Water area intelligent protection monitoring system based on iron tower visual connection platform

Abstract

The invention relates to the technical field of water area intelligent protection, and discloses a water area intelligent protection monitoring system based on an iron tower visual connection platform, and the system comprises the steps: obtaining a water area image of a supervision water area at a current moment R; identifying all to-be-cleaned targets in the supervised water area according to the water area image, and predicting actual space coordinates of each to-be-cleaned target at a future moment; obtaining a geographic map of the supervised water area, and constructing an undirected graph according to the geographic map and the actual space coordinates of all the to-be-cleaned targets; all cleaning paths about the unmanned ship in the undirected graph are obtained through a depth-first search method, and an optimal cleaning path is screened out from all the cleaning paths through a pre-configured genetic algorithm; at the future moment, the unmanned ship is controlled to clean pollutants on the monitored water area according to the optimal cleaning path; according to the invention, the phenomenon of'running away 'of the unmanned ship is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of water area intelligent protection, and more specifically, to a water area intelligent protection monitoring system based on a tower video link platform. Background Art

[0002] With the continuous enhancement of environmental awareness, aquatic ecological protection has become an important part of environmental management; traditional methods of water pollution control usually rely on manual inspections, fixed sensors or planning based on preset paths. These methods have certain limitations when dealing with complex and changeable aquatic environments.

[0003] Existing path planning technologies mostly rely on static data and fixed pollutant distribution models to design cleaning routes; however, due to the highly dynamic distribution of pollutants in waters, existing path planning methods are unable to respond to changes in the location of pollutants in real time, resulting in response lags and reduced efficiency during the cleaning process; in addition, static path planning usually fails to fully consider the time-varying characteristics of pollutants in waters, which can easily cause unmanned ships to "run empty" (that is, when an unmanned ship is performing a cleaning task, it arrives at a certain location and finds that the target garbage or pollutant is no longer at that location), resulting in blank cleaning areas or repeated cleaning areas, causing waste of resources.

[0004] Therefore, there is an urgent need for a technical solution that can predict the distribution of pollutants in waters in real time and dynamically adjust the cleaning path to improve governance efficiency and resource utilization. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a water area intelligent protection monitoring system based on a tower video link platform.

[0006] To achieve the above object, the present invention provides the following technical solutions: A water area intelligent security monitoring system based on a tower video link platform, the system comprising: A data acquisition module is used to obtain a water area image of the supervised water area at the current time R, wherein the water area image is acquired by a high-altitude camera pre-installed on the tower, and R is an integer greater than zero; An information prediction module is used to identify all targets to be cleaned in the supervised waters according to the waters image, and predict the actual spatial coordinates of each target to be cleaned at a future time Q, wherein the target to be cleaned is a pollutant; An undirected graph construction module is used to obtain a geographical map of the regulated waters, and to construct an undirected graph according to the geographical map and the actual spatial coordinates of all the targets to be cleaned up, wherein the undirected graph includes a plurality of edges and a plurality of nodes, and the nodes include an unmanned boat starting node, a pollutant unloading node, and a plurality of target nodes to be cleaned up; A path planning module is used to obtain all cleaning paths about the unmanned ship in the undirected graph by a depth-first search method, and select the optimal cleaning path from all cleaning paths by using a preconfigured genetic algorithm. Each cleaning path is connected in series with the unmanned ship starting node, the pollutant unloading node and each target node to be cleaned; The cleaning control module is used to control the unmanned boat to clean pollutants in the monitored waters according to the optimal cleaning path at the future time Q.

[0007] Furthermore, the identification of all targets to be cleaned up in the regulated waters includes: Call the YOLOv8 target detection model pre-deployed in the Tower Vision Platform; Input the water area image at the current time R into the YOLOv8 target detection model to obtain a target detection result, wherein the target detection result includes K target frames, the type of each target frame, and the first pixel coordinate at the current time R; the type includes pollutants and non-pollutants, and K is an integer greater than zero; All target boxes of pollutant type are marked as targets to be cleaned, and all targets to be cleaned are obtained.

[0008] Furthermore, the method for generating the YOLOv8 target detection model is as follows: Acquire historical pollutant detection training data, and divide the historical pollutant detection training data into a pollutant detection training set and a pollutant detection test set, wherein the historical pollutant detection training data includes a plurality of water area images and corresponding annotation data; The annotation data includes a target frame, pixel coordinates and type of the target frame; Construct an initialized YOLOv8 network, use the water area images in the pollutant detection training set as input data of the YOLOv8 network, and use the labeled data in the pollutant detection training set as output data of the YOLOv8 network, train the YOLOv8 network, and obtain a trained YOLOv8 network; The trained YOLOv8 network is validated using the pollutant detection test set, and the trained YOLOv8 network with an accuracy greater than or equal to the preset accuracy threshold is output as the pre-deployed YOLOv8 target detection model.

[0009] Furthermore, the prediction of the actual spatial coordinates of each target to be cleaned at a future time Q includes: Obtain the motion data of each target to be cleaned, and call the motion trajectory prediction model pre-deployed in the tower vision platform, wherein the motion data includes the moving direction and moving speed; The motion data, the current time R and the first pixel coordinates are input into the motion trajectory prediction model to predict the second pixel coordinates of each target to be cleaned at the future time Q; The camera calibration data of the high-altitude camera is obtained, and the actual spatial coordinates of each target to be cleaned at the future time Q are determined based on the camera calibration data and the second pixel coordinates.

[0010] Furthermore, the method for generating the motion trajectory prediction model is as follows: Acquire historical motion trajectory training data, and divide the historical motion trajectory training data into a motion trajectory training set and a motion trajectory training test set, wherein the historical motion trajectory training data includes a plurality of motion trajectory feature data and the corresponding second pixel coordinates at a future time Q; Wherein, the motion trajectory feature data includes motion data, current time R and first pixel coordinates; Construct an initialized LSTM neural network, use the motion trajectory feature data in the motion trajectory training set as input data of the LSTM neural network, and use the second pixel coordinate at the future time Q in the motion trajectory training set as output data of the LSTM neural network, train the LSTM neural network, and obtain a trained LSTM neural network; The trained LSTM neural network is model verified using the motion trajectory training and test set, and the trained LSTM neural network with a value less than or equal to the preset test error threshold is output as the pre-deployed motion trajectory model.

[0011] Furthermore, the method of selecting the optimal cleaning path from all cleaning paths by using the preconfigured genetic algorithm includes: a1: Initialize the population: randomly generate an original population, which contains G individuals, each of which represents a cleaning path, and G is an integer greater than zero; a2: Fitness evaluation: Under each individual, obtain the actual spatial coordinates of all nodes in the cleaning path at the future time Q; input the distance between each adjacent node into the pre-built fitness function to calculate the fitness of each individual; a3: Selection: Use the roulette method to select two individuals with high fitness in the original population as the father and mother; a4: Crossover: Perform crossover operation on the father and mother to produce new individuals; a5: Mutation: Perform mutation operation on the new individuals to obtain S new individuals, combine the S new individuals into a new population, replace the original population with the new population, and return to step a2; a6: Repeat steps a2 to a5 above until the fitness of the individuals in the original population or the new population is greater than or equal to the preset fitness threshold, or the number of iterations is greater than or equal to the preset maximum number of iterations threshold, and output the cleaning path represented by the corresponding individual as the optimal cleaning path.

[0012] Furthermore, the calculation formula of the pre-constructed fitness function is: ; Where: For fitness, is the actual spatial coordinate of the i-th node at the future time Q, is the actual spatial coordinate of the i+1th node at the future time Q, is the total number of nodes in the cleaning path.

[0013] A water area intelligent security monitoring method based on a tower visual communication platform is implemented based on any one of the water area intelligent security monitoring systems based on a tower visual communication platform described above, and the method comprises: Obtaining a water area image of the supervised water area at the current time R, wherein the water area image is acquired by a high-altitude camera pre-installed on the tower, where R is an integer greater than zero; Identify all targets to be cleaned in the supervised waters according to the waters image, and predict the actual spatial coordinates of each target to be cleaned at a future time Q, wherein the target to be cleaned is a pollutant; Obtaining a geographic map of the regulated waters, and constructing an undirected graph according to the geographic map and the actual spatial coordinates of all targets to be cleaned up, wherein the undirected graph includes a plurality of edges and a plurality of nodes, and the nodes include an unmanned boat starting node, a pollutant unloading node, and a plurality of target nodes to be cleaned up; All cleaning paths of the unmanned ship in the undirected graph are obtained by the depth-first search method, and the optimal cleaning path is selected from all cleaning paths by using the preconfigured genetic algorithm. Each cleaning path is connected in series with the unmanned ship starting node, the pollutant unloading node and each target node to be cleaned; At the future time Q, the unmanned boat is controlled to clean up pollutants in the monitored waters according to the optimal cleaning path.

[0014] Compared with the prior art, the present invention has the following beneficial effects: The present application discloses a water area intelligent protection monitoring system based on a tower visual communication platform, comprising: obtaining a water area image of a supervised water area at the current time R; identifying all targets to be cleaned in the supervised water area according to the water area image, and predicting the actual spatial coordinates of each target to be cleaned at a future time; obtaining a geographical map of the supervised water area, and constructing an undirected graph according to the geographical map and the actual spatial coordinates of all targets to be cleaned; obtaining all cleaning paths about unmanned ships in the undirected graph by a depth-first search method, and selecting the optimal cleaning path from all cleaning paths by using a preconfigured genetic algorithm; at a future time, controlling the unmanned ship to clean pollutants in the monitored water area according to the optimal cleaning path; based on the above technical features, the present invention accurately identifies pollutants in the water area by using a YOLOv8 target detection model, and calculates the position of pollutants at a future time by a prediction model, thereby avoiding the cleaning lag caused by failure to respond to changes in the position of pollutants in a timely manner, which is beneficial to avoiding the unmanned ship from "running empty", eliminating the phenomenon of cleaning blank areas and repeated cleaning areas, and not only improving the accuracy and real-time performance of path planning, but also greatly improving the water area management efficiency and resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A module structure diagram of a water area intelligent protection monitoring system based on a tower visual connection platform provided by the present invention; Figure 2 A flow chart of a water area intelligent protection monitoring method based on a tower visual communication platform provided by the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] Example: See Figure 1 As shown, this embodiment discloses a water area intelligent security monitoring system based on a tower video link platform, the system comprising: The data acquisition module 101 is used to acquire the water area image of the supervised water area at the current time R, wherein the water area image is acquired by a high-altitude camera pre-installed on the tower, and R is an integer greater than zero; It can be understood that deploying high-altitude cameras on communication towers can obtain a wider range of water images in real time from a higher perspective, overcoming the limitations of traditional monitoring methods and providing reliable data support for subsequent path planning. At the same time, through image transmission technology (such as video streaming or timed image uploading), the water images are transmitted in real time to the tower visual platform for processing, which is conducive to the realization of intelligent protection of regulated waters and greatly improve the intelligence level of water monitoring and protection.

[0018] The information prediction module 102 is used to identify all the targets to be cleaned in the supervised water area according to the water area image, and predict the actual spatial coordinates of each target to be cleaned at a future time Q, wherein the target to be cleaned is a pollutant; In practice, the identification of all targets to be cleaned up in the regulated waters includes: Call the YOLOv8 target detection model pre-deployed in the Tower Vision Platform; What needs to be understood is that the Tower Vision Platform is a monitoring and data analysis platform built on communication towers, with real-time data collection, transmission and intelligent analysis capabilities; the platform can achieve comprehensive monitoring of water area garbage distribution, environmental status and pollution dynamics through the high-coverage monitoring network of communication tower stations and a variety of sensing devices, such as high-altitude cameras and sensor networks, and is an important technical support for intelligent protection of water areas; Specifically, the method for generating the YOLOv8 target detection model is as follows: Acquire historical pollutant detection training data, and divide the historical pollutant detection training data into a pollutant detection training set and a pollutant detection test set, wherein the historical pollutant detection training data includes a plurality of water area images and corresponding annotation data; The annotation data includes the target frame, the pixel coordinates and type of the target frame. The type includes pollutants and non-pollutants. The pollutants specifically refer to solid waste, such as plastic products floating on the water surface. It should be understood that: the water area images in the historical pollutant detection training data are collected by high-altitude cameras pre-installed on the tower in the past; the annotated data in the historical pollutant detection training data are formed by technicians after manual annotating using an annotation tool, and the annotation tool includes but is not limited to LabelImg and SuperAnnotate, etc.; Construct an initialized YOLOv8 network, use the water area images in the pollutant detection training set as input data of the YOLOv8 network, and use the labeled data in the pollutant detection training set as output data of the YOLOv8 network, train the YOLOv8 network, and obtain a trained YOLOv8 network; Use the pollutant detection test set to verify the trained YOLOv8 network model, and output the trained YOLOv8 network with a precision greater than or equal to the preset accuracy threshold as the pre-deployed YOLOv8 target detection model; Input the water area image at the current time R into the YOLOv8 target detection model to obtain a target detection result, wherein the target detection result includes K target frames, the type of each target frame, and the first pixel coordinate at the current time R; the type includes pollutants and non-pollutants, and K is an integer greater than zero; Mark all target boxes of pollutants as targets to be cleaned, and obtain all targets to be cleaned; In implementation, the prediction of the actual spatial coordinates of each to-be-cleaned target at a future time Q includes: Obtain the motion data of each target to be cleaned, and call the motion trajectory prediction model pre-deployed in the tower vision platform, wherein the motion data includes the moving direction and moving speed; It should be noted that the motion data of each target to be cleaned can be collected by motion detection sensors (such as Doppler radar, laser radar, current meter, etc.), or obtained by analyzing and calculating continuous water area images; among them, the calculation formula for the moving direction is: ; Where: is the direction of movement, expressed in radians or degrees, is the pixel coordinate of the target to be cleaned in the current frame, is the pixel coordinate of the target to be cleaned in the current frame; the calculation formula for the moving speed is: ; Where: is the moving speed of the target to be cleaned, is the inter-frame time interval of continuous frame water area images; The motion data, the current time R and the first pixel coordinates are input into the motion trajectory prediction model to predict the second pixel coordinates of each target to be cleaned at the future time Q; Specifically, the method for generating the motion trajectory prediction model is as follows: Acquire historical motion trajectory training data, and divide the historical motion trajectory training data into a motion trajectory training set and a motion trajectory training test set, wherein the historical motion trajectory training data includes a plurality of motion trajectory feature data and the corresponding second pixel coordinates at a future time Q; Wherein, the motion trajectory feature data includes motion data, current time R and first pixel coordinates; It should be understood that the motion trajectory feature data in the historical motion trajectory training data and the corresponding second pixel coordinates at the future time Q are actually collected or recorded by the technicians based on the historical data or experimental data; Construct an initialized LSTM neural network, use the motion trajectory feature data in the motion trajectory training set as input data of the LSTM neural network, and use the second pixel coordinate at the future time Q in the motion trajectory training set as output data of the LSTM neural network, train the LSTM neural network, and obtain a trained LSTM neural network; The trained LSTM neural network is model verified using the motion trajectory training and test set, and the trained LSTM neural network with a value less than or equal to the preset test error threshold is output as the pre-deployed motion trajectory model; Obtain the camera calibration data of the high-altitude camera, and determine the actual spatial coordinates of each target to be cleaned at the future time Q based on the camera calibration data and the second pixel coordinates; It should be understood that the camera calibration data provides a mapping relationship between pixel coordinates and real three-dimensional space coordinates; by obtaining the camera's internal and external parameters through the calibration process, as well as correcting image distortion, the two-dimensional pixel coordinates on the image plane can be accurately converted to real physical positions in three-dimensional space.

[0019] An undirected graph construction module 103 is used to obtain a geographical map of the regulated waters, and to construct an undirected graph according to the geographical map and the actual spatial coordinates of all the targets to be cleaned up, wherein the undirected graph includes a plurality of edges and a plurality of nodes, and the nodes include an unmanned boat starting node, a pollutant unloading node, and a plurality of target nodes to be cleaned up; It should be noted that: the geographical map of the regulated waters is pre-stored in the system database, and the geographical map marks the actual spatial coordinates of the unmanned ship starting node and the pollutant unloading node in the physical world; It should be understood that: the undirected graph is constructed based on graph theory, that is, the positions of all targets to be cleaned, the pollutant unloading positions and the starting positions of the unmanned boats are taken as nodes of the graph to obtain a node set of the undirected graph. After that, by calculating the distance between the nodes and determining whether each two nodes can be connected according to actual conditions (such as obstacles, etc.), and assigning weights to each edge, and constructing an edge set, the collection of all edges is completed. In the undirected graph, the edge represents the connection relationship between the nodes. Finally, according to the connection relationship between all nodes and edges, a topological structure is generated to complete the construction of the directed graph.

[0020] The path planning module 104 is used to obtain all cleaning paths related to the unmanned ship in the undirected graph by a depth-first search method, and select the optimal cleaning path from all cleaning paths by using a preconfigured genetic algorithm, where each cleaning path is connected in series with the unmanned ship starting node, the pollutant unloading node and each target node to be cleaned; It should be understood that the Depth First Search (DFS) method is a graph or tree traversal algorithm. Its core idea is: starting from the starting node, prioritize the current path as deep as possible, until it cannot go any deeper, then backtrack and turn to other paths to continue exploring until all nodes are visited; In implementation, the method of selecting the optimal cleaning path from all cleaning paths using a preconfigured genetic algorithm includes: a1: Initialize the population: randomly generate an original population, which contains G individuals, each of which represents a cleaning path, and G is an integer greater than zero; a2: Fitness evaluation: For each individual (i.e., each cleaning path in the original population), obtain the actual spatial coordinates of all nodes in the cleaning path at the future time Q; input the distance between each adjacent node into the pre-built fitness function to calculate the fitness of each individual; It should be understood that the actual spatial coordinates of all nodes in the cleaning path at the future time Q, including the unmanned ship starting node, the pollutant unloading node and each target node to be cleaned, are different from the actual spatial coordinates of each target node to be cleaned in that the actual spatial coordinates of the unmanned ship starting node and the pollutant unloading node are fixed positions. Therefore, the actual spatial coordinates of the unmanned ship starting node and the pollutant unloading node at the future time Q are the same at any time; The calculation formula of the pre-constructed fitness function is: ; Where: For fitness, is the actual spatial coordinate of the i-th node at the future time Q, is the actual spatial coordinate of the i+1th node at the future time Q, is the total number of nodes in the cleaning path; a3: Selection: Use the roulette method to select two individuals with high fitness in the original population as the father and mother; The roulette wheel method is a commonly used selection method, which is used in genetic algorithms to select individuals with higher fitness to enter the next generation. It simulates the process of roulette, and each individual obtains a corresponding "roulette wheel" area according to its fitness. The higher the fitness of the individual, the larger the corresponding area, and the higher the probability of being selected. a4: Crossover: Perform crossover operation on the father and mother to produce new individuals; It should be noted that the crossover operation on the paternal parent and the maternal parent is implemented based on a crossover operation, and the crossover operation includes but is not limited to one of a single-point crossover, a uniform crossover or a sequential crossover, etc.; a5: Mutation: Perform mutation operation on the new individuals to obtain S new individuals, combine the S new individuals into a new population, replace the original population with the new population, and return to step a2; In genetic algorithms, mutation operations are used to introduce genetic diversity and prevent the algorithm from falling into a local optimum. The mutation operation on new individuals is implemented by uniform mutation or Gaussian mutation. a6: Repeat steps a2 to a5 above until the fitness of the individuals in the original population or the new population is greater than or equal to the preset fitness threshold, or the number of iterations is greater than or equal to the preset maximum number of iterations threshold, and output the cleaning path represented by the corresponding individual as the optimal cleaning path; For example: Assume that the maximum number of iterations is 100 times, and record the individual with the highest fitness in the current population and its fitness value after each iteration; if it is found that the fitness value has not changed significantly in a certain generation, it is considered that the convergence condition is met, the iteration is stopped, and the cleaning path represented by the corresponding individual is output as the optimal cleaning path.

[0021] The cleaning control module 105 is used to control the unmanned boat to clean pollutants in the monitored waters according to the optimal cleaning path at a future time Q; By adopting the YOLOv8 target detection model to accurately identify pollutants in the water area and calculating the position of pollutants at future times through the prediction model, the present invention avoids the cleaning delay caused by the failure to respond to the change of the position of pollutants in time, which is conducive to preventing the unmanned boat from "running empty" and eliminating the phenomenon of cleaning blank areas and repeated cleaning areas. It not only improves the accuracy and real-time performance of path planning, but also greatly improves the efficiency of water area management and resource utilization. Example

[0022] See also Figure 2 As shown, this embodiment discloses a water area intelligent security monitoring method based on a tower video link platform, the method comprising: S201: Obtaining a water area image of the supervised water area at the current time R, wherein the water area image is acquired by a high-altitude camera pre-installed on the tower, where R is an integer greater than zero; S202: Identify all the targets to be cleaned in the supervised waters according to the waters image, and predict the actual spatial coordinates of each target to be cleaned at a future time Q, wherein the target to be cleaned is a pollutant; S203: Obtain a geographical map of the regulated waters, and construct an undirected graph according to the geographical map and the actual spatial coordinates of all the targets to be cleaned up, wherein the undirected graph includes a plurality of edges and a plurality of nodes, and the nodes include an unmanned boat starting node, a pollutant unloading node, and a plurality of target nodes to be cleaned up; S204: Obtain all cleaning paths related to the unmanned ship in the undirected graph by a depth-first search method, and select the optimal cleaning path from all cleaning paths by using a preconfigured genetic algorithm, wherein each cleaning path is connected in series with an unmanned ship starting node, a pollutant unloading node, and each target node to be cleaned; S205: At a future time Q, the unmanned boat is controlled to clean pollutants in the monitored waters according to the optimal cleaning path.

[0023] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters, weights and thresholds in the formula are set by technicians in this field according to actual conditions.

[0024] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired network or a wireless network. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media sets. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD) or a semiconductor medium. The semiconductor medium may be a solid-state hard disk.

[0025] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0026] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0027] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only one, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0028] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0029] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0030] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0031] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A water area intelligent security monitoring system based on a tower video link platform, characterized in that: The system comprises: A data acquisition module is used to obtain a water area image of the supervised water area at the current time R, wherein the water area image is acquired by a high-altitude camera pre-installed on the tower, and R is an integer greater than zero; An information prediction module is used to identify all targets to be cleaned in the supervised waters according to the waters image, and predict the actual spatial coordinates of each target to be cleaned at a future time Q, wherein the target to be cleaned is a pollutant; An undirected graph construction module is used to obtain a geographical map of the regulated waters, and to construct an undirected graph according to the geographical map and the actual spatial coordinates of all the targets to be cleaned up, wherein the undirected graph includes a plurality of edges and a plurality of nodes, and the nodes include an unmanned boat starting node, a pollutant unloading node, and a plurality of target nodes to be cleaned up; A path planning module is used to obtain all cleaning paths about the unmanned ship in the undirected graph by a depth-first search method, and select the optimal cleaning path from all cleaning paths by using a preconfigured genetic algorithm. Each cleaning path is connected in series with the unmanned ship starting node, the pollutant unloading node and each target node to be cleaned; The cleaning control module is used to control the unmanned boat to clean pollutants in the monitored waters according to the optimal cleaning path at the future time Q.

2. The water area intelligent security monitoring system based on the tower video link platform according to claim 1 is characterized in that: The identification of all targets for cleanup in regulated waters includes: Call the YOLOv8 target detection model pre-deployed in the Tower Vision Platform; Input the water area image at the current time R into the YOLOv8 target detection model to obtain a target detection result, wherein the target detection result includes K target frames, the type of each target frame, and the first pixel coordinate at the current time R; the type includes pollutants and non-pollutants, and K is an integer greater than zero; All target boxes of pollutant type are marked as targets to be cleaned, and all targets to be cleaned are obtained.

3. The water area intelligent security monitoring system based on the tower video link platform according to claim 2 is characterized in that: The method for generating the YOLOv8 target detection model is as follows: Acquire historical pollutant detection training data, and divide the historical pollutant detection training data into a pollutant detection training set and a pollutant detection test set, wherein the historical pollutant detection training data includes a plurality of water area images and corresponding annotation data; The annotation data includes a target frame, pixel coordinates and type of the target frame; Construct an initialized YOLOv8 network, use the water area images in the pollutant detection training set as input data of the YOLOv8 network, and use the labeled data in the pollutant detection training set as output data of the YOLOv8 network, train the YOLOv8 network, and obtain a trained YOLOv8 network; The trained YOLOv8 network is validated using the pollutant detection test set, and the trained YOLOv8 network with an accuracy greater than or equal to the preset accuracy threshold is output as the pre-deployed YOLOv8 target detection model.

4. The water area intelligent security monitoring system based on the tower video link platform according to claim 3 is characterized in that: The predicting of the actual spatial coordinates of each to-be-cleaned object at a future time Q includes: Obtain the motion data of each target to be cleaned, and call the motion trajectory prediction model pre-deployed in the tower vision platform, wherein the motion data includes the moving direction and moving speed; The motion data, the current time R and the first pixel coordinates are input into the motion trajectory prediction model to predict the second pixel coordinates of each target to be cleaned at the future time Q; The camera calibration data of the high-altitude camera is obtained, and the actual spatial coordinates of each target to be cleaned at the future time Q are determined based on the camera calibration data and the second pixel coordinates.

5. The water area intelligent security monitoring system based on the tower video link platform according to claim 4 is characterized in that: The method for generating the motion trajectory prediction model is as follows: Acquire historical motion trajectory training data, and divide the historical motion trajectory training data into a motion trajectory training set and a motion trajectory training test set, wherein the historical motion trajectory training data includes a plurality of motion trajectory feature data and the corresponding second pixel coordinates at a future time Q; Wherein, the motion trajectory feature data includes motion data, current time R and first pixel coordinates; Construct an initialized LSTM neural network, use the motion trajectory feature data in the motion trajectory training set as input data of the LSTM neural network, and use the second pixel coordinate at the future time Q in the motion trajectory training set as output data of the LSTM neural network, train the LSTM neural network, and obtain a trained LSTM neural network; The trained LSTM neural network is model verified using the motion trajectory training and test set, and the trained LSTM neural network with a value less than or equal to the preset test error threshold is output as the pre-deployed motion trajectory model.

6. The water area intelligent security monitoring system based on the tower video link platform according to claim 5 is characterized in that: The method of selecting the optimal cleaning path from all cleaning paths by using a preconfigured genetic algorithm includes: a1: Initialize the population: randomly generate an original population, which contains G individuals, each of which represents a cleaning path, and G is an integer greater than zero; a2: Fitness evaluation: Under each individual, obtain the actual spatial coordinates of all nodes in the cleaning path at the future time Q; input the distance between each adjacent node into the pre-built fitness function to calculate the fitness of each individual; a3: Selection: Use the roulette method to select two individuals with high fitness in the original population as the father and mother; a4: Crossover: Perform crossover operation on the father and mother to produce new individuals; a5: Mutation: Perform mutation operation on the new individuals to obtain S new individuals, combine the S new individuals into a new population, replace the original population with the new population, and return to step a2; a6: Repeat steps a2 to a5 above until the fitness of the individuals in the original population or the new population is greater than or equal to the preset fitness threshold, or the number of iterations is greater than or equal to the preset maximum number of iterations threshold, and output the cleaning path represented by the corresponding individual as the optimal cleaning path.

7. The water area intelligent security monitoring system based on the tower video link platform according to claim 6 is characterized in that: The calculation formula of the pre-constructed fitness function is: ; Where: For fitness, is the actual spatial coordinate of the i-th node at the future time Q, is the actual spatial coordinate of the i+1th node at the future time Q, is the total number of nodes in the cleaning path.

8. A water area intelligent security monitoring method based on a tower video link platform, characterized in that: The method is implemented based on the water area intelligent security monitoring system based on the tower video link platform as described in any one of claims 1 to 7, and the method comprises: Obtaining a water area image of the supervised water area at the current time R, wherein the water area image is acquired by a high-altitude camera pre-installed on the tower, where R is an integer greater than zero; Identify all targets to be cleaned in the supervised waters according to the waters image, and predict the actual spatial coordinates of each target to be cleaned at a future time Q, wherein the target to be cleaned is a pollutant; Obtaining a geographic map of the regulated waters, and constructing an undirected graph according to the geographic map and the actual spatial coordinates of all targets to be cleaned up, wherein the undirected graph includes a plurality of edges and a plurality of nodes, and the nodes include an unmanned boat starting node, a pollutant unloading node, and a plurality of target nodes to be cleaned up; All cleaning paths of the unmanned ship in the undirected graph are obtained by the depth-first search method, and the optimal cleaning path is selected from all cleaning paths by using the preconfigured genetic algorithm. Each cleaning path is connected in series with the unmanned ship starting node, the pollutant unloading node and each target node to be cleaned; At the future time Q, the unmanned boat is controlled to clean up pollutants in the monitored waters according to the optimal cleaning path.

Images