A Satellite Remote Sensing-Based Method and System for Collaborative Monitoring of Unmanned Aerial Vehicles (UAVs)
By combining satellite remote sensing and UAV swarm monitoring, and employing image segmentation and network tag propagation algorithms to identify abnormal areas and formulate UAV scheduling strategies, efficient monitoring and rapid emergency response in complex areas have been achieved, improving detection accuracy and operational efficiency.
Patent Information
- Application Number
- CN202511958890.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-24
- Publication Date
- 2026-11-10
AI Technical Summary
Traditional ground-based monitoring struggles to cover remote areas, satellite remote sensing cannot achieve real-time monitoring of sudden disasters, and deep learning networks lack generalization ability in complex scenarios. Existing methods are inefficient and have a high false positive rate.
By acquiring image data through satellite remote sensing, identifying abnormal areas using image segmentation and network tag propagation algorithms, and combining this with UAV swarm collaborative monitoring, scheduling strategies and emergency response measures are formulated.
It enables high-precision monitoring and rapid response in complex areas, improving detection accuracy and operational efficiency, and solving the problems of low segmentation accuracy, poor scheduling efficiency, and delayed response in traditional methods.
Smart Images

Figure CN123071845A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of environmental monitoring technology, specifically relating to a method and system for UAV hive collaborative monitoring based on satellite remote sensing. Background Technology
[0002] Ecological and environmental monitoring serves as the "eyes and ears" and "early warning system" for ecological civilization construction, and is a core hub connecting ecological awareness, management decision-making, and governance actions. Against the backdrop of intensifying global climate change, sharp declines in biodiversity, and frequent extreme disasters, its importance has transcended the technical level, becoming a fundamental project supporting the sustainable development of human society.
[0003] Traditional ground-based monitoring (such as weather stations and water quality monitoring stations) relies on fixed locations, making it difficult to cover complex areas such as remote mountainous areas, uninhabited areas, and oceans, thus creating monitoring blind spots. Although satellite remote sensing can observe a wide area, it is limited by orbital cycles. For example, the revisit cycle of optical satellites is mostly 1 to 3 days, making it impossible to achieve real-time monitoring of sudden disasters (such as fires and dam breaks) or rapidly changing ecological processes (such as the spread of pests and diseases). In addition, optical remote sensing is greatly affected by cloud and lighting conditions. For example, during the rainy season in southern China, the cloud coverage of satellite images can reach as high as 40%. Although SAR can penetrate cloud layers, its accuracy in identifying targets under vegetation canopies (such as surface water) is low.
[0004] Existing methods that rely on manual interpretation of remote sensing satellite images for visual identification of fire points are inefficient and highly subjective. Deep learning networks lack generalization ability in complex scenarios. For example, fire identification may misidentify high-temperature rocks or industrial heat sources as fire points, and flood identification may miss detections in areas obscured by clouds due to missing data. They are unable to cope with the challenges of high-resolution remote sensing images with multiple sizes, dense distributions, and complex backgrounds.
[0005] Therefore, there is an urgent need to provide a collaborative monitoring method and system to improve the aforementioned deficiencies of existing technologies. Summary of the Invention
[0006] To address the aforementioned problems in the existing technology, this invention provides a method and system for collaborative monitoring of unmanned aerial vehicles (UAVs) based on satellite remote sensing. The technical problem to be solved by this invention is achieved through the following technical solution: In a first aspect, the present invention provides a method for collaborative monitoring of unmanned aerial vehicles (UAVs) based on satellite remote sensing, comprising: Image data of the target area is acquired through satellite remote sensing; the image data of the target area includes multiple images. Multiple images are pre-segmented, with each image divided into multiple regions. For the same image, the similarity between regions is calculated. If the similarity is greater than a threshold, an edge is added between the two regions; otherwise, no edge is added, thus constructing an image unweighted network. A preset network label propagation algorithm is used to divide the image unweighted network into communities, merging the regions corresponding to nodes belonging to the same community to achieve image segmentation and obtain the processed image data of the target region. Based on the processed image data of the target area and its characteristics, the anomaly level of the target area is determined. The pre-defined drone swarm responds to the anomaly level of the target area, adopts a pre-defined path planning algorithm, and formulates a drone scheduling strategy; at least one drone responds to the scheduling strategy, covers the target area, and collects video data of the target area. The pre-defined drone swarm responds to video data from the target area, makes decisions, and takes emergency response measures.
[0007] Secondly, the present invention also provides a method for collaborative monitoring of unmanned aerial vehicles (UAVs) based on satellite remote sensing, comprising: The data acquisition module is used to acquire image data of the target area through satellite remote sensing; the image data of the target area includes multiple images. The data preprocessing module is used to pre-segment multiple images, dividing each image into multiple regions. For the same image, the similarity between regions is calculated. If the similarity is greater than a threshold, an edge is added between the two regions; otherwise, no edge is added, thus constructing an image unweighted network. A preset network label propagation algorithm is used to divide the image unweighted network into communities, merging the regions corresponding to nodes belonging to the same community to achieve image segmentation and obtain the processed image data of the target region. The level determination model is used to determine the anomaly level of the target area based on the processed image data of the target area and the characteristics of the target area. The scheduling strategy formulation module is used to formulate a scheduling strategy for drones based on the preset anomaly level of the drone swarm response to the target area and a preset path planning algorithm; at least one drone responds to the scheduling strategy, covers the target area, and collects video data of the target area. The emergency response module is used to make decisions and take emergency response measures based on the video data of the target area from the pre-set drone swarm response.
[0008] The beneficial effects of this invention are: This invention provides a UAV swarm collaborative monitoring method and system based on satellite remote sensing. It acquires image data of a target area, segments the image data using a segmentation algorithm, identifies anomalies in the target area, and determines the anomaly level. Based on the anomaly level, the UAV swarm formulates a scheduling strategy for the UAVs. The UAVs respond to the scheduling strategy, cover the target area, and collect anomaly data. Based on the anomaly data, decisions are made and emergency measures are taken. This achieves a complete environmental monitoring design, solving core problems in traditional environmental monitoring such as low segmentation accuracy, poor scheduling efficiency, and delayed response. This invention offers significant advantages in improving detection accuracy, operational efficiency, and emergency response capabilities.
[0009] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description
[0010] Figure 1 This is a flowchart of a UAV hive cooperative monitoring method based on satellite remote sensing provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a UAV hive collaborative monitoring system based on satellite remote sensing provided in an embodiment of the present invention. Detailed Implementation
[0011] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.
[0012] Please see Figure 1 , Figure 1 This is a flowchart of a UAV hive cooperative monitoring method based on satellite remote sensing provided in an embodiment of the present invention. The UAV hive cooperative monitoring method based on satellite remote sensing provided by the present invention includes: S101. Acquire image data of the target area through satellite remote sensing; the image data of the target area includes multiple images.
[0013] Specifically, in this embodiment, firstly, image data of the target area is acquired via commercial satellites according to a preset cycle or event-driven mode; secondly, the image data of the target area is projected onto a geographic coordinate system to resolve geometric distortions caused by satellite motion and terrain undulations; finally, through time and spatial registration, the spectral information of the optical image is fused with the backscattering coefficient of SAR and the temperature information of infrared thermal imaging to improve the ability to identify abnormal areas.
[0014] S102. Perform pre-segmentation on multiple images, dividing each image into multiple regions. For the same image, calculate the similarity between regions. If the similarity is greater than a threshold, add an edge between the two regions; otherwise, do not add an edge, and construct an image unweighted network. Use a preset network label propagation algorithm to divide the image unweighted network into communities, and merge the regions corresponding to nodes belonging to the same community to achieve image segmentation and obtain the processed image data of the target region.
[0015] Specifically, in this embodiment, multiple images are pre-segmented, and each image is divided into multiple regions, including: Each image is segmented into each region , Each region represents an image network. One node; For the same image Get each region Color feature vector and texture feature vector Optional texture features include contrast, dissimilarity, homogeneity, energy, and correlation. Each region The color feature vector and texture feature vector are merged into a feature vector. Then, the eigenvectors are normalized to obtain the normalized eigenvectors. , is represented as: ; in, Representing the eigenvector The mean, Representing the eigenvector Standard deviation; Calculate each region The centroid is represented as: ; ; in, Indicates the region The x-coordinate of the centroid, Indicates the region The Middle The x-coordinate of each pixel. Indicates the region The ordinate of the centroid, Indicates the region The Middle The ordinate of each pixel. Indicates the region The number of pixels in the middle.
[0016] In this embodiment, for the same image, the similarity between regions is calculated. If the similarity is greater than a threshold, an edge is added between the two regions; otherwise, no edge is added. An image unweighted network is constructed, including: The feature similarity between regions is represented as: ; in, Indicates the region With the region Feature similarity between them Indicates the region The normalized feature vectors, Indicates the region The normalized eigenvectors; The location similarity between regions is represented as: ; ; in, Indicates the region With the region Positional similarity between them; Based on the similarity of features and location between regions, the similarity between regions is quantified and expressed as follows: ; in, Indicates the region With the region The similarity between them Indicates the weight.
[0017] It should be noted that in this embodiment, after each image is segmented into multiple regions, the color feature vector and texture feature vector of each region are extracted and merged into a normalized feature vector. This allows multi-feature fusion to avoid misjudgment caused by changes in lighting or occlusion of a single feature. For example, in the smoke area of a fire, adjacent regions may have large color differences due to high temperature, but the thermal infrared features and texture features are highly similar. Through similarity, they can be correctly associated, thus improving the segmentation accuracy.
[0018] In this embodiment, a preset network label propagation algorithm is used to divide the unweighted network of the image into communities, merging the regions corresponding to nodes belonging to the same community to achieve image segmentation, including: The core nodes of the image unweighted network are identified, and each core node is treated as an independent community and assigned a unique label. Calculate the similarity between core nodes and common neighbor nodes, update the label of the common neighbor node with the label of the core node with the greatest similarity, and add the common neighbor node with the updated label to the community of the corresponding core node to form a new community; where a common neighbor node corresponds to multiple core nodes. Calculate the local density of the core node's neighboring nodes and the average density of the community to which the core node belongs. If the local density of the core node's neighboring nodes is greater than the average density of the community to which the core node belongs, then propagate the core node's label to the neighboring nodes. The neighboring nodes that update the label join the community to which the core node belongs, forming a new community. Calculate the attraction value between the unlabeled node and each community, and update the label of the unlabeled node with the label of the community with the highest attraction value; The regions corresponding to nodes in the same community are merged to achieve image segmentation.
[0019] It should be noted that local density reflects the compactness of a single node's neighborhood, while average density reflects the overall compactness of the community; both together determine the priority of label propagation. This invention achieves efficient community partitioning and segmentation of unweighted image networks by quantifying node local density and community average density and dynamically adjusting label propagation rules.
[0020] In this embodiment, the local density of the neighboring nodes of the core node is calculated and expressed as: ; in, Representing neighboring nodes Local density, Represents the core node The set of domain nodes, Representing neighboring nodes The degree, Represents the core node The degree, This represents the sum of node degrees in an unweighted image network.
[0021] It should be noted that this embodiment uses a label propagation algorithm with a preset number of clusters. By using the diffusion of core node labels and density-driven label propagation rules, it can quickly merge the same community regions. Compared with traditional algorithms, this embodiment can reduce the complexity of label propagation and improve segmentation efficiency. In addition, by comparing local density with average density, it avoids the erroneous merging of low-density regions.
[0022] S103. Based on the processed image data of the target area and combined with the characteristics of the target area, determine the anomaly level of the target area.
[0023] Specifically, in this embodiment, the anomaly level of the target area is determined based on the disaster type, development speed, impact range, and sensitive targets. The anomaly levels include yellow, orange, and red alerts. A yellow alert triggers a single drone to operate; an orange alert triggers a group of drones to operate in formation; and a red alert triggers all drones to operate collaboratively. Optionally, the disaster type includes fire (thermal radiation intensity), flooding (inundated area or water depth), and crop disease (NDVI decline rate); the development speed includes the rate of disaster change; the impact range includes the area covered by the disaster; and sensitive targets include whether residential areas, ecological protection zones, or important infrastructure are involved.
[0024] S104. The preset drone swarm responds to the anomaly level of the target area, adopts a preset path planning algorithm, and formulates a drone scheduling strategy; at least one drone responds to the scheduling strategy, covers the target area, and collects video data of the target area.
[0025] Specifically, in this embodiment, the scheduling strategy of the UAV swarm needs to take into account global efficiency (coverage range, task completion time), local smoothness (UAV maneuverability safety) and task adaptability (the matching degree between payload and task). The following will be elaborated on three levels: macro level, micro level and payload adaptation mechanism. By combining the improved ant colony algorithm, cubic Bezier curve and modular payload management, a closed-loop scheduling with global optimization and local smoothness can be achieved.
[0026] In this embodiment, the method further includes: setting a preset anomaly level for the drone cellular network to receive the target area, analyzing the anomaly type, querying the drone payload list, matching the task type with the drone sensor capabilities, and determining the target task type.
[0027] The drone swarm automatically loads the corresponding drone sensor capabilities based on the mission type, such as disease detection or fire monitoring. By identifying the mission type and its anomaly level, it determines the types of sensors the drone needs to carry, such as multispectral, thermal imaging, and visible light cameras. For example, fires require thermal imagers, disease detection requires multispectral cameras, and floods require LiDAR and RGB cameras.
[0028] In this embodiment, aiming to maximize global task coverage and minimize energy consumption, an improved ant colony algorithm is used to generate the initial path for the drones. This addresses the problems of slow convergence and susceptibility to local optima inherent in traditional ant colony algorithms. A preset drone swarm response is used to determine the anomaly level of the target area, and a preset path planning algorithm is employed to formulate a drone scheduling strategy, including: For any given drone, by combining pheromone concentration, heuristic factors, and constraint penalties, the drone's starting point from the node is obtained. Transfer to node probability , is represented as: ; in, Indicates that the drone is from the node Transfer to node edge The concentration of pheromones reflects the quality of historical pathways. Represents the heuristic factor. , Represents a node To the node Euclidean distance, This represents the complexity cost of the terrain, such as slope and obstacle density. Indicates the terrain adaptation weight. Indicates the weight of terrain factors. Indicates pheromone concentration; optional. , Indicates the heuristic factor weights, optional. This strengthens the preference for short paths and low-complexity regions. Represents the set of unvisited nodes. Represents a node. Represents the constraint cost, if the edge Crossing communication dead zones or dangerous terrain areas incurs a penalty of 10. 3 Otherwise, it is 0. The sensitivity coefficient representing the constraint penalty is optional. This is used to balance the probabilities of normal paths and penalized paths; On a global scale, the pheromone concentration is increased for the optimal path, i.e., the path with the largest coverage area and lowest energy consumption. This is expressed as: ; ; in, This indicates the pheromone concentration after the global update. This indicates the global pheromone concentration that has not been updated. Indicates the pheromone evaporation coefficient, which is optional. , Indicates a constant, optional. , This represents the overall score of the optimal path, which is coverage area × 0.6 + energy consumption × 0.4.
[0029] Within a local range, edges on non-optimal paths are weakly reinforced to prevent invalid paths from being completely eliminated, thus updating the pheromone concentration. This is represented as: ; ; in, This indicates the pheromone concentration after a local update. This indicates the concentration of pheromones that have not been updated locally. Represents a uniformly distributed random number. , Represents a node Transfer to node The Euclidean distance; Based on the globally updated and locally updated pheromone concentrations, update the drone's slave node. Transfer to node probability Generate the initial path for the drone.
[0030] In this embodiment, the initial path generated by the macroscopic layer may contain polyline segments, such as straight lines connecting adjacent nodes, causing the UAV to frequently make sharp turns, resulting in abrupt changes in curvature and exceeding its maneuverability. The microscopic layer uses a cubic Bezier curve to smooth the initial path of the UAV, ensuring that the rate of curvature change meets requirements while satisfying dynamic constraints. The parametric equations of the cubic Bezier curve... Represented as: ; in, Represents a random number. , Indicates the starting point of the initial trajectory. Indicates the end point of the initial track. and These are different control points on the initial trajectory, used to adjust the curvature and direction of the curve; Set curvature constraints, curvature Represented as: ; in, This represents the first derivative and indicates the direction of the tangent. The second derivative represents the direction of bending. A constraint on the rate of change of curvature is set. The rate of change of curvature is obtained by differentiating the cubic Bezier curve, and is expressed as: ; in, Indicates the rate of change of curvature; optional. ≤5° / meter.
[0031] The initial path of the drone is divided into several line segments according to the node sequence. For each line segment... The control points are solved using an optimization algorithm. and This makes the curve corresponding to each line segment... Satisfying curvature constraints and rate of change of curvature constraints; adjacent smooth segments and At the connection point The first derivative is continuous at this point.
[0032] In this embodiment, the preset drone swarm response to the anomaly level of the target area adopts a preset path planning algorithm to formulate a drone scheduling strategy, and also includes: When the anomaly level of the target area is a yellow alert, the initial path of a single drone is generated using a preset path planning algorithm. When the anomaly level of the target area is orange or red, a preset path planning algorithm is used to generate initial paths for multiple drones; The collaborative generation of the initial path of the UAV employs a task partitioning and time offset strategy, dividing the target area into... There are three overlapping sub-regions with an overlap rate of 10%, and each drone is responsible for the main path of one sub-region; based on the initial position of the drone and the distance to the sub-region, each drone is assigned a specific path. The delayed takeoff time, This avoids congestion caused by multiple drones in the same area.
[0033] In this embodiment, the performance of the improved ant colony algorithm is verified through simulation experiments. Taking a forest fire monitoring task as an example, the input is: target area area 2km². 2 Terrain complexity (slope 10°~25°), communication coverage (signal blind spots 15%), number of drones: 5; Output: initial path coverage area 1.95km. 2 Coverage rate was 97.5%, longest single-path duration was 28 seconds, and there were no intersections or conflicts in the initial paths of multiple drones; parameter sensitivity: when As the value increases from 1 to 1.5, the average path length decreases by 8%, indicating that drones prefer shorter paths, but the sensitivity to complexity costs increases. At this time, the pheromone evaporation rate is moderate to avoid premature convergence.
[0034] S105. The pre-set drone swarm responds to video data from the target area, makes decisions, and takes emergency response measures.
[0035] In summary, the UAV swarm collaborative monitoring method based on satellite remote sensing provided by this invention acquires image data of the target area, segments the image data of the target area using a segmentation algorithm, identifies anomalies in the target area, and determines the anomaly level of the target area. Based on the anomaly level of the target area, the UAV swarm formulates a scheduling strategy for the UAVs, and the UAVs respond to the scheduling strategy, covering the target area and collecting anomaly data of the target area. Based on the anomaly data of the target area, further decisions are made and emergency measures are taken. In this way, a full-link design for environmental monitoring is realized, solving the core problems of low segmentation accuracy, poor scheduling efficiency, and delayed response in traditional environmental monitoring. This application has significant advantages in improving detection accuracy, operational efficiency, and emergency response capabilities.
[0036] Based on the same inventive concept, please refer to Figure 2 , Figure 2 This is a schematic diagram of a UAV hive cooperative monitoring system based on satellite remote sensing provided in an embodiment of the present invention. The present invention also provides a UAV hive cooperative monitoring system based on satellite remote sensing, applied to the UAV hive cooperative monitoring method based on satellite remote sensing provided in the above embodiments of the present invention. Examples of the monitoring method are described above and will not be repeated here. The system includes: The data acquisition module 201 is used to acquire image data of the target area through satellite remote sensing; wherein, the image data of the target area includes multiple images; The data preprocessing module 202 is used to pre-segment multiple images, dividing each image into multiple regions. For the same image, the similarity between regions is calculated. If the similarity is greater than a threshold, an edge is added between the two regions; otherwise, no edge is added, and an image unweighted network is constructed. A preset network label propagation algorithm is used to divide the image unweighted network into communities, merging the regions corresponding to nodes belonging to the same community to achieve image segmentation and obtain the processed image data of the target region. The level determination model 203 is used to determine the anomaly level of the target area based on the processed image data of the target area and the characteristics of the target area. The scheduling strategy formulation module 204 is used to formulate a scheduling strategy for drones based on the preset anomaly level of the drone swarm response to the target area and a preset path planning algorithm; at least one drone responds to the scheduling strategy, covers the target area, and collects video data of the target area. The emergency response module 205 is used to make decisions and take emergency response measures based on the video data of the target area in the pre-set drone swarm response.
[0037] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations are intended to cover non-exclusive inclusion, such that an article or device comprising a list of elements includes not only those elements but also other elements not expressly listed. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the article or device comprising said element. Terms such as "connected" or "linked" are not limited to physical or mechanical connections but can include electrical connections, whether direct or indirect. The orientations or positional relationships indicated by terms such as "upper," "lower," "left," and "right" are based on the orientations or positional relationships shown in the accompanying drawings and are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as limiting the invention.
[0038] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.
[0039] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.
Claims
1. A method for collaborative monitoring of unmanned aerial vehicles (UAVs) based on satellite remote sensing, characterized in that, include: Image data of the target area is acquired through satellite remote sensing; wherein, the image data of the target area includes multiple images; The multiple images are pre-segmented, with each image divided into multiple regions. For the same image, the similarity between the regions is calculated. If the similarity is greater than a threshold, an edge is added between the two regions; otherwise, no edge is added, thus constructing an image unweighted network. A preset network label propagation algorithm is used to divide the image unweighted network into communities, merging the regions corresponding to nodes belonging to the same community to achieve image segmentation and obtain the processed image data of the target region. Based on the processed image data of the target area and its characteristics, the anomaly level of the target area is determined. The pre-defined drone swarm responds to the anomaly level of the target area, employs a pre-defined path planning algorithm, and formulates a drone scheduling strategy; at least one of the drones responds to the scheduling strategy, covers the target area, and collects data from the target area; The preset drone swarm responds to data from the target area, makes decisions, and takes emergency response measures.
2. The UAV hive cooperative monitoring method based on satellite remote sensing according to claim 1, characterized in that, The multiple images are pre-segmented, with each image divided into multiple regions, including: Each image is segmented into each region , Each region represents an image network. One node; For the same image Get each region Color feature vector and texture feature vector ; Each region The color feature vector and texture feature vector are merged into a feature vector. The feature vector is then normalized to obtain the normalized feature vector. , represented as: ; in, Representing the eigenvector The mean, Representing the eigenvector Standard deviation; Calculate each region The centroid is represented as: ; ; in, Indicates the area The x-coordinate of the centroid, Indicates the area The Middle The x-coordinate of each pixel. Indicates the area The ordinate of the centroid, Indicates the area The Middle The ordinate of each pixel. Indicates the area The number of pixels in the middle.
3. The UAV hive cooperative monitoring method based on satellite remote sensing according to claim 2, characterized in that, For the same image, the similarity between the regions is calculated. If the similarity is greater than a threshold, an edge is added between the two regions; otherwise, no edge is added. An unweighted image network is constructed, including: The feature similarity between the regions is calculated and expressed as: ; in, Indicates the area With the region Feature similarity between them Indicates the area The normalized feature vectors, Indicates the area The normalized eigenvectors; The positional similarity between the regions is calculated and expressed as: ; ; in, Indicates the area With the region Positional similarity between them; The similarity between the regions is quantified based on their feature similarity and location similarity, and is expressed as follows: ; in, Indicates the area With the region The similarity between them Indicates the weight.
4. The UAV hive cooperative monitoring method based on satellite remote sensing according to claim 1, characterized in that, A preset network label propagation algorithm is used to divide the unweighted network of the image into communities, merging the regions corresponding to nodes belonging to the same community to achieve image segmentation, including: The core nodes of the image unweighted network are determined, and each core node is treated as an independent community and assigned a unique label. Calculate the similarity between the core node and its common neighbor nodes, update the label of the common neighbor node with the label of the core node with the highest similarity, and add the common neighbor node with the updated label to the community of the corresponding core node to form a new community; wherein, the common neighbor node corresponds to multiple core nodes; Calculate the local density of the neighboring nodes of the core node, and calculate the average density of the community to which the core node belongs. If the local density of the neighboring nodes of the core node is greater than the average density of the community to which the core node belongs, then propagate the label of the core node to the neighboring nodes, and the neighboring nodes with updated labels join the community to which the core node belongs, forming a new community. Calculate the attraction value between the unlabeled node and each community, and update the label of the unlabeled node with the label of the community with the highest attraction value; The regions corresponding to nodes in the same community are merged to achieve image segmentation.
5. The UAV hive cooperative monitoring method based on satellite remote sensing according to claim 1, characterized in that, Based on the processed image data of the target area and its characteristics, the anomaly level of the target area is determined, including: Based on the disaster type, development speed, impact range, and sensitive targets in the target area, the anomaly level of the target area is determined. The anomaly level includes yellow alert, orange alert, and red alert. A yellow alert triggers a single drone to operate, an orange alert triggers some drones to operate in formation, and a red alert triggers all drones to operate collaboratively.
6. The UAV hive cooperative monitoring method based on satellite remote sensing according to claim 1, characterized in that, Also includes: The preset drone cellular system receives the anomaly level of the target area, analyzes the anomaly type, queries the drone payload list, matches the task type with the drone sensor capabilities, and determines the target task type.
7. The UAV hive cooperative monitoring method based on satellite remote sensing according to claim 1, characterized in that, The pre-defined drone swarm response to the anomaly level of the target area employs a pre-defined path planning algorithm to formulate a drone scheduling strategy, including: For any given drone, obtain the drone's slave node. Transfer to node probability , represented as: ; in, Indicates that the drone comes from the node Transfer to node edge pheromone concentration, Represents the heuristic factor. Indicates the terrain adaptation weight. Indicates the weight of terrain factors. Indicates pheromone concentration. Represents the heuristic factor weights. Represents the set of unvisited nodes. Represents a node. Indicates the cost of constraint. The sensitivity coefficient representing the constraint penalty; Globally, the pheromone concentration is increased along the optimal path to update the pheromone concentration, expressed as: ; ; in, This indicates the pheromone concentration after a global update. This indicates the global pheromone concentration that has not been updated. Indicates the pheromone evaporation coefficient. Represents a constant. This represents the overall score of the optimal path; Within a local range, edges on non-optimal paths are weakly enhanced to update the pheromone concentration, expressed as: ; ; in, This indicates the pheromone concentration after a local update. This indicates the concentration of pheromones that have not been updated locally. Represents a uniformly distributed random number. Represents a node Transfer to node The Euclidean distance; Based on the globally updated and locally updated pheromone concentrations, update the drone's slave node. Transfer to node probability Generate the initial path for the drone.
8. The UAV hive cooperative monitoring method based on satellite remote sensing according to claim 7, characterized in that, Also includes: A cubic Bezier curve is used to smooth the initial path of the UAV. The parametric equation of the cubic Bezier curve is... Represented as: ; in, Represents a random number. , Indicates the starting point of the initial trajectory. Indicates the end point of the initial track. and These are different control points on the initial trajectory; Set curvature constraints, the curvature Represented as: ; in, Denotes the first derivative. Represents the second derivative; A curvature change rate constraint is set, which is obtained by differentiating the cubic Bezier curve and expressed as: ; in, Indicates the rate of change of curvature; The initial path of the drone is divided into several line segments according to the node order. For each line segment... The control points are solved using an optimization algorithm. and This makes the curve corresponding to each line segment... Satisfying curvature constraints and rate of change of curvature constraints; adjacent smooth segments and At the connection point The first derivative is continuous at this point.
9. The UAV hive cooperative monitoring method based on satellite remote sensing according to claim 7, characterized in that, The pre-defined drone swarm response to the anomaly level of the target area, employing a pre-defined path planning algorithm, formulates a drone scheduling strategy, and also includes: When the anomaly level of the target area is a yellow warning, a preset path planning algorithm is used to generate the initial path of a single drone. When the anomaly level of the target area is orange or red, a preset path planning algorithm is used to generate initial paths for multiple drones; for The collaborative generation of the initial path of the UAV employs a task partitioning and time offset strategy, dividing the target area into... The system consists of several overlapping sub-regions, with each drone responsible for the main path within one sub-region. Based on the distance between the drone's initial position and the sub-region, each drone is assigned a specific path. The delayed takeoff time.
10. A UAV hive cooperative monitoring system based on satellite remote sensing, characterized in that, include: The data acquisition module is used to acquire image data of a target area through satellite remote sensing; wherein, the image data of the target area includes multiple images; The data preprocessing module is used to pre-segment the multiple images, dividing each image into multiple regions; for the same image, the similarity between the regions is calculated. If the similarity is greater than a threshold, an edge is added between the two regions; otherwise, no edge is added, thus constructing an image unweighted network; a preset network label propagation algorithm is used to divide the image unweighted network into communities, merging the regions corresponding to nodes belonging to the same community to achieve image segmentation and obtain the processed image data of the target region; The level determination model is used to determine the anomaly level of the target area based on the processed image data of the target area and the characteristics of the target area. The scheduling strategy formulation module is used to formulate a scheduling strategy for drones based on a preset drone swarm response to an anomaly level in the target area and a preset path planning algorithm; at least one of the drones responds to the scheduling strategy, covers the target area, and collects video data of the target area; An emergency response module is used for the preset drone swarm to respond to video data from the target area, make decisions, and take emergency response measures.
