera, including analog or digital signals, and then determine the technical parameters of the surveillance camera, including resolution and frame rate; Select the corresponding video capture card or video server according to the number and quality of videos to be captured, install compatible drivers and software on the selected video capture card and video server, and configure the software to identify and control the connected surveillance cameras; Then, all surveillance cameras are synchronized and adjusted with the timestamp through the network time protocol service. The synchronization adjustment of the timestamp is carried out by calculating the timestamp synchronization error. For IP cameras, add the IP address of the camera through the software, establish a connection and start receiving the video stream. For analog cameras, ensure that the video signal is correctly captured by the capture card; Configure the storage path and storage strategy for video data, including loop recording and event-triggered recording, and configure user permissions to ensure that only authorized users can access video data.

5. The digital video monitoring processing system based on artificial intelligence according to claim 1, It is characterized in that The specific operation steps of the video preprocessing module for preprocessing the collected video data are as follows: Receive the raw video data transmitted from the video acquisition module and adjust the frame rate and resolution of the video according to processing requirements and system performance; Apply denoising algorithms to reduce noise in the video, adjust the contrast and brightness of the video, use edge detection algorithms to enhance edge information in the video, and apply histogram equalization technology to improve the contrast of the video; A background subtraction algorithm is used to separate the foreground object and the background, and the format of the video data is converted for subsequent processing and storage, and then the video data is encoded and compressed.

6. The digital video monitoring processing system based on artificial intelligence according to claim 1, It is characterized in that The specific operation steps of the intelligent analysis module using the deep learning algorithm to perform real-time analysis of video content are as follows: Set the target type and behavior to be identified. The target type includes people or cars, and the behavior includes movement or intrusion. Select the corresponding deep learning framework, including TensorFlow or PyTorch, to develop and train the model. Collect existing video data for annotation, including target detection boxes and behavior labels, select the corresponding deep learning model architecture, including convolutional neural network or recurrent neural network, use the annotated data to train the model, and optimize the model parameters; Optimize the trained model and deploy it to the video analysis server or edge device, and connect the video data stream output by the video preprocessing module to the intelligent analysis module; The model monitors targets in the video, including people and cars, in real time and outputs the location and attributes of the targets.

7. The digital video monitoring processing system based on artificial intelligence according to claim 1, It is characterized in that The anomaly detection module sets the detection threshold according to the application scenario, and the specific operation steps for motion detection and abnormal behavior identification are as follows: Determine the specific needs of the monitoring scenario, including the types of anomalies that need to be detected, including motion, intrusion, and artifacts; configure detection parameters according to the application scenario, including sensitivity and time window; Collect video data in normal and abnormal scenarios for algorithm training and testing. Use the collected data to train models to distinguish between normal and abnormal behaviors and set detection thresholds to determine what degree of change or behavior constitutes an abnormality. Through the motion detection algorithm based on background subtraction and optical flow technology, the abnormal behavior recognition algorithm is integrated, including the algorithm based on behavior pattern analysis, and the real-time video stream is connected to the anomaly detection module for real-time analysis; The preset logic judgment code determines whether an abnormal event occurs based on the model output and threshold. When an abnormal event is detected, an early warning is generated and relevant personnel are notified.

Citation Information

Cited By

  • Intelligent security and protection monitoring platform based on computer and system thereof

    CN119360309A

  • Intelligent water treatment monitoring device and system

    CN119804801A

  • Coal mining data management system and method based on video artificial intelligence

    CN119851183A

  • A coal mining data management system and method based on video artificial intelligence

    CN119851183B

  • Parking lot monitoring method and device based on image recognition, equipment and medium

    CN120014546A