CN116665386A - A method for intelligent video surveillance linkage detection and early warning - Google Patents

Abstract

The invention discloses an intelligent video monitoring linkage detection early warning method, relates to the field of video linkage monitoring of hydropower stations, and solves the problems that an existing video monitoring system of a hydropower station cannot intelligently identify risks and track people or objects with risks. The method has the characteristics of being convenient for intelligently monitoring the hydropower station, effectively identifying the risk points of the monitoring picture, tracking and displaying the risk targets, and effectively and timely alarming the risks.

Description

A method for intelligent video surveillance linkage detection and early warning

technical field

The invention relates to the field of video linkage monitoring of hydropower stations, in particular to an intelligent video monitoring linkage detection and early warning method.

Background technique

A hydropower station is composed of a hydraulic system, a mechanical system, and an electric energy generating device. It is a water conservancy project that realizes the conversion of water energy to electric energy. The sustainability of electric energy production requires that the utilization of hydropower in hydropower stations be uninterrupted. Through the construction of hydropower reservoir system, artificially adjust and change the distribution of hydropower resources in time and space, and realize the sustainable utilization of hydropower resources. In order to effectively convert the water energy in the reservoir into electrical energy, a hydroelectric power station needs to be realized through a hydromechanical system, which is mainly composed of a pressure diversion pipe, a water turbine, a generator, and a draft tube.

With the wide application of video surveillance technology in the safety production of hydropower stations, the scope of monitoring personnel management is increasing, the management content is increasing, and the emergencies and abnormal events faced are becoming more and more complex. The difficulty and importance of monitoring It is also becoming more and more prominent that it is becoming more and more difficult to analyze and understand the astonishing amount of video data collected only by manpower. When staring at a monitor continuously, even a very dedicated monitor will experience a significant decline in his ability to sustain his focus on the monitor. Therefore, the in-depth application of video surveillance is not only how to effectively collect on-site surveillance images, but how to effectively use new intelligent technology to analyze a large amount of video information that appears in front of the surveillance personnel, and to eliminate many non-threatening and irrelevant Screen out the video information, so that the really important information can be presented to the user more effectively, improve the efficiency and accuracy of video monitoring, assist the monitoring personnel to find and deal with problems in time, and further ensure the safe and reliable operation of the hydropower station.

However, there is no such way to intelligently identify risks and track people or objects at risk, so as to ensure accurate and efficient discovery of risk points and timely disposal, so as to improve the stability and safety of hydropower station operation.

Contents of the invention

The purpose of the present invention is to provide an intelligent video monitoring linkage detection and early warning method, which solves the problem that the existing video monitoring system of a hydropower station cannot intelligently identify risks, and track people or things at risk, so as to ensure accurate risk points Efficiently discover and deal with problems in a timely manner.

In order to achieve the above object, the present invention provides the following technical solution: an intelligent video monitoring linkage detection and early warning method, which is characterized in that it includes the following steps:

S1: Build a video surveillance system, group the cameras in multiple systems with serial numbers, and set up a main and secondary screen display system that can distinguish between large and small images;

S2: Establish the existing BP convolutional neural network training system, and search for risk picture data that have an impact on the operation of the hydropower station in the big data, input them into the BP convolutional neural network training system for training, and obtain video pictures that can Risk prediction model, and then expanded to a video screen risk prediction system;

S3: Establish the existing BP convolutional neural network training system, determine the pictures taken by multiple grouped cameras, collect the moving directions of people and objects in the monitoring pictures taken by each grouped camera, and which will appear in the subsequent Organize the camera, and use this data to train the prediction model that can predict the movement trajectory of people and objects in the portrait through the BP convolutional neural network training system, and then expand it into a picture trajectory prediction and tracking system;

S4: In the specific linkage detection and early warning, first analyze the video captured by the camera, and use the video screen risk prediction system to predict whether there is a risk of endangering the operation of the hydropower station. If there is, an alarm will be issued, and the alarm will be displayed through the screen;

S5: When the video picture risk prediction system judges that there is risk, then further use the picture trajectory prediction and tracking system to predict which grouping camera the next trajectory of the person or object in the portrait will appear in;

S6: Control the backstage display to present the pictures taken by the risky grouped cameras in the form of a large picture, while other grouped cameras are displayed or hidden in a form smaller than the picture, and when the target disappears in the big picture, according to Predict which group camera screen will appear in the next step, and use the camera's shooting screen to display the big picture screen.

Preferably, the video monitoring system in the S1 also includes an integrated monitoring management platform and a camera terminal, each of the camera terminals is connected to the integrated monitoring management platform, and the integrated monitoring management platform is also respectively connected to an intelligent video analysis system and a camera terminal. Grouping system, the large database is connected with a BP convolutional neural network training system, and the BP convolutional neural network training system is respectively connected with a video picture risk prediction system and a picture trajectory prediction tracking system, and the intelligent video analysis system is connected successively There is a video picture risk prediction system and a picture trajectory prediction and tracking system. The camera grouping system is connected with a monitoring picture switching system and a main and secondary picture display system in turn. The video picture risk prediction system is connected with an alarm system, and the alarm system is connected with a The main and secondary screen display system, the system for predicting and tracking the portrait trajectory is connected with a monitoring screen switching system.

Preferably, the integrated monitoring and management platform is used for centralized management of each camera terminal, the camera grouping system is used for numbering each camera according to the shooting location angle, and the monitoring screen switching system is used for switching the main and secondary screen display systems The main and sub screens displayed on the screen.

Preferably, the intelligent picture risk prediction system is used to predict the risk of the video picture through the prediction model trained by the BP convolutional neural network training system, and the picture trajectory prediction tracking system is used to train the system through the BP convolutional neural network The track prediction model trained is used to predict the track of people or objects in the video.

Preferably, the warning system is used to display a warning when there is a risk in the video picture, and the monitoring picture switching system is used to switch different cameras according to the trajectory prediction of people or objects in the picture, and through the main and secondary picture display system Display the difference.

Preferably, the data input into the BP convolutional neural network training system in S2 to train the video picture risk prediction model includes the behavior of people who are at risk of harm to the hydropower station, people's walking tracks, people's body movements, people's carried Objects, human expressions, volumes, shapes, and trajectories of objects.

Preferably, the data input to the BP convolutional neural network training system in S2 to train the video picture risk prediction model also includes fire data, building structure crack data, and flood data.

Preferably, the data input to the BP convolutional neural network training system in the S3 training picture trajectory prediction and tracking system includes the angle at which people or objects disappear in this picture, and the data in which one of the group cameras appears .

Preferably, the image trajectory prediction and tracking system in S3 can also predict that a target person or object will appear in the images of the two cameras.

Preferably, the display format of the primary and secondary screen display system in S6 is 1+12 or 2+8.

Compared with related technologies, an intelligent video surveillance linkage detection and early warning method provided by the present invention has the following beneficial effects:

1. The present invention provides a method for intelligent video monitoring linkage detection and early warning, by collecting data in a large database in advance, and setting up a BP convolutional neural network training system, and training by inputting different data sets, and training video images respectively The risk prediction system and the screen trajectory prediction and tracking system, and in specific applications, can be used to predict whether there is a risk in the surveillance video screen, and to predict and track the trajectory of people or objects at risk, so as to facilitate data analysis. To screen and mainly identify events that affect the safe operation of hydropower stations or violate management regulations and their accompanying persons.

2. The present invention provides an intelligent video monitoring linkage detection and early warning method. By setting up an intelligent video analysis system, it is convenient to analyze and identify the captured pictures of the camera, and then through the video picture risk prediction system and the picture trajectory prediction and tracking system. Carry out analysis, and when the risk is predicted by the video screen risk prediction system, the screen trajectory prediction and tracking system will conduct further trajectory prediction and tracking of the judged risk target, that is, people or objects, and at the same time send a signal to the alarm system for alarm , and the alarm system controls the main and secondary screen display system to display in it, so as to facilitate timely alarming of risk points.

3. The present invention provides an intelligent video monitoring linkage detection and early warning method. By setting up a camera grouping system, each camera can be sequenced and numbered according to the shooting area, and it is also connected with a monitoring screen switching system, which is convenient. In specific use, when the screen trajectory prediction and tracking system predicts which group camera a person or object will move to next, and the main and sub-picture display system shows that the target in the main screen has disappeared, the predicted group camera will shoot The screen displayed in the main and sub-screen display system is displayed on the main screen.

The method has the characteristics of convenient and intelligent monitoring of hydropower stations, effective identification of risk points on the monitoring screen, tracking and display of risk targets, and effective and timely warning of risks.

Description of drawings

Fig. 1 is a system diagram of the present invention.

Fig. 2 is a flowchart of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, not all of them; based on The embodiments of the present invention and all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

Embodiment one:

Please refer to Figures 1-2, the present invention provides a technical solution: a method for intelligent video surveillance linkage detection and early warning, which is characterized in that it includes the following steps:

S1: Build a video surveillance system, group the cameras in multiple systems with serial numbers, and set up a main and secondary screen display system that can distinguish between large and small images;

S2: Establish the existing BP convolutional neural network training system, and search for risk picture data that have an impact on the operation of the hydropower station in the big data, input them into the BP convolutional neural network training system for training, and obtain video pictures that can Risk prediction model, and then expanded to a video screen risk prediction system;

S3: Establish the existing BP convolutional neural network training system, determine the pictures taken by multiple grouped cameras, collect the moving directions of people and objects in the monitoring pictures taken by each grouped camera, and which will appear in the subsequent Organize the camera, and use this data to train the prediction model that can predict the movement trajectory of people and objects in the portrait through the BP convolutional neural network training system, and then expand it into a picture trajectory prediction and tracking system;

S4: In the specific linkage detection and early warning, first analyze the video captured by the camera, and use the video screen risk prediction system to predict whether there is a risk of endangering the operation of the hydropower station. If there is, an alarm will be issued, and the alarm will be displayed through the screen;

S5: When the video picture risk prediction system judges that there is risk, then further use the picture trajectory prediction and tracking system to predict which grouping camera the next trajectory of the person or object in the portrait will appear in;

S6: Control the backstage display to present the pictures taken by the risky grouped cameras in the form of a large picture, while other grouped cameras are displayed or hidden in a form smaller than the picture, and when the target disappears in the big picture, according to Predict which group camera screen will appear in the next step, and use the camera's shooting screen to display the big picture screen.

The video monitoring system in S1 also includes a comprehensive monitoring management platform and camera terminals. Each camera terminal is connected to the comprehensive monitoring management platform. The comprehensive monitoring management platform is also connected to an intelligent video analysis system and a camera grouping system. The convolutional neural network training system and the BP convolutional neural network training system are respectively connected with the video image risk prediction system and the image trajectory prediction and tracking system. The intelligent video analysis system is connected with the video image risk prediction system and the image trajectory prediction and tracking system in turn, and the camera grouping The system is sequentially connected to the monitoring screen switching system and the main and secondary screen display system, the video screen risk prediction system is connected to the alarm system, the alarm system is connected to the main and secondary screen display system, and the image trajectory prediction and tracking system is connected to the monitoring screen switching system.

The comprehensive monitoring and management platform is used for centralized management of each camera terminal, the camera grouping system is used to number each camera according to the shooting location angle, and the monitoring screen switching system is used to switch the main and secondary screens displayed on the main and secondary screen display system.

The intelligent picture risk prediction system is used to predict the risk of the video picture through the prediction model trained by the BP convolutional neural network training system, and the picture trajectory prediction and tracking system is used to predict the trajectory prediction model trained by the BP convolutional neural network training system Predict the trajectory of people or objects in the video picture.

The alarm system is used to display an alarm when there is a risk in the video screen, and the monitoring screen switching system is used to switch different cameras according to the trajectory prediction of people or objects in the screen, and display them differently through the main and secondary screen display systems.

The data input into the BP convolutional neural network training system in S2 to train the video picture risk prediction model includes the behavior of people who are at risk of harming the hydropower station, people’s walking trajectories, people’s body movements, things carried by people, people’s Expression, the volume of the object, the shape of the object, the trajectory of the object.

The data input to the BP convolutional neural network training system in S2 to train the video picture risk prediction model also includes fire data, building structure crack data, and flood data.

In S3, the data input to the BP convolutional neural network training system to train the trajectory prediction and tracking system of the picture includes the angle at which the person or object disappears in the picture, and the data of which one of the grouped cameras appears.

The image trajectory prediction and tracking system in S3 can also predict that the target person or object will appear in the images of the two cameras.

The display form of the main and sub-picture display system in S6 is 1+12 or 2+8.

In this embodiment:

By collecting data in a large database in advance, and setting up a BP convolutional neural network training system, and by inputting different data sets for training, and respectively training a video screen risk prediction system and a screen trajectory prediction and tracking system, and in specific applications , it is possible to predict whether there is a risk in the monitoring video screen through the two, and to predict and track the trajectory of risky people or objects, so as to facilitate the screening of data, and mainly identify those that affect the safe operation of the hydropower station or violate the management The normative event and its accompanying persons;

By setting up an intelligent video analysis system, it is convenient to analyze and identify the captured images of the camera, and then analyze them sequentially through the video image risk prediction system and the image trajectory prediction and tracking system. When the risk is predicted by the video image risk prediction system , then through the screen trajectory prediction and tracking system, further trajectory prediction and tracking will be carried out on the judged risk target, that is, people or objects, and at the same time, a signal will be sent to the alarm system for alarm, and the alarm system will control the main and secondary screen display systems to display in it , so as to facilitate timely alarming of risk points;

By setting up a camera grouping system, it is realized that each camera can be sequenced and numbered according to the shooting area, and it is also connected with a monitoring screen switching system, which is convenient for specific use. When the screen trajectory prediction and tracking system predicts people or If the object will move to which grouping camera in the next step, and the target in the main screen in the main and secondary screen display system has disappeared, the predicted screen shot by the grouping camera will be displayed in the main and secondary screen display system, and the main screen show.

The method has the characteristics of convenient and intelligent monitoring of hydropower stations, effective identification of risk points on the monitoring screen, tracking and display of risk targets, and effective and timely warning of risks.

Claims (10)

1. A method for intelligent video surveillance linkage detection and early warning, characterized in that, comprising the following steps: S1: Build a video surveillance system, group the cameras in multiple systems with serial numbers, and set up a main and secondary screen display system that can distinguish between large and small images; S2: Establish the existing BP convolutional neural network training system, and search for risk picture data that have an impact on the operation of the hydropower station in the big data, input them into the BP convolutional neural network training system for training, and obtain video pictures that can Risk prediction model, and then expanded to a video screen risk prediction system; S3: Establish the existing BP convolutional neural network training system, determine the pictures taken by multiple grouped cameras, collect the moving directions of people and objects in the monitoring pictures taken by each grouped camera, and which will appear in the subsequent Organize the camera, and use this data to train the prediction model that can predict the movement trajectory of people and objects in the portrait through the BP convolutional neural network training system, and then expand it into a picture trajectory prediction and tracking system; S4: In the specific linkage detection and early warning, first analyze the video captured by the camera, and use the video screen risk prediction system to predict whether there is a risk of endangering the operation of the hydropower station. If there is, an alarm will be issued, and the alarm will be displayed through the screen; S5: When the video picture risk prediction system judges that there is risk, then further use the picture trajectory prediction and tracking system to predict which grouping camera the next trajectory of the person or object in the portrait will appear in; S6: Control the backstage display to present the pictures taken by the risky grouped cameras in the form of a large picture, while other grouped cameras are displayed or hidden in a form smaller than the picture, and when the target disappears in the big picture, according to Predict which group camera screen will appear in the next step, and use the camera's shooting screen to display the big picture screen. 2. A kind of intelligent video surveillance linkage detection early warning method according to claim 1, is characterized in that, the video surveillance system among the described S1 also comprises comprehensive monitoring management platform, camera terminal, and each described camera terminal is all connected with comprehensive monitoring The management platform is connected, and the integrated monitoring and management platform is also connected with an intelligent video analysis system and a camera grouping system, and the large database is connected with a BP convolutional neural network training system, and the BP convolutional neural network training system is respectively connected with a video The picture risk prediction system and the picture trajectory prediction and tracking system, the intelligent video analysis system is connected to the video picture risk prediction system and the picture trajectory prediction and tracking system in turn, and the camera grouping system is connected to the monitoring picture switching system and the main and secondary picture display in turn system, the video image risk prediction system is connected to an alarm system, the alarm system is connected to a main and secondary image display system, and the image trajectory prediction and tracking system is connected to a monitoring image switching system. 3. A kind of intelligent video monitoring linkage detection and early warning method according to claim 1, characterized in that, the integrated monitoring management platform is used for centralized management of each camera terminal, and the camera grouping system is used for each camera according to The shooting location angles are numbered, and the monitoring screen switching system is used to switch the main and secondary screens displayed on the main and secondary screen display system. 4. A kind of intelligent video surveillance linkage detection and early warning method according to claim 1, characterized in that, the intelligent picture risk prediction system is used for the risk of the video picture by the prediction model trained by the BP convolutional neural network training system Forecasting, the picture trajectory prediction and tracking system is used to predict the trajectory of people or objects in the video picture through the BP convolutional neural network training system to the trained trajectory prediction model. 5. A method for intelligent video monitoring linkage detection and early warning according to claim 1, wherein the warning system is used to display a warning when a risk occurs in the video picture, and the monitoring picture switching system is used to Different cameras are switched according to the trajectory prediction of people or objects in the middle, and are displayed differently through the main and secondary screen display systems. 6. A kind of intelligent video monitoring linkage detection and early warning method according to claim 1, is characterized in that, the data input into the BP convolutional neural network training system in the S2 to train the video picture risk prediction model includes information that is useful to the hydropower station. Hazard risks include human behavior, human walking trajectory, human body movements, objects carried by people, human expressions, volume of objects, shape of objects, and trajectory of objects. 7. A kind of intelligent video surveillance linkage detection and early warning method according to claim 1, is characterized in that, the data of training video picture risk prediction model in the BP convolutional neural network training system that is input in the described S2 also includes fire situation Data, building structure crack data, flood data. 8. A kind of intelligent video surveillance linkage detection and early warning method according to claim 1, is characterized in that, in the described S3, the data input into the BP convolutional neural network training system in the training picture trajectory prediction and tracking system includes people or objects The angle that disappears in this screen is the data of which of the group cameras appears. 9. A method for intelligent video monitoring linkage detection and early warning according to claim 1, characterized in that the picture trajectory prediction and tracking system in S3 can also predict that a target person or object will appear in the pictures of the two cameras. 10. An intelligent video surveillance linkage detection and early warning method according to claim 1, characterized in that the display format of the main and sub-picture display system in S6 is 1+12 or 2+8.