Natural park intelligent monitoring method and system based on edge calculation

Abstract

The invention discloses a natural park intelligent monitoring method and system based on edge computing, and relates to the technical field of video monitoring, a park monitoring range is determined, a plurality of monitoring points are arranged in the park monitoring range, edge servers are configured corresponding to the monitoring points to serve as edge computing nodes, and AI models are configured corresponding to the edge computing nodes; park videos are collected through edge computing nodes, the park videos are analyzed through an AI model, abnormal data are obtained, and early warning is carried out; and dividing the park video exceeding the preset storage time period in the edge computing node and the park video corresponding to the abnormal data according to the preset time period to obtain a plurality of unit videos. According to the invention, the situation that an external network obtains the video box to obtain accurate park videos can be avoided, good confidentiality is achieved, inaccurate park videos can be perceived through a video base point and cannot be combined, and accurate and safe park videos can be transmitted and provided.

Description

Technical Field

[0001] This invention relates to the field of video surveillance technology, specifically to a method and system for intelligent monitoring of natural parks based on edge computing. Background Technology

[0002] With the acceleration of urbanization, the construction of smart cities has become an important way to improve urban management efficiency and the quality of life for residents. As an important component of smart cities, the safety and management level of parks has attracted much attention. Parks are usually large in area, have complex environments, and attract a large number of visitors. Traditional manpower patrols and simple monitoring equipment are insufficient to meet management needs. Introducing edge computing can enable rapid response to events within parks. Due to the inherent security issues within parks, it is necessary to provide the collected park video to a remote data center for analysis and storage. Therefore, a system is needed to ensure the security of park video during transmission. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for intelligent monitoring of natural parks based on edge computing, so as to solve the problems in the background technology.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a smart monitoring method for natural parks based on edge computing, comprising the following steps: Determine the park's monitoring range, deploy multiple monitoring points within the park's monitoring range, configure edge servers as edge computing nodes for each monitoring point, and configure AI models for each edge computing node. Park videos are collected by edge computing nodes, and the videos are analyzed by AI models to obtain abnormal data and issue warnings. The abnormal data includes abnormal events and their corresponding location information. Park videos that exceed the preset storage period and park videos corresponding to abnormal data in the edge computing nodes are divided into multiple unit videos according to the preset period. Multiple unit videos are segmented and interleaved in an interleaved data plane. By sorting the data plane and interleaving the multiple unit videos, video boxes are obtained. The video boxes are then transmitted to a remote data center for storage via edge computing nodes.

[0005] In a preferred embodiment, the steps of determining the park monitoring range, deploying multiple monitoring points within the park monitoring range, configuring edge servers as edge computing nodes for each monitoring point, and configuring AI models for each edge computing node include: Collect park environmental and geographic information to construct a park geographic model; The effective monitoring range of the monitoring points is determined, and the park monitoring range of the park geographical model is divided into multiple unit ranges based on the effective monitoring range. Edge computing nodes are deployed in each unit range. Configure AI models for the corresponding edge computing nodes.

[0006] In a preferred embodiment, the step of acquiring park video via edge computing nodes, analyzing the park video using an AI model, obtaining abnormal data, and issuing early warnings includes: The park video is collected by monitoring points, and the location information of the monitoring points is combined with the park video and provided to the edge server. By analyzing park videos using AI models on edge servers, abnormal events are identified. These abnormal events are then combined with their corresponding location information to form abnormal data, which is then used to issue early warnings.

[0007] In a preferred embodiment, the step of dividing park videos exceeding a preset storage period and park videos corresponding to abnormal data in the edge computing node into multiple unit videos according to the preset time period includes: Set the storage period for park videos on edge computing nodes, and use park videos that exceed the preset storage period as the first uploaded videos; The park video corresponding to the abnormal data obtained from the analysis will be used as the second uploaded video; The first and second uploaded videos are divided into preset time periods and arranged in the order of division to obtain multiple unit videos.

[0008] In a preferred embodiment, the steps of segmenting and interleaving multiple unit videos in an interleaved data plane, recording the multiple unit videos in an interleaved manner by sorting the data plane to obtain video boxes, and transmitting the video boxes to a remote data center for storage via an edge computing node include: Multiple unit videos are divided into multiple local videos by the same number of frames. Matching video base points are set between adjacent local videos according to the position of the frame segmentation. The video base points are then data-bound to the frame features in the corresponding local videos. Multiple local videos corresponding to multiple unit videos are stored in the sorting package respectively, and the connection relationship between the sorting package and the sorting data surface is established; Multiple sorting packages corresponding to a single unit video are interleaved in an interleaved data plane. The interleaving records between multiple sorting packages are recorded in the sorting data plane. The interleaved data plane and the sorting data plane are connected to obtain the video box. Edge computing nodes transmit video boxes to remote data centers.

[0009] In a preferred embodiment, the step of storing multiple local videos corresponding to multiple unit videos in a sorting package and establishing a connection between the sorting package and the sorting data plane includes: Multiple sorting points are set in the edge computing node, and the sorting points are arranged in a network position. Adjacent sorting points are connected to form a sorting data surface in the form of a grid connection. Set up a sorting package with the same number of sorting points, and store the multiple local videos corresponding to multiple unit videos in the sorting package in order. Connect the multiple sorting packages corresponding to a single unit video with the adjacent sorting points in the sorting data plane.

[0010] In a preferred embodiment, the step of connecting the interlaced data plane and the combed data plane to obtain the video box includes: Construct an interlaced data surface, in which an interlaced data surface is obtained by connecting multiple interlaced points in a grid connection manner, and there is a one-to-one correspondence between the interlaced points of the interlaced data surface and the combing points of the combed data surface; The positions of the sorting packets corresponding to a single unit video are interleaved, and the interleaved data packets are stored in the interleaving points in the interleaved data plane in the interleaved order; Multiple sorting points connected to a sorting package of a single unit video record the interleaving record of the corresponding sorting packages. The interleaving record includes the interleaving order and the corresponding interleaving position between the multiple sorting packages. By connecting the intersection points of the interlaced data surface with the sorting points of the sorted data surface according to the corresponding relationship, a video box is obtained.

[0011] In a preferred embodiment, the step of the edge computing node transmitting the video box to a remote data center includes: Edge computing nodes transmit video boxes to remote data centers; After the remote data center receives the video box, it restores the position of the sorting package according to the interlacing record recorded at the sorting point, extracts the local video in the sorting package, and combines multiple local videos corresponding to a single unit video in sequence based on the video base point to obtain the unit video. If the video base points between multiple local videos cannot be matched, they cannot be combined, which means that the abnormal video is stored in the remote data center and the management terminal is notified. The park video is obtained by combining multiple unit videos according to the division order.

[0012] This invention also provides an intelligent monitoring system for nature parks based on edge computing, comprising: The deployment module is used to determine the monitoring range of the park, deploy multiple monitoring points within the monitoring range, configure edge servers as edge computing nodes for the corresponding monitoring points, and configure AI models for the corresponding edge computing nodes. The early warning module, connected to the deployment module, is used to collect park videos through edge computing nodes, analyze the park videos through AI models, obtain abnormal data, and issue early warnings. The abnormal data includes abnormal events and their corresponding location information. The video processing module, connected to the early warning model, is used to divide park videos that exceed the preset storage period and park videos corresponding to abnormal data in the edge computing nodes into multiple unit videos according to the preset period. The transmission and storage module, connected to the video processing module, is used to segment and interleave multiple unit videos in an interleaved data plane. By sorting the data plane and interleaving the multiple unit videos, a video box is obtained. The video box is then transmitted to a remote data center for storage via an edge computing node.

[0013] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention interleave the transmission of park videos after image segmentation. During the transmission of park videos, it can prevent external networks from obtaining accurate park videos, thus providing good confidentiality. At the same time, inaccurate park videos can be detected through video base points and cannot be combined. It can transmit accurate and secure park videos and has a good function for park situation analysis. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0015] Figure 1 This is a flowchart of the method of the present invention.

[0016] Figure 2 This is a system block diagram of the present invention.

[0017] Figure 3 This is a logic block diagram of the video box of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0019] Example 1, please refer to Figure 1 and Figure 3 As shown in the figure, the intelligent monitoring method for natural parks based on edge computing described in this embodiment includes the following steps: S1. Determine the park's monitoring range, deploy multiple monitoring points within the park's monitoring range, configure edge servers as edge computing nodes for the corresponding monitoring points, and configure AI models for the corresponding edge computing nodes. S2. Collect park videos through edge computing nodes, analyze the park videos through AI models, obtain abnormal data and issue warnings. The abnormal data includes abnormal events and their corresponding location information. S3. Divide the park videos that exceed the preset storage period and the park videos corresponding to abnormal data in the edge computing nodes into multiple unit videos according to the preset period. S4. Divide and interleave multiple unit videos in an interleaved data plane. By sorting the data plane, interleave and record multiple unit videos to obtain video boxes. Transmit the video boxes to a remote data center for storage through edge computing nodes. As described in steps S1-S4 above, the park video is divided into time periods and then the images are segmented. During the transmission of the park video, it is possible to prevent external networks from obtaining the video box and thus obtain accurate park video, which has good confidentiality. At the same time, inaccurate park video can be detected through video base points and cannot be combined. It can transmit accurate and secure park video and has a good function for park situation analysis.

[0020] In one embodiment, step S1, which involves determining the park monitoring range, deploying multiple monitoring points within the park monitoring range, configuring edge servers as edge computing nodes for each monitoring point, and configuring AI models for each edge computing node, includes: S11. Collect park environmental and geographic information to construct a park geographic model; S12. Determine the effective monitoring range of the monitoring points. Based on the effective monitoring range, divide the park monitoring range of the park geographic model into multiple unit ranges. Deploy edge computing nodes in each unit range. S13. Configure AI models for corresponding edge computing nodes; As described in steps S11-S13 above, nature parks, as important places for ecological protection, scientific research and monitoring, and public visits, need to take into account both ecological safety (such as fire early warning) and visitor safety (such as rescue of lost people and violation of rules). By configuring corresponding edge computing cloud servers at monitoring points, data transmission to remote data centers can be eliminated. Edge computing (deploying computing power at edge nodes close to the data source) effectively solves the pain points of judgment delays due to long transmission distances, achieving low-latency and highly reliable intelligent monitoring. Furthermore, AI models (such as YOLO-based target detection and CNN-based fire identification) are set up in the edge computing nodes and stored on the edge servers. Anomaly identification and analysis can be performed using the AI ​​models of the edge computing nodes, leading to early warnings. The monitoring points (high-definition cameras (including infrared night vision)) can be deployed in the park according to the effective monitoring range, which is the area that the monitoring point can monitor. The park's geographical model is then divided into multiple unit ranges based on the effective monitoring range. Edge computing nodes are deployed in each unit range (with edge servers configured as edge computing nodes corresponding to the monitoring points). The placement of these edge computing nodes in the actual park is consistent, but can also be designed according to specific circumstances (park obstruction issues, and unsuitable locations for edge computing nodes, such as water or building obstructions).

[0021] In one embodiment, step S2, which involves collecting park video via an edge computing node, analyzing the park video using an AI model, obtaining abnormal data, and issuing an early warning, includes: S21. Collect park video through monitoring points and provide the location information of the monitoring points, combined with the park video, to the edge server; S22. Analyze park videos using AI models in edge servers to identify abnormal events, combine these abnormal events with their corresponding location information as abnormal data, and issue early warnings for the abnormal data. As described in steps S21-S22 above, park videos within the effective monitoring range are collected from monitoring points. Since the location of each monitoring point is known when it is set up, the location information of the corresponding ground where the monitoring point is located is used as the location information for the park video collection. The park video is provided by the monitoring point to the edge server for storage and analysis. The AI ​​model in the edge server analyzes the park video to obtain abnormal events. The abnormal events are combined with the corresponding location information as abnormal data. The specific formation process of the AI ​​model here is as follows: through model pruning, quantization, and knowledge distillation, the complex cloud model is compressed into a lightweight version suitable for edge devices to ensure real-time operation on low-computing-power hardware; dynamic inference strategy: adjust the model accuracy according to the complexity of the scene (e.g., use high-precision detection in densely populated tourist areas and reduce computing power consumption in remote areas). After identifying abnormal data, an alert is issued. There are various methods for issuing alerts, including audible and visual warnings, notifications to park management, and contacting the fire department. The specific method can be a combination of these methods based on the actual information of the abnormal data. For example, if a fire occurs in the park, audible and visual warnings, notifications to park management, and contact with the fire department should be issued simultaneously. If there is a malfunction in park equipment, audible and visual warnings and notifications to park management are required. The selection of this notification method is pre-set based on various malfunctions; upon receiving the abnormal data, the appropriate method is directly matched and selected to issue an alert.

[0022] In one embodiment, step S3, which divides park videos exceeding a preset storage period and park videos corresponding to abnormal data in the edge computing node into multiple unit videos according to preset time periods, includes: S31. Set the storage period for park videos by the edge computing node, and use park videos that exceed the preset storage period as the first uploaded videos; S32. Use the park video corresponding to the abnormal data obtained from the analysis as the second uploaded video; S33. Divide the first uploaded video and the second uploaded video into preset time periods and arrange them in the order of division to obtain multiple unit videos.

[0023] As described in steps S31-S33 above, since the edge computing node acts as an independent temporary storage edge cloud server, it can perform simple park video analysis. However, long-term analysis of the park as a whole requires processing by a remote data center. Due to the limited operating and storage capabilities of the edge computing node, a storage period is set for the edge computing node. For example, the edge computing node is set to store park videos within 10 minutes of the current time. Through continuous collection, park videos exceeding 10 minutes of the current time will be uploaded to the remote data center. For example, if the edge computing node provides a storage duration of 12 minutes for park videos, the two minutes of park videos exceeding 10 minutes will be uploaded. After uploading, the edge computing node waits until the corresponding two-minute duration of park videos is collected before uploading the two minutes of videos exceeding 10 minutes. In addition, besides the 10 minutes exceeding the current time... Park videos outside of a certain time frame are uploaded. Additionally, if any park videos show abnormal data, the corresponding videos are immediately uploaded to a remote data center. This allows the remote data center to detect and further analyze and store the videos with abnormal data, resulting in a good response time. Before uploading, the first and second uploaded videos are divided into preset time periods and arranged in the order of division to create individual video units. For example, if the first / second uploaded video is 2 minutes long and the preset time period is 10 seconds, the 2-minute park video is divided into 12 video units. If the preset time period is 11 seconds, the 2-minute park video is divided into 10 video units. Any extra 10 seconds (less than 11 seconds) are also considered as one video unit, resulting in 11 video units. These multiple video units are arranged in the order of division.

[0024] In one embodiment, step S4, which involves segmenting and interleaving multiple unit videos in an interleaved data plane, recording the multiple unit videos in an interleaved manner by sorting the data plane to obtain video boxes, and transmitting the video boxes to a remote data center for storage via an edge computing node, includes: S41. Divide multiple unit videos into multiple local videos by the same number of screen segments. Set matching video base points between adjacent local videos according to the position of screen segmentation. Data bind the video base points to the screen features in the corresponding local videos. S42. Store the multiple local videos corresponding to the multiple unit videos in the sorting package respectively, and establish the connection relationship between the sorting package and the sorting data surface; S43. Distribute multiple sorting packages corresponding to a single unit video in an interleaved data plane. Record the interleaving records between multiple sorting packages through the sorting data plane. Connect the interleaved data plane and the sorting data plane to obtain the video box. S44, Edge computing nodes transmit video boxes to remote data centers; In one embodiment, step S42, which involves storing multiple local videos corresponding to multiple unit videos in a sorting package and establishing a connection between the sorting package and the sorting data plane, includes: S421. Set up multiple sorting points in the edge computing node, arrange the network positions among the multiple sorting points, and connect adjacent sorting points to form a sorting data surface in the form of a grid connection. S422. Set up a sorting package (virtual machine) with the same number of sorting points. Store multiple local videos corresponding to multiple unit videos in the sorting package in order. Connect the multiple sorting packages corresponding to a single unit video with adjacent sorting points in the sorting data plane.

[0025] In one embodiment, step S43, which connects the interlaced data plane and the combed data plane to obtain the video box, includes: S431. Construct an interlaced data surface, wherein the interlaced data surface is obtained by connecting multiple interlaced points (virtual machines) in a mesh connection form, and there is a one-to-one correspondence between the interlaced points of the interlaced data surface and the sorting points of the sorted data surface; S432. Interleave the sorting packets corresponding to a single unit video, and store the interleaved data packets in the interleaved data plane in the interleaved order; S433. Through multiple sorting points (virtual machines) connected to the sorting package of a single unit video, the interleaving record of the corresponding connected sorting package is recorded, wherein the interleaving record includes the interleaving order between multiple sorting packages and the corresponding interleaving position; S434. Connect the intersection points of the interlaced data surface with the sorting points of the sorted data surface according to the corresponding relationship to obtain the video box.

[0026] In one embodiment, step S44, where the edge computing node transmits the video box to a remote data center, includes: S441, Edge computing nodes transmit video boxes to remote data centers; S442. After the remote data center receives the video box, it restores the position of the sorting package according to the interlacing record recorded at the sorting point, extracts the local video in the sorting package, and combines multiple local videos corresponding to a single unit video in sequence based on the video base point to obtain the unit video. If the video base points between multiple local videos cannot be matched, they cannot be combined, which means that the video is abnormal and stored in the remote data center and the management terminal is prompted. S443. Combine multiple unit videos according to the division order to obtain the park video.

[0027] As described in steps S41-S44 above, after obtaining multiple video units divided according to time periods, in order to protect the security of the images during transmission and to determine whether the images have been tampered with, it is necessary to divide the video units according to the frame. For example, if a video unit is 1920×1080, it is then horizontally divided into four 960×540 units. The process begins by analyzing a local video segment. The image features within this segment are then read, and video base points are configured for adjacent segments. These base points serve as identifiers; for example, each segment of two adjacent local videos is assigned an identifier, and these identifiers are used for matching. If a local video needs to be segmented from two or more other local videos, then video base points must be set for each adjacent segment. If a local video is segmented from other local videos on both sides, then video base points must be set for both sides of the corresponding local video. The video base points are then data-bound to the image features within their respective local videos. If the image features in a local video change, the video base points will change accordingly. In this case, when stitching the local videos together, the video base points cannot match, preventing stitching. This indicates an anomaly in the local video; it could be due to tampering during transmission or network issues causing image loss, resulting in a change in the video base points. In either case, the segment of the video is incorrect and unusable. After obtaining multiple local videos, they need to be stored in sorting packages. The sorting package records which local videos belong to a single unit video. The sorting packages containing multiple local videos within a single unit video are connected by sorting points in the nearest sorting data plane. Although all sorting packages containing local videos are connected through a sorting data plane, the sorting packages containing local videos within the same unit video are set close together. Multiple sorting points are set in the edge computing node, and the network positions of the multiple sorting points are arranged. Adjacent sorting points are connected to form a grid-connected sorting data plane. The same number of sorting packages as sorting points are set. This sorting package is a virtual machine that can be used to store local videos. The multiple local videos corresponding to multiple unit videos are stored in sorting packages in order (this order is the order in which the unit videos are divided into local videos), and this storage order is recorded. The multiple sorting packages corresponding to a single unit video are connected to adjacent sorting points in the sorting data plane, also in the order of the layout videos.An interleaved data surface is set up, which has the same technical architecture as the sorted data surface. These are referred to as interleaved points. Multiple interleaved points are connected in a grid-like manner to form the interleaved data surface. Before the sorted package is stored in the interleaved data surface, the interleaved data surface and the sorted data surface are corresponding. Connecting the interleaved points of the interleaved data surface with the sorted points of the sorted data surface according to their correspondence yields video boxes. Multiple sorted packages are independent and clearly defined. When stored in the interleaved data surface, the positions of the sorted packages corresponding to a single video unit are interchanged and interleaved, forming an interchanged interleaved path as an interleaved record. For example, if there are four sorted packages corresponding to a single video unit, arranged sequentially, they are denoted as... The sorting packets a1, a2, a3, and a4 are stored as s1, s2, s3, and s4, respectively. The sorting points connected to these packets are sL1, sL2, sL3, and sL4, respectively. The corresponding interleaving points are sj1, sj2, sj3, and sj4. The sorting packets are then interleaved by swapping the positions of s2 and s1, then s1 and s4, and finally s4 and s3. This process of swapping positions is repeated a preset number of times. This entire process is the interleaving record. Using the above example, the final sorting packet order is: s2, s3, s4, and s1. s2, s3, s4, and s1 are stored sequentially as sj1, sj2, sj3, and sj4. The interlacing record is recorded through corresponding sorting points. To restore the original position, the reverse interlacing record is used. When the remote data center receives the video box, it restores the position of the sorting package according to the interlacing record recorded at the sorting points. Local videos within the sorting package are extracted, and multiple local videos corresponding to a single unit video are combined sequentially based on video base points to obtain a unit video. If the video base points of multiple local videos cannot match, they cannot be combined, indicating an abnormal video stored in the remote data center and alerting the management end. Multiple unit videos are combined according to the division order to obtain the park video. By dividing the park video into time periods and then segmenting the image, the transmission of park video can prevent external networks from obtaining accurate park video, providing good confidentiality. Furthermore, inaccurate park video can be detected through video base points, preventing its combination. This ensures accurate and secure transmission of park video, providing excellent park situation analysis capabilities.

[0028] Example 2, please refer to Figure 2 As shown in this embodiment, a nature park intelligent monitoring system based on edge computing includes: The deployment module is used to determine the monitoring range of the park, deploy multiple monitoring points within the monitoring range, configure edge servers as edge computing nodes for the corresponding monitoring points, and configure AI models for the corresponding edge computing nodes. The early warning module, connected to the deployment module, is used to collect park videos through edge computing nodes, analyze the park videos through AI models, obtain abnormal data, and issue early warnings. The abnormal data includes abnormal events and their corresponding location information. The video processing module, connected to the early warning model, is used to process park videos that exceed the preset storage period and park data corresponding to abnormal data from edge computing nodes. The videos are divided into multiple video units according to preset time periods; The transmission and storage module, connected to the video processing module, is used to segment and interleave multiple unit videos in an interleaved data plane. By sorting the data plane and interleaving the multiple unit videos, a video box is obtained. The video box is then transmitted to a remote data center for storage via an edge computing node.

[0029] It should be noted that dividing the park video by time period and then segmenting the image ensures that external networks cannot obtain accurate park video during transmission, thus providing good confidentiality. Furthermore, inaccurate park video can be detected and prevented from being combined using video baselines, ensuring accurate and secure transmission of park video and providing effective park situation analysis.

[0030] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for intelligent monitoring of natural parks based on edge computing, characterized in that, Includes the following steps: Determine the park's monitoring range, deploy multiple monitoring points within the park's monitoring range, configure edge servers as edge computing nodes for each monitoring point, and configure AI models for each edge computing node. Park videos are collected by edge computing nodes, and the videos are analyzed by AI models to obtain abnormal data and issue warnings. The abnormal data includes abnormal events and their corresponding location information. Park videos that exceed the preset storage period and park videos corresponding to abnormal data in the edge computing nodes are divided into multiple unit videos according to the preset period. Multiple unit videos are segmented and interleaved in an interleaved data plane. By sorting the data plane and interleaving the multiple unit videos, video boxes are obtained. The video boxes are then transmitted to a remote data center for storage via edge computing nodes.

2. The intelligent monitoring method for natural parks based on edge computing according to claim 1, characterized in that, The steps of determining the park's monitoring range, deploying multiple monitoring points within that range, configuring edge servers as edge computing nodes at each monitoring point, and configuring AI models on each edge computing node include: Collect park environmental and geographic information to construct a park geographic model; The effective monitoring range of the monitoring points is determined, and the park monitoring range of the park geographical model is divided into multiple unit ranges based on the effective monitoring range. Edge computing nodes are deployed in each unit range. Configure AI models for the corresponding edge computing nodes.

3. The intelligent monitoring method for natural parks based on edge computing according to claim 1, characterized in that, The steps of collecting park videos through edge computing nodes, analyzing the park videos using AI models, obtaining abnormal data, and issuing early warnings include: The park video is collected by monitoring points, and the location information of the monitoring points is combined with the park video and provided to the edge server. By analyzing park videos using AI models on edge servers, abnormal events are identified. These abnormal events are then combined with their corresponding location information to form abnormal data, which is then used to issue early warnings.

4. The intelligent monitoring method for natural parks based on edge computing according to claim 1, characterized in that, The step of dividing park videos exceeding a preset storage period and park videos corresponding to abnormal data in the edge computing nodes into multiple unit videos according to preset time periods includes: Set the storage period for park videos on edge computing nodes, and use park videos that exceed the preset storage period as the first uploaded videos; The park video corresponding to the abnormal data obtained from the analysis will be used as the second uploaded video; The first and second uploaded videos are divided into preset time periods and arranged in the order of division to obtain multiple unit videos.

5. The intelligent monitoring method for natural parks based on edge computing according to claim 1, characterized in that, The steps of segmenting and interleaving multiple unit videos in an interleaved data plane, recording the multiple unit videos in an interleaved manner by sorting the data plane to obtain video boxes, and transmitting the video boxes to a remote data center for storage via edge computing nodes include: Multiple unit videos are divided into multiple local videos by the same number of frames. Matching video base points are set between adjacent local videos according to the position of the frame segmentation. The video base points are then data-bound to the frame features in the corresponding local videos. Multiple local videos corresponding to multiple unit videos are stored in the sorting package respectively, and the connection relationship between the sorting package and the sorting data surface is established; Multiple sorting packages corresponding to a single unit video are interleaved in an interleaved data plane. The interleaving records between multiple sorting packages are recorded in the sorting data plane. The interleaved data plane and the sorting data plane are connected to obtain the video box. Edge computing nodes transmit video boxes to remote data centers.

6. The intelligent monitoring method for natural parks based on edge computing according to claim 5, characterized in that, The step of storing multiple local videos corresponding to multiple unit videos in a sorting package and establishing the connection between the sorting package and the sorting data surface includes: Multiple sorting points are set in the edge computing node, and the sorting points are arranged in a network position. Adjacent sorting points are connected to form a sorting data surface in the form of a grid connection. Set up a sorting package with the same number of sorting points, and store the multiple local videos corresponding to multiple unit videos in the sorting package in order. Connect the multiple sorting packages corresponding to a single unit video with the adjacent sorting points in the sorting data plane.

7. The intelligent monitoring method for natural parks based on edge computing according to claim 6, characterized in that, The step of connecting the interlaced data plane and the combed data plane to obtain the video box includes: Construct an interlaced data surface, in which an interlaced data surface is obtained by connecting multiple interlaced points in a grid connection manner, and there is a one-to-one correspondence between the interlaced points of the interlaced data surface and the combing points of the combed data surface; The positions of the sorting packets corresponding to a single unit video are interleaved, and the interleaved data packets are stored in the interleaving points in the interleaved data plane in the interleaved order; Multiple sorting points connected to a sorting package of a single unit video record the interleaving record of the corresponding linked sorting packages. The interleaving record includes the interleaving order and the corresponding interleaving position between the multiple sorting packages. By connecting the intersection points of the interlaced data surface with the sorting points of the sorted data surface according to the corresponding relationship, a video box is obtained.

8. The intelligent monitoring method for natural parks based on edge computing according to claim 7, characterized in that, The step of the edge computing node transmitting the video box to the remote data center includes: Edge computing nodes transmit video boxes to remote data centers; After the remote data center receives the video box, it restores the position of the sorting package according to the interlacing record recorded at the sorting point, extracts the local video in the sorting package, and combines multiple local videos corresponding to a single unit video in sequence based on the video base point to obtain the unit video. If the video base points between multiple local videos cannot be matched, they cannot be combined, which means that the video is abnormal and stored in the remote data center and the management terminal is notified. The park video is obtained by combining multiple unit videos according to the division order.

9. An edge computing-based intelligent monitoring system for natural parks, used to implement the edge computing-based intelligent monitoring method for natural parks as described in any one of claims 1-8, characterized in that, include: The deployment module is used to determine the monitoring range of the park, deploy multiple monitoring points within the monitoring range, configure edge servers as edge computing nodes for the corresponding monitoring points, and configure AI models for the corresponding edge computing nodes. The early warning module, connected to the deployment module, is used to collect park videos through edge computing nodes, analyze the park videos through AI models, obtain abnormal data, and issue early warnings. The abnormal data includes abnormal events and their corresponding location information. The video processing module, connected to the early warning model, is used to divide park videos that exceed the preset storage period and park videos corresponding to abnormal data in the edge computing nodes into multiple unit videos according to the preset period. The transmission and storage module, connected to the video processing module, is used to segment and interleave multiple unit videos in an interleaved data plane. By sorting the data plane and interleaving the multiple unit videos, a video box is obtained. The video box is then transmitted to a remote data center for storage via an edge computing node.

Images