Community remote monitoring video transmission method and system

Abstract

This invention relates to the field of video surveillance processing technology, specifically disclosing a method and system for transmitting video during remote community monitoring. The method includes the following steps: segmenting the monitoring video into frames at edge nodes to obtain video frames; acquiring the movement trajectory of personnel based on the video frames; obtaining the trajectory sorting; grouping personnel based on the trajectory sorting; determining the target trajectory corresponding to each group; generating an image set; determining the target video; and transmitting the image set and target video to a data center; acquiring the image set A containing the undetermined image; and extracting the target video corresponding to image set A for display. This invention can reduce the operational pressure on the monitoring system.

Description

Technical Field

[0001] This invention relates to the field of video surveillance processing technology, specifically to a method and system for remote video transmission in communities. Background Technology

[0002] With the continuous advancement of technology, remote monitoring is becoming increasingly prevalent in community management. By installing high-definition cameras, sensors, and other intelligent devices within the community, management personnel can monitor the dynamic situation of public areas in real time, such as key areas like entrances, corridors, and parking lots. This allows for the rapid detection of abnormal behavior, and by linking with intelligent devices such as alarm systems and access control systems, automatic early warning and remote control can be achieved, enabling timely handling of various security risks.

[0003] With the widespread application of surveillance equipment in communities, the number of devices and image clarity are constantly improving. Numerous high-definition cameras and advanced sensors have been installed in various areas, resulting in a dramatic increase in real-time video data transmission. This ever-increasing data volume puts significant pressure on network bandwidth, leading to a noticeable decrease in upload speed. Simultaneously, the aggregation of large amounts of data also severely tests the storage and processing capabilities of the central server, further increasing the burden on system operation. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for remote video transmission in communities, solving the following technical problems:

[0005] With the widespread application of surveillance equipment in communities, the number of devices and image clarity are constantly improving. Numerous high-definition cameras and advanced sensors have been installed in various areas, resulting in a dramatic increase in real-time video data transmission. This ever-increasing data volume puts significant pressure on network bandwidth, leading to a noticeable decrease in upload speed. Simultaneously, the aggregation of large amounts of data also severely tests the storage and processing capabilities of the central server, further increasing the burden on system operation.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A method for remotely monitoring video transmission in a community includes the following steps:

[0008] The monitoring video of a single monitoring device is acquired, and the monitoring video is transmitted to a preset edge node. The monitoring video is divided into frames at the edge node to obtain video frames. The position of a single person in a single video frame is acquired, and the positions are connected in chronological order to obtain the movement trajectory of the person corresponding to the single monitoring device.

[0009] The movement trajectories are sorted according to the timeline, denoted as trajectory sorting. The personnel are grouped based on the trajectory sorting, and the target trajectories corresponding to the groups are determined.

[0010] The person corresponding to a single target trajectory is marked as a target person, the face image of the target person is obtained, and an image set CJH = (C1, C2, ..., Cm) is generated, where Cm represents the face image of the m-th target person. All the surveillance videos corresponding to the target trajectory are recorded as target videos, and the image set and the target videos are transmitted to the data center together.

[0011] The system acquires a preset facial image of a monitoring personnel, which is designated as a pending image. It then acquires an image set A containing the pending image and extracts the target video corresponding to the image set A for display.

[0012] As a further aspect of the present invention: determining the target trajectory corresponding to the group includes:

[0013] Based on the trajectory sorting, a candidate trajectory is generated. When the similarity between any two candidate trajectories is greater than a preset similarity threshold, they are divided into the same group, and the candidate trajectories in the same group are marked as candidate trajectories. The total similarity between the target trajectory and the other candidate trajectories is maximized.

[0014] As a further aspect of the present invention: generating a trajectory to be determined based on the trajectory sorting includes:

[0015] Obtain any two adjacent movement trajectories in the trajectory sorting and label them as movement trajectory i and movement trajectory i+1 respectively. The sorting position of movement trajectory i in the trajectory sorting is to the left of the sorting position of movement trajectory i+1 in the trajectory sorting.

[0016] By connecting the end point of the movement trajectory i and the start point of the movement trajectory i+1 with a straight line, a new movement trajectory is obtained to replace the movement trajectory i and the movement trajectory i+1;

[0017] Obtain a new trajectory sort and repeat the above steps until there is only one moving trajectory in a certain trajectory sort. Record the moving trajectory as an undetermined trajectory.

[0018] As a further aspect of the present invention: obtaining the similarity of the undetermined trajectory includes:

[0019] Obtain the trajectory sorting corresponding to the undetermined trajectory j and the undetermined trajectory k, and mark them as trajectory sorting J1 and trajectory sorting K1 respectively. Obtain the movement trajectory at the y-th position in the trajectory sorting J1 and trajectory sorting K1, and record them as sub-trajectory J1y and sub-trajectory K1y respectively.

[0020] In a preset spatial rectangular coordinate system, generate the curve corresponding to the sub-trajectory J1y, obtain the tangent slope of each point on the curve, sort the tangent slopes in ascending order according to the target distance to obtain slope sorting P1, the target distance represents the distance between the point on the tangent line corresponding to the tangent slope and the starting point, and the starting point represents the point on the curve corresponding to the starting point of the sub-trajectory.

[0021] Obtain the slope sort P2 corresponding to the sub-trajectory K1y, and calculate the similarity between the sub-trajectory J1y and the sub-trajectory K1y:

[0022]

[0023] Where B represents the total number of tangent slopes in the slope sorting, and P1b and P2b represent the tangent slope at the b-th position in the slope sorting P1 and the tangent slope at the b-th position in the slope sorting P2, respectively.

[0024] The total similarity is calculated as the degree of similarity.

[0025] As a further aspect of the present invention: the process of grouping the personnel also includes:

[0026] Calculate the first difference DY = P1tot - P2tot, where P1tot and P2tot represent the total number of tangent slopes in the slope sorting P1 and the slope sorting P2, respectively;

[0027] When the first difference is greater than a preset first difference threshold, the first difference is marked as an abnormal difference, and the total number of abnormal differences is counted. When the total number is greater than a preset total threshold, the undetermined trajectory j and the undetermined trajectory k are not divided into the same group.

[0028] As a further aspect of the present invention: the process of grouping the personnel also includes:

[0029] Calculate the second difference DR = TOTj - TOTk, where TOTj and TOTk represent the total number of movement trajectories corresponding to the undetermined trajectory j and the undetermined trajectory k, respectively.

[0030] When the second difference is greater than the preset second difference threshold, the undetermined trajectory j and the undetermined trajectory k are not divided into the same group.

[0031] As a further aspect of the present invention: after transmitting the image set and the target video together to the data center, the method further includes:

[0032] The data center deletes the image set and the target video whose total storage time reaches a preset duration.

[0033] A community remote monitoring video transmission system, characterized in that it includes:

[0034] Edge transmission module: acquires the monitoring video of a single monitoring device, transmits the monitoring video to a preset edge node, divides the monitoring video into frames at the edge node to obtain video frames, acquires the position of a single person in a single video frame, connects the positions in chronological order according to the timeline, and obtains the movement trajectory of the person corresponding to the single monitoring device.

[0035] Central transmission module: Sorts the movement trajectories in chronological order, denoted as trajectory sorting; groups the personnel based on the trajectory sorting; and determines the target trajectory corresponding to each group.

[0036] The person corresponding to a single target trajectory is marked as a target person, the face image of the target person is obtained, and an image set CJH = (C1, C2, ..., Cm) is generated, where Cm represents the face image of the m-th target person. All the surveillance videos corresponding to the target trajectory are recorded as target videos, and the image set and the target videos are transmitted to the data center together.

[0037] Display module: Acquires a preset facial image of a monitoring person, denoted as a pending image, acquires the image set A containing the pending image, and the data center extracts the target video corresponding to the image set A for display.

[0038] The beneficial effects of this invention compared to the prior art are as follows:

[0039] 1) This invention performs frame segmentation, trajectory extraction, and grouping at edge nodes. Only the video segment corresponding to the final determined target trajectory and the facial image of the target person are uploaded, instead of uploading all the original video data to the server. This effectively reduces the bandwidth occupied by data transmission over the network.

[0040] 2) This invention groups the trajectories of multiple people and uses a target trajectory to represent the passage data of the same group of people. Only necessary target videos and facial images are uploaded, avoiding the occupation of server resources by repeated or redundant monitoring data, thereby extending the effective working life of the server, reducing the server retrieval and analysis delay caused by excessive data volume, and improving the overall system operating efficiency.

[0041] 3) In practical monitoring applications, it is often necessary to review the activity trajectories or time periods of specific individuals. Traditional methods require a "carpet-bombing" search through massive amounts of video files to find the target person's image, which is not only time-consuming and labor-intensive but also prone to false positives or false negatives. In this invention, by grouping and labeling personnel trajectories at edge nodes, the central side only needs to determine the group to which the person belongs based on their face (image to be determined) and then quickly lock onto the corresponding target video for playback and analysis. This significantly shortens the query path, especially in community environments with a large number of monitored personnel and complex scenarios, where retrieval speed is greatly improved. Attached Figure Description

[0042] The invention will now be further described with reference to the accompanying drawings.

[0043] Figure 1 This is a flowchart illustrating a community remote monitoring video transmission method according to the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Please see Figure 1 As shown, this invention provides a method for remote monitoring video transmission in a community, comprising the following steps:

[0046] The monitoring video of a single monitoring device is acquired, and the monitoring video is transmitted to a preset edge node. The monitoring video is divided into frames at the edge node to obtain video frames. The position of a single person in a single video frame is acquired, and the positions are connected in chronological order to obtain the movement trajectory of the person corresponding to the single monitoring device.

[0047] It should be noted that in this invention, the real-time video stream of a single monitoring device can be obtained through the RTSP protocol. After decoding the streaming media data using a video decoder, the video data is stably transmitted to a preset edge node via network transmission protocols such as TCP / IP or UDP. Upon reaching the edge node, the cv2.VideoCapture module in the OpenCV library is used for frame segmentation. This method can read the video frame by frame at a set frame rate, thereby decomposing the continuous video stream into a series of static images. Each frame provides an independent data unit for subsequent processing. After obtaining each video frame, a deep learning-based target detection algorithm, such as YOLO or SSD, is used to detect people in the image, and the position information of a single person in the frame is accurately extracted through image segmentation and localization technology. Subsequently, by introducing a target tracking algorithm, such as a Kalman filter or optical flow tracking, the positions of the same people detected in consecutive frames are matched and associated. The position information of the people in each frame is connected sequentially along the time axis using the frame timestamp, thereby forming the complete movement trajectory of the person, and finally realizing the trajectory extraction of the person corresponding to a single monitoring device.

[0048] The movement trajectories are sorted according to the timeline, denoted as trajectory sorting. The personnel are grouped based on the trajectory sorting, and the target trajectories corresponding to the groups are determined.

[0049] In a preferred embodiment of the present invention, determining the target trajectory corresponding to the group includes:

[0050] Based on the trajectory sorting, a candidate trajectory is generated. When the similarity between any two candidate trajectories is greater than a preset similarity threshold, they are divided into the same group, and the candidate trajectories in the same group are marked as candidate trajectories. The total similarity between the target trajectory and the other candidate trajectories is maximized.

[0051] It is worth noting that, based on the recorded trajectories, candidate trajectories are generated. The similarity between candidate trajectories is evaluated. When the similarity between any two candidate trajectories exceeds a preset threshold, they are grouped into the same group, and all candidate trajectories in that group are marked as candidate trajectories. Subsequently, by comparing the total similarity between candidate trajectories in each group, the candidate trajectory with the highest total similarity is selected as the target trajectory. This ensures that the selected target trajectory best represents the movement characteristics of all people in the group, thereby further improving the accuracy and efficiency of subsequent video transmission and processing.

[0052] In a preferred embodiment, generating a trajectory to be determined based on the trajectory sorting includes:

[0053] Obtain any two adjacent movement trajectories in the trajectory sorting and label them as movement trajectory i and movement trajectory i+1 respectively. The sorting position of movement trajectory i in the trajectory sorting is to the left of the sorting position of movement trajectory i+1 in the trajectory sorting.

[0054] By connecting the end point of the movement trajectory i and the start point of the movement trajectory i+1 with a straight line, a new movement trajectory is obtained to replace the movement trajectory i and the movement trajectory i+1;

[0055] Obtain a new trajectory sort and repeat the above steps until there is only one moving trajectory in a certain trajectory sort. Record the moving trajectory as an undetermined trajectory.

[0056] It is important to note that, firstly, by traversing the sorted trajectory data, two adjacent movement trajectories are obtained in sequence and marked as movement trajectory i and movement trajectory i+1, respectively, where movement trajectory i is located before movement trajectory i+1 in the data list; then, by using the coordinates of the endpoint of movement trajectory i and the starting point of movement trajectory i+1 as the two endpoints of a straight line, a new straight line segment is generated using a simple straight line equation as the fusion result of the two trajectories, thereby replacing the original two trajectories, and the updated trajectory data is re-sorted. This process is implemented through iterative loops until only one movement trajectory remains in the sorted data, which is recorded as the undetermined trajectory;

[0057] In another preferred embodiment, obtaining the similarity of the undetermined trajectory includes:

[0058] Obtain the trajectory sorting corresponding to the undetermined trajectory j and the undetermined trajectory k, and mark them as trajectory sorting J1 and trajectory sorting K1 respectively. Obtain the movement trajectory at the y-th position in the trajectory sorting J1 and trajectory sorting K1, and record them as sub-trajectory J1y and sub-trajectory K1y respectively.

[0059] In a preset spatial rectangular coordinate system, generate the curve corresponding to the sub-trajectory J1y, obtain the tangent slope of each point on the curve, sort the tangent slopes in ascending order according to the target distance to obtain slope sorting P1, the target distance represents the distance between the point on the tangent line corresponding to the tangent slope and the starting point, and the starting point represents the point on the curve corresponding to the starting point of the sub-trajectory.

[0060] Obtain the slope sort P2 corresponding to the sub-trajectory K1y, and calculate the similarity between the sub-trajectory J1y and the sub-trajectory K1y:

[0061]

[0062] Where B represents the total number of tangent slopes in the slope sorting, and P1b and P2b represent the tangent slope at the b-th position in the slope sorting P1 and the tangent slope at the b-th position in the slope sorting P2, respectively.

[0063] The total similarity is calculated as the degree of similarity.

[0064] Understandably, to precisely assess the similarity between the trajectories to be determined, the corresponding sequences of trajectory j and trajectory k in the original trajectory sorting are first obtained, and the y-th moving trajectory is extracted from each sequence as sub-trajectory J1y and sub-trajectory K1y. Then, curve fitting is performed on each sub-trajectory in a preset spatial rectangular coordinate system to generate a curve. The tangent slope at each point on the curve is calculated using the difference method or other numerical methods. Then, the trajectories are sorted in ascending order according to the distance from the starting point to the point corresponding to each tangent slope, resulting in slope sorting P1 and P2. Finally, the similarity between the two sub-trajectories is calculated based on the total number B of tangent slopes in the slope sort and the difference of tangent slopes at each corresponding position. The similarity of each sub-trajectories is then summed to obtain the total similarity. This method, through the merging of continuous trajectories and precise similarity measurement, achieves effective fusion of trajectories within groups and accurate selection of target trajectories, thereby improving the efficiency of subsequent video transmission and processing, reducing redundant data, and optimizing the overall system performance.

[0065] In a preferred embodiment, the process of grouping the personnel further includes:

[0066] Calculate the first difference DY = P1tot - P2tot, where P1tot and P2tot represent the total number of tangent slopes in the slope sorting P1 and the slope sorting P2, respectively;

[0067] When the first difference is greater than a preset first difference threshold, the first difference is marked as an abnormal difference, and the total number of abnormal differences is counted. When the total number is greater than a preset total threshold, the undetermined trajectory j and the undetermined trajectory k are not divided into the same group.

[0068] In another preferred embodiment, the process of grouping the personnel further includes:

[0069] Calculate the second difference DR = TOTj - TOTk, where TOTj and TOTk represent the total number of movement trajectories corresponding to the undetermined trajectory j and the undetermined trajectory k, respectively.

[0070] When the second difference is greater than the preset second difference threshold, the undetermined trajectory j and the undetermined trajectory k are not divided into the same group.

[0071] The person corresponding to a single target trajectory is marked as a target person, the face image of the target person is obtained, and an image set CJH = (C1, C2, ..., Cm) is generated, where Cm represents the face image of the m-th target person. All the surveillance videos corresponding to the target trajectory are recorded as target videos, and the image set and the target videos are transmitted to the data center together.

[0072] Understandably, based on the determined target trajectory, the corresponding personnel are identified and marked at edge nodes to ensure the system accurately distinguishes the target personnel. Using a pre-trained personnel recognition algorithm, personnel matching the target trajectory can be quickly filtered from a large amount of personnel data. Subsequently, the built-in face detection module (e.g., using a detection algorithm based on convolutional neural networks) extracts the target personnel's face images from real-time video frames and performs preprocessing such as normalization and image enhancement to generate a clear sequence of face images, which are then assembled into an image set. Simultaneously, by performing timestamp synchronization processing on the video data within the time period covered by the target trajectory, all monitoring videos within that segment are integrated and identified as the target video, ensuring that the video content accurately corresponds to the personnel trajectory. Finally, using data compression and encryption technology, the generated image set and the target video are packaged and transmitted to the data center to achieve secure data storage and efficient retrieval. This not only ensures accurate confirmation of the target personnel's identity and precise matching of key video data but also greatly reduces the redundancy of transmitted data and improves the efficiency of subsequent system retrieval and analysis.

[0073] Obtain a preset facial image of a monitoring personnel, denoted as a pending image, obtain the image set A containing the pending image, and the data center extracts the target video corresponding to the image set A for display;

[0074] In a preferred embodiment of the present invention, after transmitting the image set and the target video together to the data center, the method further includes:

[0075] The data center deletes the image set and the target video whose total storage time reaches a preset duration.

[0076] A community remote monitoring video transmission system includes:

[0077] Edge transmission module: acquires the monitoring video of a single monitoring device, transmits the monitoring video to a preset edge node, divides the monitoring video into frames at the edge node to obtain video frames, acquires the position of a single person in a single video frame, connects the positions in chronological order according to the timeline, and obtains the movement trajectory of the person corresponding to the single monitoring device.

[0078] Central transmission module: Sorts the movement trajectories in chronological order, denoted as trajectory sorting; groups the personnel based on the trajectory sorting; and determines the target trajectory corresponding to each group.

[0079] The person corresponding to a single target trajectory is marked as a target person, the face image of the target person is obtained, and an image set CJH = (C1, C2, ..., Cm) is generated, where Cm represents the face image of the m-th target person. All the surveillance videos corresponding to the target trajectory are recorded as target videos, and the image set and the target videos are transmitted to the data center together.

[0080] Display module: Acquires a preset facial image of a monitoring person, denoted as a pending image, acquires the image set A containing the pending image, and the data center extracts the target video corresponding to the image set A for display.

[0081] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for transmitting video for remote monitoring in a community, characterized in that, Includes the following steps: The monitoring video of a single monitoring device is acquired, and the monitoring video is transmitted to a preset edge node. The monitoring video is divided into frames at the edge node to obtain video frames. The position of a single person in a single video frame is acquired, and the positions are connected in chronological order to obtain the movement trajectory of the person corresponding to the single monitoring device. The movement trajectories are sorted according to the timeline, denoted as trajectory sorting. The personnel are grouped based on the trajectory sorting, and the target trajectories corresponding to the groups are determined. The person corresponding to a single target trajectory is marked as a target person, and the face image of the target person is obtained to generate an image set CJH = (C1, C2, ..., C...). m ), C m The image represents the face of the m-th target person. All the surveillance videos corresponding to the target trajectory are recorded as target videos. The image set and the target videos are transmitted together to the data center. Obtain a preset facial image of a monitoring personnel, denoted as a pending image, obtain the image set A containing the pending image, and the data center extracts the target video corresponding to the image set A for display; Determining the target trajectory corresponding to the group includes: Based on the trajectory sorting, undetermined trajectories are generated. When the similarity between any two undetermined trajectories is greater than a preset similarity threshold, they are divided into the same group, and the undetermined trajectories in the same group are marked as candidate trajectories. The total similarity between the target trajectory and the other candidate trajectories is maximized. Obtaining the similarity of the undetermined trajectories includes: Obtain the trajectory sorts corresponding to the undetermined trajectories j and k, and label them as trajectory sort J1 and trajectory sort K1, respectively. Obtain the movement trajectory at the y-th position in the trajectory sorts J1 and K1, and denote it as sub-trajectory J1, respectively. y Sum trajectory K1 y ; The sub-trajectory J1 is generated in a preset spatial rectangular coordinate system. y The corresponding curve is obtained, and the tangent slope of each point on the curve is obtained. The tangent slopes are sorted in ascending order according to the target distance to obtain the slope sort P1. The target distance represents the distance between the point on the tangent line corresponding to the tangent slope and the starting point. The starting point represents the point on the curve corresponding to the starting point of the sub-trajectory. Obtain the sub-trajectory K1 y The corresponding slope sorting P2 is used to calculate the sub-trajectory J1. y and the sub-trajectory K1 y Similarity between: ； Where B represents the total number of tangent slopes in the slope sorting, and P1 b P2 b These represent the tangent slope at the b-th position in the slope sorting P1 and the tangent slope at the b-th position in the slope sorting P2, respectively. The total similarity is calculated as the degree of similarity.

2. The community remote monitoring video transmission method according to claim 1, characterized in that, Generating a trajectory to be determined based on the trajectory sorting includes: Obtain any two adjacent movement trajectories in the trajectory sorting and label them as movement trajectory i and movement trajectory i+1 respectively. The sorting position of movement trajectory i in the trajectory sorting is to the left of the sorting position of movement trajectory i+1 in the trajectory sorting. By connecting the end point of the movement trajectory i and the start point of the movement trajectory i+1 with a straight line, a new movement trajectory is obtained to replace the movement trajectory i and the movement trajectory i+1; Obtain a new trajectory sort and repeat the above steps until there is only one moving trajectory in a certain trajectory sort. Record the moving trajectory as an undetermined trajectory.

3. The community remote monitoring video transmission method according to claim 1, characterized in that, The process of grouping the personnel also includes: Calculate the first difference DY=P1 tot -P2 tot P1 tot P2 tot These represent the total number of tangent slopes in the slope sorting P1 and the slope sorting P2, respectively. When the first difference is greater than a preset first difference threshold, the first difference is marked as an abnormal difference, and the total number of abnormal differences is counted. When the total number is greater than a preset total threshold, the undetermined trajectory j and the undetermined trajectory k are not divided into the same group.

4. The community remote monitoring video transmission method according to claim 1, characterized in that, The process of grouping the personnel also includes: Calculate the second difference DR=TOT j -TOT k TOT j TOT k These represent the total number of movement trajectories corresponding to the undetermined trajectory j and the undetermined trajectory k, respectively; When the second difference is greater than the preset second difference threshold, the undetermined trajectory j and the undetermined trajectory k are not divided into the same group.

5. A method for remote monitoring video transmission in a community according to claim 1, characterized in that, After transmitting the image set and the target video together to the data center, the following steps are also included: The data center deletes the image set and the target video whose total storage time reaches a preset duration.

6. A community remote monitoring video transmission system, characterized in that, include: Edge transmission module: acquires the monitoring video of a single monitoring device, transmits the monitoring video to a preset edge node, divides the monitoring video into frames at the edge node to obtain video frames, acquires the position of a single person in a single video frame, connects the positions in chronological order according to the timeline, and obtains the movement trajectory of the person corresponding to the single monitoring device. Central transmission module: Sorts the movement trajectories in chronological order, denoted as trajectory sorting; groups the personnel based on the trajectory sorting; and determines the target trajectory corresponding to each group. The person corresponding to a single target trajectory is marked as a target person, and the face image of the target person is obtained to generate an image set CJH = (C1, C2, ..., C...). m ), C m The image represents the face of the m-th target person. All the surveillance videos corresponding to the target trajectory are recorded as target videos. The image set and the target videos are transmitted together to the data center. Display module: acquires a preset facial image of a monitoring person, denoted as a pending image, acquires the image set A where the pending image is located, and the data center extracts the target video corresponding to the image set A for display; Determining the target trajectory corresponding to the group includes: Based on the trajectory sorting, undetermined trajectories are generated. When the similarity between any two undetermined trajectories is greater than a preset similarity threshold, they are divided into the same group, and the undetermined trajectories in the same group are marked as candidate trajectories. The total similarity between the target trajectory and the other candidate trajectories is maximized. Obtaining the similarity of the undetermined trajectories includes: Obtain the trajectory sorts corresponding to the undetermined trajectories j and k, and label them as trajectory sort J1 and trajectory sort K1, respectively. Obtain the movement trajectory at the y-th position in the trajectory sorts J1 and K1, and denote it as sub-trajectory J1, respectively. y Sum trajectory K1 y ; The sub-trajectory J1 is generated in a preset spatial rectangular coordinate system. y The corresponding curve is obtained, and the tangent slope of each point on the curve is obtained. The tangent slopes are sorted in ascending order according to the target distance to obtain the slope sort P1. The target distance represents the distance between the point on the tangent line corresponding to the tangent slope and the starting point. The starting point represents the point on the curve corresponding to the starting point of the sub-trajectory. Obtain the sub-trajectory K1 y The corresponding slope sorting P2 is used to calculate the sub-trajectory J1. y and the sub-trajectory K1 y Similarity between: ； Where B represents the total number of tangent slopes in the slope sorting, and P1 b P2 b These represent the tangent slope at the b-th position in the slope sorting P1 and the tangent slope at the b-th position in the slope sorting P2, respectively. The total similarity is calculated as the degree of similarity.

Images