A public place security video surveillance management system and method

Abstract

The present invention relates to the technical field of surveillance video management, and specifically to a public place security video surveillance management system and method, comprising the steps of acquiring video data from all cameras within a public place surveillance area, and screening out candidate video streams for subsequent security video surveillance management from the acquired video data; calculating the similarity between video frames in the surveillance area and the candidate video streams, and performing image matching based on the calculation results; and detecting abnormal behavior through a cross-validation algorithm based on a constructed set of matching images of the public place surveillance area. The present invention automatically screens out candidate video streams by calculating the similarity between video frames and the difference in camera view angles, thereby reducing the processing burden of irrelevant video streams. With the help of multi-camera data streams and scene layout information, cross-validation is used to ensure complete coverage of events within the surveillance area. Combined with an anomaly detection algorithm, potential abnormal behaviors and safety hazards can be effectively identified, thereby improving recognition accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of surveillance video management, and in particular to a public place security video surveillance management system and method. Background Art

[0002] Public security video surveillance systems are designed to achieve real-time monitoring and post-management of people, vehicles, and objects. They ensure the safety of public places through a combination of efficient video surveillance equipment, data transmission networks, intelligent analysis technologies, and management platforms. However, at this stage, they face the following challenges:

[0003] With the widespread deployment of surveillance equipment, a large amount of video data is generated. Manual screening and analysis is not only time-consuming but may also miss key information.

[0004] Due to the lack of effective multi-view verification and scene layout correlation analysis, existing systems are prone to false positives or false negatives in complex scenarios;

[0005] Current technologies lack effective image matching methods when dealing with dynamic targets (such as fast-moving people or vehicles) and complex backgrounds. Summary of the Invention

[0006] In view of the above-mentioned problems, the present invention is proposed.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for managing security video surveillance in public places, comprising the following steps:

[0008] Obtain video data from all cameras within the public surveillance area and filter out candidate video streams for subsequent security video surveillance management from the acquired video data. Specifically:

[0009] By deploying multiple cameras to acquire video data from the surveillance area, a deviation factor is derived based on the differences between the video frames of the cameras in the surveillance area. The deviation factor is used to determine the similarity between video streams, including determining the similarity between images based on the structural similarity between each pair of video frames and determining the geometric similarity between video frames based on the difference in viewing angles between cameras. At the same time, a deviation factor threshold is set to screen candidate video streams.

[0010] Calculate the similarity between the video frame corresponding to the surveillance area video and the video frame corresponding to the candidate video stream, and perform image matching based on the calculation results. Based on the image matching results, obtain the matching image of the public place surveillance area;

[0011] In addition, for the constructed public place monitoring area matching image collection, multi-camera data streams are used and combined with camera layout information for cross-validation, while anomaly detection algorithms are used to detect potential abnormal behaviors and safety hazards.

[0012] As a preferred solution of the public place security video surveillance management method of the present invention, wherein: each pair of video frames as well as The structural similarity between them determines the similarity between images, as follows:

[0013]

[0014] in, Respectively as well as The mean of Respectively as well as The variance of express and The covariance between represents the correction term, Represents the calculated structural similarity between video frames, which is used to determine the similarity between images.

[0015] As a preferred solution of the public place security video surveillance management method of the present invention, the geometric similarity between video frames is determined by the viewing angle difference between the cameras as follows:

[0016]

[0017] in, Represents the camera and The viewing angle vector, represents the dot product operation, Respectively represent the modulus corresponding to the viewing angle vector, It represents the difference in viewing angle between two cameras and is the geometric similarity between the video frames corresponding to the cameras.

[0018] As a preferred solution of the public place security video surveillance management method of the present invention, the setting of the deviation factor threshold to screen the candidate video streams is specifically as follows:

[0019] According to the determined image similarity and geometric similarity, a deviation factor is constructed, specifically:

[0020]

[0021] in, Represents the calculated structural similarity between video frames. To determine the similarity between images, Represents the difference in viewing angle between the two cameras, which is the geometric similarity between the video frames corresponding to the cameras. Represents the weight coefficient, which is set by the implementer according to the actual application scenario. Indicates the calculated deviation factor, which is used to implement video stream selection, specifically:

[0022] Setting the Deviation Factor Threshold ,

[0023] If the calculated deviation factor meets the deviation factor threshold, the formula , means that at the same moment, the two frames of video corresponding to the camera have similar contents. At this time, the video data collected by the camera is used as the candidate video stream for subsequent video surveillance management;

[0024] If the calculated deviation factor meets the deviation factor threshold, the formula , indicating that at the same moment, the content of the two frames of video corresponding to the camera deviates. At this time, the video data collected by the camera is marked as an invalid video stream.

[0025] As a preferred solution of the public place security video surveillance management method of the present invention, the similarity between the video stream of the monitoring area and the candidate video stream is calculated as follows:

[0026] Arbitrarily select the video data collected by the camera in the monitoring area as the video data in the monitoring area, and construct the video data in the monitoring area into the form of video frames, then,

[0027]

[0028] in, Indicates a point in time, Indicates time The monitoring frame at Indicates time The monitoring frame at A set of video frames representing the monitored area;

[0029] The filtered candidate videos are used to construct a candidate video stream frame set, then, ,in, Indicates the candidate video streams, Indicates a point in time, a set of video frames representing a candidate video stream;

[0030] The two video frames are converted into corresponding feature vectors through the same convolutional neural network as well as , respectively represent monitoring frames The feature vector and candidate video frame The eigenvector of

[0031] According to the transformed eigenvector, the Euclidean distance between the two is calculated, then,

[0032]

[0033] in, Indicates the dimension sequence number of the feature vector, represents the upper limit of the dimension of the feature vector, 、 Respectively represent The feature vectors corresponding to the two videos in dimension, Indicates the calculated Euclidean distance, which is used to determine the degree of matching between two frames.

[0034] As a preferred solution of the public place security video surveillance management method of the present invention, the image matching according to the calculation results is specifically as follows:

[0035] Set the matching degree judgment threshold ,

[0036] If the Euclidean distance between the two video frames and the threshold value satisfy the formula , indicating that the two video frames match;

[0037] If the Euclidean distance between the two video frames and the threshold value satisfy the formula , indicating that the two video frames do not match;

[0038] Traverse all video frame pairs , and the results that meet the matching degree form a set of matching images, then,

[0039]

[0040] in, Represents the formed matching image set of public place monitoring area, represents the calculated Euclidean distance, Indicates the matching degree judgment threshold, Represents a video frame pair.

[0041] As a preferred solution of the public place security video surveillance management method of the present invention, the cross-validation is specifically as follows:

[0042] Build scene layout model based on camera deployment information , including, camera position , camera angle of view And scene partition information ;

[0043] Based on the historical abnormal behavior database, set the abnormal behavior label set ;

[0044] Before performing abnormal behavior detection, cross-validate the constructed scene layout, specifically:

[0045]

[0046] in, Indicates the Camera and The distance between the cameras, Indicates the distance threshold between the two cameras. Indicates scene partition information, Used to limit the division of scene partitions to not exceed the scene partition range information, Indicates the viewing angle of the camera. Indicates the minimum number of covered cameras. Indicates the indicator function, when the value is 1, it means Covered by the camera's view, Indicates the total number of deployed cameras, Indicates the total number of scene partitions.

[0047] As a preferred solution of the public place security video surveillance management method of the present invention, the method of using anomaly detection algorithm to detect potential abnormal behaviors and safety hazards is specifically as follows:

[0048] If the behavior feature vector and behavior tag collection The feature comparison results in satisfy the formula When , it means that the behavior feature extracted from the video data of the current monitoring area is consistent with the label in the constructed abnormal behavior label set, and the current behavior feature is abnormal behavior, where, Indicates the structural similarity between two features, Indicates the set structural similarity threshold;

[0049] If the behavior feature vector and behavior tag collection The feature comparison results in satisfy the formula , it means that the behavior features extracted from the video data of the current monitoring area do not match the labels in the constructed abnormal behavior label set, and the current behavior features are not abnormal behaviors.

[0050] A public place security video surveillance management system, including a video data acquisition and processing module, an image matching module and an abnormal behavior recognition module.

[0051] The video data acquisition and processing module is used to acquire camera video data within the monitoring area of ​​the public place and filter out candidate video streams from the video data;

[0052] The image matching module performs image matching on the video frame corresponding to the surveillance area video and the candidate video stream, and acquires the matching image of the public place surveillance area based on the image matching result;

[0053] The abnormal behavior identification module detects potential abnormal behaviors and safety hazards based on the constructed public place monitoring area matching image set.

[0054] Beneficial effects of the present invention:

[0055] The present invention automatically screens out candidate video streams by calculating the similarity between video frames and the difference in camera angles, reducing the processing burden of irrelevant video streams, thereby significantly improving the efficiency of the security video surveillance system;

[0056] Leveraging multi-camera data streams and scene layout information, cross-validation ensures complete coverage of events within the monitoring area. Combined with anomaly detection algorithms, it can effectively identify potential abnormal behaviors and safety hazards, improving recognition accuracy.

[0057] The deep learning-based feature extraction method can adapt to complex and dynamically changing monitoring environments and accurately perform image matching even in the case of multiple overlapping targets or changing lighting conditions.

[0058] By constructing a matching image set and using an efficient similarity measurement method (such as Euclidean distance), the present invention significantly improves the speed and accuracy of image matching, providing technical support for real-time monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them:

[0060] Figure 1 This is a schematic diagram of the overall method steps of a public place security video surveillance management method of the present invention. DETAILED DESCRIPTION

[0061] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0062] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0063] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0064] The present invention is described in detail with reference to schematic diagrams. For ease of illustration, cross-sectional views of device structures may be partially enlarged and not to scale when describing embodiments of the present invention. Furthermore, the schematic diagrams are merely illustrative and should not limit the scope of the present invention. Furthermore, in actual production, the three-dimensional dimensions of length, width, and depth should be included.

[0065] At the same time, in the description of the present invention, it should be noted that the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0066] In this disclosure, unless otherwise specified or limited, the terms "mounted," "connected," and "connected" should be interpreted broadly. For example, they may refer to fixed, removable, or integral connections. They may also refer to mechanical, electrical, or direct connections, indirect connections through an intermediary, or internal communication between two components. Those skilled in the art will understand the specific meanings of these terms in this disclosure.

[0067] Example 1: Reference Figure 1 , as an embodiment of the present invention, provides a public place security video surveillance management method, comprising the following steps,

[0068] S1: Obtain video data from all cameras in the surveillance area of ​​a public place, and filter out candidate video streams for subsequent security video surveillance management from the acquired video data.

[0069] Specifically, the method of screening out candidate video streams for subsequent security video surveillance management from the acquired video data is to acquire the video data of the monitoring area by deploying multiple cameras, and derive a deviation factor based on the difference between the video frames of the cameras in the monitoring area. At the same time, a deviation factor threshold is set, and the calculated deviation factor is compared with the deviation factor threshold. The video stream whose deviation factor is lower than the deviation factor threshold is used as the candidate video stream, thereby completing the screening of the candidate video streams.

[0070] Furthermore, the deviation factor is used to determine the similarity between video streams, and is specifically calculated as follows:

[0071] Setting monitoring area cameras, each camera ,in, Represents the serial number of the camera and satisfies the formula , each camera captures a video frame at the same point in time , Indicates the Cameras in The image frame captured at the moment;

[0072] In order to synchronize the video streams of different cameras in the monitoring area, the frames of each video stream are time-aligned, then, , and satisfy the formula , indicating that the synchronization between the two cameras is completed;

[0073] By calculating each pair of video frames as well as The structural similarity between them determines the similarity between images, which is calculated as follows:

[0074]

[0075] in, Respectively as well as The mean of Respectively as well as The variance of express and The covariance between Indicates a correction term to prevent the denominator from being zero. The specific value is set by the implementer according to the actual application scenario. Represents the calculated structural similarity between video frames, which is used to determine the similarity between images;

[0076] The geometric similarity between video frames is determined by calculating the difference in viewing angles between cameras. The specific calculation is as follows:

[0077]

[0078] in, Represents the camera and The viewing angle vector, represents the dot product operation, Respectively represent the modulus corresponding to the viewing angle vector, Represents the perspective difference between the two cameras, which is the geometric similarity between the video frames corresponding to the cameras;

[0079] According to the determined image similarity and geometric similarity, a deviation factor is constructed, specifically:

[0080]

[0081] in, Represents the calculated structural similarity between video frames. To determine the similarity between images, Represents the difference in viewing angle between the two cameras, which is the geometric similarity between the video frames corresponding to the cameras. Represents the weight coefficient, which is set by the implementer according to the actual application scenario. Indicates the calculated deviation factor, which is used to implement video stream selection, specifically:

[0082] Setting the Deviation Factor Threshold ,

[0083] If the calculated deviation factor meets the deviation factor threshold, the formula , means that at the same moment, the two frames of video corresponding to the camera have similar contents. At this time, the video data collected by the camera is used as the candidate video stream for subsequent video surveillance management;

[0084] If the calculated deviation factor meets the deviation factor threshold, the formula , indicating that at the same moment, the content of the two frames of video corresponding to the camera deviates. At this time, the video data collected by the camera is marked as an invalid video stream.

[0085] S2: Calculate the similarity between the video frame of the monitoring area and the candidate video stream, and perform image matching based on the calculation results.

[0086] Specifically, the similarity between the surveillance area video frame and the candidate video stream is calculated by performing similarity calculation on the video frame corresponding to the surveillance area video and the video frame corresponding to the candidate video stream, and performing image matching based on the calculation result. The matching image of the public place surveillance area is obtained based on the image matching result, as follows:

[0087] Arbitrarily select the video data collected by the camera in the monitoring area as the video data in the monitoring area, and construct the video data in the monitoring area into the form of video frames, then,

[0088]

[0089] in, Indicates a point in time, Indicates time The monitoring frame at Indicates time The monitoring frame at A set of video frames representing the monitored area;

[0090] The filtered candidate videos are used to construct a candidate video stream frame set, then, ,in, Indicates the candidate video streams, Indicates a point in time, a set of video frames representing a candidate video stream;

[0091] The two video frames are converted into corresponding feature vectors through the same convolutional neural network as well as , respectively represent monitoring frames The feature vector and candidate video frame The eigenvector of

[0092] According to the transformed eigenvector, the Euclidean distance between the two is calculated, then,

[0093]

[0094] in, Indicates the dimension sequence number of the feature vector, represents the upper limit of the dimension of the feature vector, 、 Respectively represent The feature vectors corresponding to the two videos in dimension, Represents the calculated Euclidean distance, which is used to determine the degree of matching between two frames. Specifically:

[0095] Set the matching degree judgment threshold ,

[0096] If the Euclidean distance between the two video frames and the threshold value satisfy the formula , indicating that the two video frames match;

[0097] If the Euclidean distance between the two video frames and the threshold value satisfy the formula , indicating that the two video frames do not match;

[0098] Traverse all video frame pairs , and the results that meet the matching degree form a set of matching images, then,

[0099]

[0100] in, Represents the formed matching image set of public place monitoring area, represents the calculated Euclidean distance, Indicates the matching degree judgment threshold, Represents a video frame pair.

[0101] S3: Based on the constructed public place monitoring area matching image set, abnormal behavior detection is achieved through cross-validation algorithm.

[0102] Specifically, the detection of abnormal behavior is achieved by matching image sets against the constructed public place surveillance area, using multi-camera data streams and combining them with camera layout information for cross-validation, while using anomaly detection algorithms to detect potential abnormal behavior and safety hazards. The specific implementation is as follows:

[0103] Build scene layout model based on camera deployment information , including, camera position , camera angle of view And scene partition information ;

[0104] Based on the historical abnormal behavior database, set the abnormal behavior label set .

[0105] Before abnormal behavior detection, cross-validate the constructed scene layout, specifically:

[0106]

[0107] in, Indicates the Camera and The distance between the cameras, Indicates the distance threshold between the two cameras, which is set by the implementer according to the actual application scenario. Indicates scene partition information, Used to limit the division of scene partitions to not exceed the scene partition range information, Indicates the viewing angle of the camera. Indicates the minimum number of covered cameras. Indicates the indicator function, when the value is 1, it means Covered by camera view.

[0108] It should be noted that for the cross-validation of the scene layout, including the verification of distance, scene information and camera view coverage, if all are passed, it means that the constructed scene layout model is completed. Otherwise, it means that the constructed scene layout model is not completed. The camera position information is readjusted until the scene model is completed, and abnormal behavior detection is performed based on the constructed scene model.

[0109] Furthermore, the abnormal behavior detection is based on the monitoring area video data under the scene model, and the behavior feature vector of each frame is extracted by deep learning algorithm. , and compare the extracted behavior feature vector with the set abnormal behavior label set The features in are compared, specifically:

[0110] If the behavior feature vector and behavior tag collection The feature comparison results in satisfy the formula When , it means that the behavior feature extracted from the video data of the current monitoring area is consistent with the label in the constructed abnormal behavior label set, and the current behavior feature is abnormal behavior;

[0111] If the behavior feature vector and behavior tag collection The feature comparison results in satisfy the formula , it means that the behavior features extracted from the video data of the current monitoring area do not match the labels in the constructed abnormal behavior label set, and the current behavior features are not abnormal behaviors.

[0112] A public place security video surveillance management system, including a video data acquisition and processing module, an image matching module and an abnormal behavior recognition module.

[0113] The video data acquisition and processing module is used to acquire camera video data within the monitoring area of ​​the public place and filter out candidate video streams from the video data;

[0114] The image matching module performs image matching on the video frame corresponding to the surveillance area video and the candidate video stream, and acquires the matching image of the public place surveillance area based on the image matching result;

[0115] The abnormal behavior identification module detects potential abnormal behaviors and safety hazards based on the constructed public place monitoring area matching image set.

[0116] Furthermore, if the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0117] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0118] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, and then editing, interpreting, or processing in another suitable manner as necessary, and then storing it in a computer memory.

[0119] Additionally, in order to provide a concise description of exemplary embodiments, all features of an actual embodiment may not be described (ie, those features that are not relevant to the best mode presently contemplated for carrying out the invention or those that are not relevant to implementing the invention).

[0120] It will be appreciated that in the development of any actual embodiment, as in any engineering or design project, numerous implementation-specific decisions may be made. Such a development effort may be complex and time-consuming, but will, for those of ordinary skill having the benefit of this disclosure, be a routine undertaking of design, fabrication, and production without undue experimentation.

[0121] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for managing security video surveillance in public places, characterized by: The following steps are included: Obtain video data from all cameras within the public surveillance area and filter out candidate video streams for subsequent security video surveillance management from the acquired video data. Specifically: By deploying multiple cameras to acquire video data from the surveillance area, a deviation factor is derived based on the differences between the video frames of the cameras in the surveillance area. The deviation factor is used to determine the similarity between video streams, including determining the similarity between images based on the structural similarity between each pair of video frames and determining the geometric similarity between video frames based on the difference in viewing angles between cameras. At the same time, a deviation factor threshold is set to screen candidate video streams. The setting of the deviation factor threshold to screen the candidate video streams is as follows: According to the determined image similarity and geometric similarity, a deviation factor is constructed, specifically: in, Represents the calculated structural similarity between video frames. To determine the similarity between images, Represents the difference in viewing angle between the two cameras, which is the geometric similarity between the video frames corresponding to the cameras. Represents the weight coefficient, which is set by the implementer according to the actual application scenario. Indicates the calculated deviation factor, which is used to implement video stream selection, specifically: Setting the Deviation Factor Threshold , If the calculated deviation factor meets the deviation factor threshold, the formula , means that at the same moment, the two frames of video corresponding to the camera have similar contents. At this time, the video data collected by the camera is used as the candidate video stream for subsequent video surveillance management; If the calculated deviation factor meets the deviation factor threshold, the formula , which means that at the same moment, the content of the two frames of video corresponding to the camera deviates. At this time, the video data collected by the camera is marked as an invalid video stream; Calculate the similarity between the video frame corresponding to the surveillance area video and the video frame corresponding to the candidate video stream, and perform image matching based on the calculation results. Based on the image matching results, obtain the matching image of the public place surveillance area; Traverse all video frame pairs , and the results that meet the matching degree form a set of matching images, then, in, Represents the formed matching image set of the public place monitoring area, represents the calculated Euclidean distance, Indicates the matching degree judgment threshold, Represents a video frame pair; In addition, for the constructed public place monitoring area matching image collection, multi-camera data streams are used and combined with camera layout information for cross-validation, while anomaly detection algorithms are used to detect potential abnormal behaviors and safety hazards.

2. A method for managing public place security video surveillance according to claim 1, characterized in that: Each pair of video frames as well as The structural similarity between them determines the similarity between images, as follows: in, Respectively as well as The mean of Respectively as well as The variance of express and The covariance between represents the correction term, Represents the calculated structural similarity between video frames, which is used to determine the similarity between images.

3. A method for managing public place security video surveillance according to claim 2, characterized in that: The geometric similarity between video frames is determined by the difference in viewing angles between the cameras as follows: in, Represents the camera and The viewing angle vector, represents the dot product operation, Respectively represent the modulus corresponding to the viewing angle vector, It represents the difference in viewing angle between two cameras and is the geometric similarity between the video frames corresponding to the cameras.

4. A method for managing public security video surveillance according to claim 3, characterized in that: The similarity between the monitored area video stream and the candidate video stream is calculated as follows: Arbitrarily select the video data collected by the camera in the monitoring area as the video data in the monitoring area, and construct the video data in the monitoring area into the form of video frames, then, in, Indicates a point in time, Indicates time The monitoring frame at Indicates time The monitoring frame at A set of video frames representing the monitored area; The filtered candidate videos are used to construct a candidate video stream frame set, then, ,in, Indicates the candidate video streams, Indicates a point in time, a set of video frames representing a candidate video stream; The two video frames are converted into corresponding feature vectors through the same convolutional neural network as well as , respectively represent monitoring frames The feature vector and candidate video frame The eigenvector of According to the transformed eigenvector, the Euclidean distance between the two is calculated, then, in, Indicates the dimension sequence number of the feature vector, represents the upper limit of the dimension of the feature vector, Respectively represent The feature vectors corresponding to the two videos in dimension, Indicates the calculated Euclidean distance, which is used to determine the degree of matching between two frames.

5. A method for managing public place security video surveillance according to claim 4, characterized in that: The image matching according to the calculation results is specifically as follows: Set the matching degree judgment threshold , If the Euclidean distance between the two video frames and the threshold value satisfy the formula , indicating that the two video frames match; If the Euclidean distance between the two video frames and the threshold value satisfy the formula , indicating that the two video frames do not match.

6. A method for managing public place security video surveillance according to claim 5, characterized in that: The cross validation is specifically as follows: Build scene layout model based on camera deployment information , Including camera position , camera angle of view And scene partition information ; Based on the historical abnormal behavior database, set the abnormal behavior label set ; Before performing abnormal behavior detection, cross-validate the constructed scene layout, specifically: in, Indicates the Camera and The distance between the cameras, Indicates the distance threshold between the two cameras. Indicates scene partition information, Used to limit the division of scene partitions to not exceed the scene partition range information, Indicates the viewing angle of the camera. Indicates the minimum number of covered cameras. Indicates the indicator function, when the value is 1, it means Covered by the camera's view, Indicates the total number of deployed cameras, Indicates the total number of scene partitions.

7. A method for managing public place security video surveillance according to claim 6, characterized in that: The use of anomaly detection algorithms to detect potential abnormal behaviors and security risks is specifically as follows: If the behavior feature vector and behavior tag collection The feature comparison results in satisfy the formula When , it means that the behavior feature extracted from the video data of the current monitoring area is consistent with the label in the constructed abnormal behavior label set, and the current behavior feature is abnormal behavior, where, Represents the structural similarity between two features, Indicates the set structural similarity threshold; If the behavior feature vector and behavior tag collection The feature comparison results in satisfy the formula , it means that the behavior features extracted from the video data of the current monitoring area do not match the labels in the constructed abnormal behavior label set, and the current behavior features are not abnormal behaviors.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images