Ship Navigation Early Warning Method and System Based on Electronic Fence

Abstract

The present invention provides a ship navigation warning method and system based on an electronic fence, which relates to the field of ship navigation warning. The method includes: inputting an image into a preset image segmentation model to obtain a target ship, an obstacle, and the current water level height; inputting the current water level height into a preset water area model to obtain a dynamic electronic fence; determining the current speed according to the position information of the target ship, and determining the current course change speed according to the movement track of the target ship; inputting the current speed, the current course change speed, a first distance, and a second distance into a preset ship navigation warning model to obtain a risk assessment result, and determining a course warning strategy for the target ship according to the risk assessment result. The present invention comprehensively considers the current speed, course, and distances from obstacles and the electronic fence, improves the warning accuracy, and intelligently selects a dynamic or static electronic fence mode according to the change range of the water level height, not only considering the real-time nature of the warning but also being able to reduce the calculation amount as much as possible.

Description

Technical Field

[0001] The present invention relates to the field of ship navigation warning, and particularly to a ship navigation warning method and system based on an electronic fence. Background Art

[0002] In existing ship navigation, marine radar is generally used for obstacle avoidance and manual observation for navigation. The radar display is not intuitive enough, and the manual observation has large errors and strong limitations. Although on-board devices such as radar and Automatic Identification System (AIS) can collect a large amount of navigation information, existing systems often lack efficient information integration and intelligent analysis capabilities, and cannot comprehensively and real-time evaluate the navigation environment, thus unable to provide optimal decision-making support for crew members.

[0003] The interface of traditional ship navigation systems is complex, the operation is cumbersome, and the information display method is single, not intuitive and easy to understand, increasing the work burden of crew members and the risk of misoperation. Although ship automation technology has developed to a certain extent, it is still in the primary stage as a whole. Especially for the autonomous decision-making and response capabilities in complex navigation environments, it is still highly dependent on manual judgment and operation. In the case of being far from ports or coastlines, once an emergency occurs on a ship, the existing remote monitoring and emergency response mechanisms often have time delays and cannot quickly and effectively provide support and assistance. Traditional ship navigation mainly relies on radar and manual observation for obstacle avoidance and navigation, but this method is often not accurate and intuitive enough for defining navigation safety areas.

[0004] Existing ship navigation technologies have deficiencies to varying degrees in aspects such as information integration, human-computer interaction, automation level, remote monitoring and emergency response, and electronic fence technology. There is an urgent need to improve the safety, efficiency, and intelligence level of ship navigation. At present, there is no technical solution that can solve the above technical problems, and there is no ship navigation warning method and system based on an electronic fence. Summary of the Invention

[0005] The present invention provides a ship navigation warning method and system based on an electronic fence, integrating the electronic fence technology. The current speed, the current course change speed, the distance from obstacles, and the distance from the electronic fence are jointly used as warning reference factors, and a preset ship navigation warning model is used for multi-dimensional analysis and prediction, enhancing the safety and efficiency of water traffic.

[0006] In a first aspect, the present invention provides a ship navigation warning method based on an electronic fence, including:

[0007] For any frame image in a sequence of consecutive frame images, input the image into a preset image segmentation model to obtain the target ship, obstacles, and the current water level height output by the preset image segmentation model;

[0008] According to the current water area where the target ship is located, determine the target mode of the electronic fence. When it is determined that the target mode is the dynamic mode, input the current water level height into a preset water area model to obtain the dynamic electronic fence corresponding to the current water level height output by the preset water area model. The target mode includes the dynamic mode and the static mode;

[0009] Determine the current speed of the target ship according to the position information of the target ship in each frame image at different timestamps, determine the current course change speed of the target ship according to the movement trajectory of the target ship in each frame image at different timestamps, determine the first distance between the target ship and the obstacle, and determine the second distance between the target ship and the dynamic electronic fence;

[0010] Input the current speed, the current course change speed, the first distance, and the second distance into a preset ship navigation warning model to obtain the risk assessment result output by the preset ship navigation warning model. According to the risk assessment result, determine the course warning strategy of the target ship.

[0011] According to the ship navigation warning method based on an electronic fence provided by the present invention, before inputting the image into the preset image segmentation model, the method further includes:

[0012] When the current brightness is greater than or equal to the preset brightness, collect the sequence of consecutive frame images using a visible light camera;

[0013] When the current brightness is less than the preset brightness, collect a sequence of initial images using an infrared camera, perform super-resolution processing on the sequence of initial images to obtain a sequence of processed images, and perform noise suppression and image enhancement processing on the sequence of processed images to obtain the sequence of consecutive frame images.

[0014] According to the ship navigation warning method based on an electronic fence provided by the present invention, the step of inputting the image into the preset image segmentation model to obtain the target ship, obstacles, and the current water level height output by the preset image segmentation model includes:

[0015] Input the image into the preset image segmentation model to obtain the target ship and obstacles output by the preset image segmentation model;

[0016] Use a preset edge detection algorithm to identify the image to obtain a water surface segmentation line, and determine the current water level height according to the water surface segmentation line and a preset reference line;

[0017] The preset image segmentation model is determined after being trained according to all image samples and the annotation information on each image sample, and the annotation information includes ship information and obstacle information.

[0018] According to the ship navigation warning method based on an electronic fence provided by the present invention, determining the target mode of the electronic fence according to the current water area where the target ship is located includes:

[0019] When the change range of the water level height corresponding to the current water area is less than a preset range, determining that the target mode is a static mode;

[0020] When the change range of the water level height corresponding to the current water area is greater than or equal to the preset range, determining that the target mode is a dynamic mode.

[0021] According to the ship navigation warning method based on an electronic fence provided by the present invention, after determining the target mode of the electronic fence, the method further includes:

[0022] When it is determined that the target mode is a static mode, according to the current water level height, a static electronic fence corresponding to the current water level height is determined from the corresponding relationship between the preset water level height and the preset electronic fence. The static electronic fence is in a used state within a preset time period after being determined, and is used to determine the second distance according to the target ship and the static electronic fence in each frame of image at different timestamps.

[0023] According to the ship navigation warning method based on an electronic fence provided by the present invention, before inputting the current water level height into a preset water area model to obtain the dynamic electronic fence corresponding to the current water level height output by the preset water area model, the method further includes:

[0024] Obtain all sample images at different water level heights;

[0025] For each sample image at each water level height, perform grid processing on the waterway water area in the sample image. For each grid in the sample image, determine the depth value of the grid according to the water level depth corresponding to the grid. Traverse all grids to form a grid array, and mark the electronic fence at the boundary of the grid array with a preset value. The depth value is a value greater than zero, and the preset value is a value less than or equal to zero. Traverse all sample images to obtain all sample images after grid processing;

[0026] Train a preset initial model according to all sample images after grid processing and the water level height corresponding to each sample image to obtain the preset water area model.

[0027] According to the ship navigation warning method based on an electronic fence provided by the present invention, determining the current speed of the target ship according to the position information of the target ship in each frame of image at different timestamps includes:

[0028] Converting the position information of the target ship in each frame of image into actual position coordinates by using longitude and latitude;

[0029] Determining the displacement of the target ship and the time stamp difference in at least two frames of images;

[0030] Determining the current speed of the target ship according to the displacement and the time stamp difference.

[0031] According to the ship navigation warning method based on an electronic fence provided by the present invention, determining the current course change speed of the target ship according to the movement track of the target ship in each frame of image at different timestamps includes:

[0032] Constructing a position sequence according to the position information of the target ship under a preset continuous duration;

[0033] Using the least squares method to process and fit the position sequence to obtain the movement track of the target ship;

[0034] Determining a first direction vector of the target ship at the previous position and a second direction vector of the target ship at the current position according to the movement track;

[0035] Determining the current course change speed of the target ship according to the first direction vector and the second direction vector.

[0036] According to the ship navigation warning method based on an electronic fence provided by the present invention, determining the course warning strategy of the target ship according to the risk assessment result includes:

[0037] In the case where the risk assessment result is greater than a first threshold and less than a second threshold, generating a reminder instruction for instructing the ship to return to within the route;

[0038] In the case where the risk assessment result is greater than or equal to the second threshold, generating an alarm instruction for issuing an emergency alarm and sending the alarm instruction to a preset department for processing.

[0039] In a second aspect, a ship navigation warning system based on an electronic fence is provided, including:

[0040] A first input unit for inputting any frame of image in a continuous frame of images into a preset image segmentation model to obtain the target ship, obstacles, and the current water level height output by the preset image segmentation model;

[0041] A first determination unit, which is configured to determine a target mode of an electronic fence according to a current water area where the target ship is located. When it is determined that the target mode is a dynamic mode, the current water level height is input into a preset water area model to obtain a dynamic electronic fence corresponding to the current water level height output by the preset water area model. The target mode includes a dynamic mode and a static mode.

[0042] A second determination unit, which is configured to determine a current speed of the target ship according to position information of the target ship in each frame of image at different timestamps, determine a current course change speed of the target ship according to a movement track of the target ship in each frame of image at different timestamps, determine a first distance between the target ship and the obstacle, and determine a second distance between the target ship and the dynamic electronic fence.

[0043] A second input unit, which is configured to input the current speed, the current course change speed, the first distance, and the second distance into a preset ship navigation warning model to obtain a risk assessment result output by the preset ship navigation warning model, and determine a course warning strategy of the target ship according to the risk assessment result.

[0044] Through a preset image segmentation model, the present invention can accurately segment a target ship, an obstacle, and a current water level height from continuous frame images, providing an accurate data basis for subsequent analysis and warning. According to the change range of the water level height of the current water area, a dynamic or static electronic fence mode is intelligently selected. The dynamic mode can adapt to water level changes, improving the accuracy and real-time performance of the warning; the static mode is suitable for situations where the water level changes little, reducing the calculation amount; considering multiple dimensional factors comprehensively, the current speed, course, distance from the obstacle and the dynamic / static electronic fence are input into a preset ship navigation warning model to obtain a risk assessment result. According to the assessment result, corresponding reminder or alarm instructions are generated to timely notify the ship driver or relevant departments to take corresponding measures, effectively preventing the occurrence of navigation accidents.

[0045] The present invention formulates warning strategies at different levels according to different ranges of the risk assessment result. For lower risks, reminder instructions are generated; for high risks, alarm instructions are generated and sent to a preset department for processing. Using this multi-level warning strategy can more effectively cope with different levels of navigation risks, comprehensively applying various technical means such as image processing, water area model, dynamic analysis, and risk assessment, realizing comprehensive monitoring and warning of ship navigation safety, and having high practical value and broad application prospects. Description of the Drawings

[0046] To more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0047] Figure 1 is a schematic flowchart of a ship navigation early warning method based on an electronic fence provided by the present invention;

[0048] Figure 2 is a schematic structural diagram of a ship navigation early warning system based on an electronic fence provided by the present invention;

[0049] Figure 3 is a schematic structural diagram of an electronic device provided by the present invention. Detailed implementation manners

[0050] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0051] Figure 1 is a schematic flowchart of a ship navigation early warning method based on an electronic fence provided by the present invention. The ship navigation early warning method based on an electronic fence includes:

[0052] Step 101: For any frame image in a sequence of frame images, input the image into a preset image segmentation model to obtain the target ship, obstacles, and the current water level height output by the preset image segmentation model;

[0053] Step 102: Determine the target mode of the electronic fence according to the current water area where the target ship is located. When it is determined that the target mode is the dynamic mode, input the current water level height into a preset water area model to obtain the dynamic electronic fence corresponding to the current water level height output by the preset water area model. The target mode includes the dynamic mode and the static mode;

[0054] Step 103: Determine the current speed of the target ship according to the position information of the target ship in each frame image at different timestamps, determine the current heading change speed of the target ship according to the movement trajectory of the target ship in each frame image at different timestamps, determine the first distance between the target ship and the obstacle, and determine the second distance between the target ship and the dynamic electronic fence;

[0055] Step 104: Input the current speed, the current course change speed, the first distance, and the second distance into a preset ship navigation warning model to obtain a risk assessment result output by the preset ship navigation warning model. According to the risk assessment result, determine the course warning strategy for the target ship.

[0056] In step 101, the present invention obtains consecutive frame images from a ship navigation monitoring system, selects each frame thereof as a processing object, and inputs the selected image into a pre-trained image segmentation model. The preset image segmentation model is based on deep learning techniques, such as convolutional neural networks (CNNs), and can automatically identify and segment the target ship, obstacles (such as reefs, other ships, etc.), and the water surface area in the image. Using the water surface area in the image, in combination with a preset edge detection algorithm and a reference line, calculate the current water level height.

[0057] Optionally, before inputting the image into the preset image segmentation model, the method further includes:

[0058] When the current brightness is greater than or equal to a preset brightness, collect the consecutive frame images using a visible light camera;

[0059] When the current brightness is less than the preset brightness, collect consecutive initial images using an infrared camera, process the consecutive initial images using super-resolution to obtain consecutive processed images, and perform noise suppression and image enhancement processing on the consecutive processed images to obtain the consecutive frame images.

[0060] Optionally, in the ship navigation warning method based on an electronic fence, the present invention adds adaptive processing of the image acquisition environment to ensure that high-quality image data can be obtained under different lighting conditions, thereby improving the accuracy of subsequent image segmentation and target recognition. Before starting to collect consecutive frame images, first detect the lighting brightness of the current environment, which can be achieved through a photosensitive sensor or a brightness analysis module in an image processing algorithm, and compare the detected current brightness with the preset brightness.

[0061] If the current brightness is greater than or equal to the preset brightness, continuous frame images are collected using a visible light camera, which can capture rich color and detail information when the light is sufficient; if the current brightness is less than the preset brightness, an infrared camera is used instead to collect continuous initial images. The infrared camera can image by capturing the infrared radiation emitted by an object in a low-light environment and has good night vision and penetration capabilities. More specifically, for the continuous initial images collected by the infrared camera, super-resolution processing is first performed. Super-resolution processing is a technology that improves the image resolution through algorithms, which can improve the image quality and make the image clearer. Then, noise suppression processing is performed on the images after super-resolution processing. Since there may be more noise in infrared images, noise suppression processing can reduce the impact of this noise on the image quality. Finally, image enhancement processing is performed. Image enhancement processing aims to improve the visual effect of the image, enhance characteristics such as the contrast and brightness of the image, and make the image more suitable for subsequent processing and analysis.

[0062] By selecting appropriate image acquisition methods under different lighting conditions and combining image preprocessing techniques, the present invention can significantly improve the quality of the collected images, contribute to the accuracy of subsequent image segmentation and target recognition, and enable the ship navigation warning system to work properly under various lighting conditions, including daytime, night, and foggy days, enhancing the environmental adaptability of the system and improving its reliability and stability in practical applications.

[0063] Optionally, inputting the image into a preset image segmentation model to obtain the target ship, obstacles, and the current water level height output by the preset image segmentation model includes:

[0064] Inputting the image into a preset image segmentation model to obtain the target ship and obstacles output by the preset image segmentation model;

[0065] Identifying the image using a preset edge detection algorithm to obtain a water surface segmentation line, and determining the current water level height according to the water surface segmentation line and a preset reference line;

[0066] The preset image segmentation model is determined after being trained according to all image samples and the annotation information on each image sample, and the annotation information includes ship information and obstacle information.

[0067] Optionally, the preset image segmentation model is trained through machine learning or deep learning algorithms. During the training process, a large number of image samples are used, and these samples contain features such as ships, obstacles, and water surfaces in various scenarios. For each image sample, detailed annotation work is carried out, that is, the ship information and obstacle information in the image are annotated. By continuously iterating and optimizing the training process, a preset image segmentation model that can accurately identify and segment the target ship and obstacles in the image is finally obtained.

[0068] After determining the preset image segmentation model, the image is input into the preset image segmentation model to obtain the target ship and obstacles output by the preset image segmentation model. After obtaining the image segmentation result, the entire image is further processed using a preset edge detection algorithm. The edge detection algorithm can identify the boundaries of different regions in the image, especially the boundary between the water surface and the surrounding environment. According to the detected water surface segmentation line and in combination with the preset reference line, the current water level height is calculated. The preset reference line can be a fixed or adjustable reference line used to determine the zero point or reference point of the water level height. The above calculation process may involve the conversion from pixels to actual distances and can also be corrected according to parameters such as the position, angle, and focal length of the camera.

[0069] The present invention can more accurately identify and segment the target ship and obstacles in the image. By using the method of combining the edge detection algorithm and the preset reference line to calculate the water level height, it can effectively reduce the influence of factors such as light and shadow on the calculation result. By optimizing key steps such as image segmentation, target recognition, and water level height calculation, the performance of the entire ship navigation warning system is improved.

[0070] In step 102, determining the target mode of the electronic fence according to the current water area where the target ship is located includes:

[0071] In the case where the change range of the water level height corresponding to the current water area is less than the preset range, determining the target mode as the static mode;

[0072] In the case where the change range of the water level height corresponding to the current water area is greater than or equal to the preset range, determining the target mode as the dynamic mode.

[0073] The present invention designs two target modes, including the dynamic mode and the static mode. According to the current water area where the target ship is located, the change range of the water level height in this water area is analyzed. If the change range of the water level height is less than the preset threshold, the static mode is selected; if the change range of the water level height is greater than or equal to the preset threshold, the dynamic mode is selected. In the dynamic mode, the current water level height is input into the preset water area model. The preset water area model can output the dynamic electronic fence boundary that matches the current water level height based on historical data and algorithms.

[0074] Optionally, after determining the target mode of the electronic fence, the method further includes:

[0075] When it is determined that the target mode is the static mode, according to the current water level height, a static electronic fence corresponding to the current water level height is determined from the correspondence between the preset water level height and the preset electronic fence. The static electronic fence is in a used state within a preset time period after determination, and is used to determine the second distance according to the target ship and the static electronic fence in each frame of image at different timestamps.

[0076] Optionally, the present invention adopts different usage methods for different target modes. According to the water area characteristics where the target ship is located and the current water level height, the target mode of the electronic fence is determined. If the water level change in the water area is small and meets the conditions of the static mode, the static mode is selected. After it is determined to be the static mode, the system will find the corresponding static electronic fence from the correspondence between the preset water level height and the electronic fence according to the current water level height. The correspondence between the preset water level height and the preset electronic fence may be a database, a lookup table or any form of data structure, which stores the preset static electronic fence information at different water level heights. Further, after the static electronic fence corresponding to the current water level height is determined, the static electronic fence will be in a used state within the next preset time period, that is, a new static electronic fence will not be re-determined, but the position of the target ship in each frame of image at different timestamps will be continuously monitored within the preset time period, and the second distance between the target ship and the static electronic fence will be calculated. In the static mode, since the electronic fence is preset and remains unchanged within a certain time period, the system does not need to frequently recalculate the position and shape of the electronic fence, which helps to reduce the consumption of computing resources and improve the operation efficiency of the system.

[0077] Optionally, before inputting the current water level height into the preset water area model to obtain the dynamic electronic fence corresponding to the current water level height output by the preset water area model, the method further includes:

[0078] Obtain all sample images at different water level heights;

[0079] For each sample image at each water level height, perform grid processing on the water area of the channel in the sample image. For each grid in the sample image, determine the depth value of the grid according to the water level depth corresponding to the grid. Traverse all grids to form a grid array, and mark the electronic fence at the boundary of the grid array using a preset value. The depth value is a value greater than zero, and the preset value is a value less than or equal to zero. Traverse all sample images to obtain all sample images after grid processing;

[0080] Train a preset initial model according to all sample images after grid processing and the water level height corresponding to each sample image to obtain the preset water area model.

[0081] Optionally, this embodiment discloses in detail the specific process of training the preset water area model, so that when the current water level height is given, a corresponding dynamic electronic fence can be output. Specifically, first, a sufficient number of data samples are collected to cover different water level height situations to ensure the generalization ability of the model. Each sample image is divided into grids. Usually, the size of the grid can be set according to the resolution of the image and the required accuracy. For each grid, a depth value is determined according to its position in the water area and the corresponding water depth. The depth value is a value greater than zero, indicating the water depth within the grid. All grids in the image are traversed to form a grid array.

[0082] Furthermore, a preset value used to represent non-water area or dangerous area is determined. The preset value is less than or equal to zero. At the boundary of the grid array, the positions of the electronic fence are marked according to actual needs. The grid depth values corresponding to these positions will be set to the preset value. The electronic fence can also be marked at the boundary of the grid array using the preset value. All sample images are traversed, and the above grid processing and electronic fence marking are performed on each image. The grid-processed sample images and the corresponding water level heights are used as inputs, and the marked electronic fence is used as the output target. These data are used to train the initial model. After continuously adjusting the parameters of the model and going through multiple iterative trainings, the preset water area model is obtained.

[0083] According to the actual situation of the navigation area, an electronic fence is set in the system. The electronic fence can be static, such as the boundary of the waterway, the no-navigation area, etc., or dynamic, such as a safety area dynamically adjusted according to tides and weather changes. For example, in an area where the waterway and riverbed often change but the water level changes little, a static electronic fence can be used; while in an area where the waterway and riverbed change little, but the water level often changes due to tides or rainfall, a dynamic electronic fence can be used; and the dynamic electronic fence is also divided into an intelligent dynamic electronic fence and a pattern-matching dynamic electronic fence. Their inputs are both horizontal plane cutting data. The intelligent dynamic electronic fence generates a dynamic electronic fence based on the horizontal plane data through a trained water area model; the pattern-matching dynamic electronic fence matches the electronic fence under different preset horizontal planes according to the horizontal plane data.

[0084] Optionally, for areas with relatively simple river channel conditions, it can be obtained through prior knowledge. Collect the electronic fence setting schemes at different water levels and input them into the deep learning model for training. Finally, the effect of obtaining the intelligent dynamic electronic fence range by inputting the water level line can be achieved. Its principle is similar to automatically generating frames between key frames of a video. The present invention can use an algorithm model (Region-based Convolutional Neural Networks, R-CNN). As an optional embodiment, the detailed algorithm of the intelligent dynamic electronic fence adopted by the present invention further includes: First, collect the top views of the same section of the waterway at different horizontal plane heights to form a set, sort the pictures according to the horizontal plane height, and identify and segment the water area in the images through image recognition; Then, starting from the image with the lowest horizontal plane, the waterway waters in each image can be meshed to form an array such as the following:

[0085] [5,10,15,20],

[0086] [4,8,12,16],

[0087] [3,7,11,15],

[0088] [2,6,10,14].

[0089] In the above array, the larger the number, the deeper the water and the easier it is to navigate. The numbers at the obstacles and the electronic fence can be coded as 0 or negative numbers; The horizontal plane corresponding to each image and the above-mentioned water area meshing array are used as paired data and input into the R-CNN algorithm model for training; After the model training is completed, by inputting the horizontal plane height, the algorithm model can output the passable area of the trained waterway area and the range of the intelligent dynamic electronic fence, monitor the ship position in real time, and compare it with the electronic fence. Once the ship approaches or enters the preset prohibited or restricted area, the system immediately triggers the warning mechanism.

[0090] In step 103, the present invention obtains the current speed of the ship by comparing the position information of the target ship in consecutive images at different timestamps, using longitude and latitude conversion and displacement calculation, and also obtains the current heading change speed of the target ship by fitting the movement trajectory of the target ship, and then calculates the first distance between the target ship and the obstacle, and the second distance between the target ship and the dynamic electronic fence, so as to obtain all the evaluation parameters for realizing risk assessment.

[0091] Optionally, the determining the current speed of the target ship according to the position information of the target ship in each frame of image at different timestamps includes:

[0092] Using longitude and latitude conversion to convert the position information of the target ship in each frame of image into actual position coordinates;

[0093] Determine the displacement of the target ship and the time - stamp difference in at least two frames of images;

[0094] Determine the current speed of the target ship according to the displacement and the time - stamp difference.

[0095] In an optional embodiment, since the position information of the target ship usually directly given in an image or video frame is based on image coordinates, and these coordinates cannot directly reflect the position of the target ship in the real world. Therefore, first, the position information in the image or the pixel position obtained through image recognition technology needs to be converted into actual geographical coordinates. To calculate the speed, the displacement of the target ship within a period of time is required. The present invention realizes this by comparing the positions of the target ship in at least two frames of images. First, obtain the time interval between these two frames of images, that is, the time - stamp difference, from the time - stamp information of the images; then, use the previously obtained actual position coordinates to calculate the displacement of the target ship between these two frames of images. The calculation of the displacement can be the straight - line distance; finally, calculate the current speed of the target ship by dividing the displacement by the time, that is, divide the obtained displacement by the time - stamp difference, and the average speed of the target ship within this period of time can be obtained. If the time interval is very short, this average speed can be approximately regarded as the current speed of the target ship.

[0096] Optionally, the determining the current course - change speed of the target ship according to the movement track of the target ship in each frame of image at different time stamps includes:

[0097] Construct a position sequence according to the position information of the target ship under a preset continuous duration;

[0098] Use the least - squares method to fit the position sequence to obtain the movement track of the target ship;

[0099] According to the movement track, determine the first direction vector of the target ship at the previous position and the second direction vector of the target ship at the current position;

[0100] Determine the current course - change speed of the target ship according to the first direction vector and the second direction vector.

[0101] Optionally, the present invention needs to collect the position information of the target ship within a preset continuous time period. The position information can be extracted from continuous image frames through preset image processing techniques, such as object detection algorithms, tracking algorithms, etc. Arrange all the position information in chronological order to construct a position sequence. Each position information usually includes the coordinates of the ship in the image or the converted geographical coordinates. Use the least squares method to fit the constructed position sequence. In an optional embodiment, a linear or non-linear model can be used to fit the position data of the ship, so as to obtain its motion trajectory. The fitting result will be a smooth curve describing the moving path of the target ship during this period. Once the motion trajectory is obtained, the direction vector can be determined according to the points on the trajectory.

[0102] Generally, two adjacent points on the trajectory will be selected to calculate the direction vector. The first direction vector is an angular vector pointing from the previous position to a certain position before the current position, while the second direction vector is an angular vector pointing from the current position to its subsequent predicted position. Finally, the vector angle corresponding to the first direction vector and the vector angle corresponding to the second direction vector are used to determine the current heading change speed of the target ship. The heading change amount is the difference between the angles of the two direction vectors divided by the difference between the time points corresponding to the two direction vectors obtained from the time stamps of the image or data, that is, the angle difference divided by the time difference, to obtain the current heading change speed.

[0103] In step 104, the current speed, the current heading change speed, the first distance, and the second distance are input into a preset ship navigation warning model, and a risk assessment result is output. According to the risk assessment result, a corresponding heading warning strategy is formulated. The preset ship navigation warning model uses different speeds, different heading change speeds, different first distances, and different second distances as sample data. Each sample data corresponds to a sample risk assessment result. Based on the different sample risk assessment results corresponding to different speeds, heading change speeds, first distances, and second distances, the initial warning model is trained, and finally the preset ship navigation warning model is obtained.

[0104] Optionally, determining the heading warning strategy of the target ship according to the risk assessment result includes:

[0105] In the case where the risk assessment result is greater than the first threshold and less than the second threshold, a reminder instruction is generated. The reminder instruction is used to instruct the ship to return to the route.

[0106] In the case where the risk assessment result is greater than or equal to the second threshold, an alarm instruction is generated. The alarm instruction is used to issue an emergency alarm and send the alarm instruction to a preset department for processing.

[0107] Optionally, when the risk assessment result is less than the first threshold, it indicates that the target ship is in a safe navigation state, and continuous monitoring is sufficient; when the risk assessment result is greater than the first threshold but less than the second threshold, it indicates that the target ship is in a certain risk state but has not reached the emergency level. At this time, the system generates a reminder instruction to indicate to the ship operator to pay attention to the current situation and recommends that they take appropriate measures to bring the ship back into the predetermined route to avoid potential risks; when the risk assessment result is greater than or equal to the second threshold, it indicates that the target ship faces extremely high risks and dangerous situations may have occurred or are about to occur. At this time, the system generates an alarm instruction to immediately notify the ship operator and relevant parties to take emergency measures, such as the maritime department, port management department, etc. The alarm content may include the specific location of the ship, the type of risk, recommended countermeasures, etc. After generating the reminder or alarm instruction, the system should continuously monitor the navigation state and risk changes of the target ship. As an optional embodiment, if the ship returns to the route according to the reminder instruction or the risk level decreases, the warning state can be adjusted accordingly.

[0108] Optionally, the present invention combines the results of image recognition, target segmentation, and electronic fence monitoring to intelligently analyze the navigation state of the ship, evaluate key parameters such as the ship's navigation speed, course change, distance from obstacles, and distance from the electronic fence, and determine whether there are potential safety risks. Among them, by analyzing the change in the ship's position in consecutive frames and combining the timestamp, the navigation speed of the ship is calculated; using the historical data of the ship's position, the navigation trajectory of the ship is fitted by methods such as the least squares method or Kalman filter to determine its course change; using the scale in the image or combining radar data, the actual distance between the ship and the obstacle and the actual distance between the ship and the electronic fence are calculated; the navigation speed, course change, first distance, and second distance are used as sample parameters to construct a comprehensive evaluation model of the ship's navigation state, and different risk factors are assigned different safety risk weights, such as the impact degree of factors such as excessive speed, sharp course change, and being too close to obstacles / electronic fences on the ship's safety.

[0109] Optionally, collect videos of risk behaviors, manually score the risk for its sailing speed, course change speed, and distance from obstacles / electronic fences, and input the results of the manual scoring into machine learning model training. Since this correspondence is non-linear, machine learning is used to predict risks. The model (Deep Factorization Machines, DeepFM) can be selected. Finally, the trained model can realize the function of intelligent risk scoring through inputs such as sailing speed, course change speed, and distance from obstacles / electronic fences. Based on the fused safety risk weights, machine learning is used for risk assessment to determine whether there are potential safety risks in the current sailing state. According to the analysis results, the system automatically generates sailing suggestions or issues alarms. For ships with a slight deviation from the route, the system can issue a reminder; for ships with a serious deviation or about to enter a dangerous area, an emergency alarm is issued, and relevant departments are automatically contacted for intervention. All key data and events during the ship's voyage are recorded, including image recognition results, target segmentation information, electronic fence monitoring records, alarm trigger situations, etc. A sailing report is generated based on the recorded data to provide data support for subsequent sailing analysis, safety assessment, and management decision-making.

[0110] The present invention adopts advanced image super-resolution processing technology, significantly improving the clarity of image details, making the ship target recognition and segmentation more precise, and thus achieving unprecedented accuracy in ship overloading determination, providing a solid guarantee for sailing safety; it can capture ship images in real time and immediately perform high-speed processing and analysis to ensure the discovery and reporting of ship overloading situations in the first time, effectively shortening the response time and improving the efficiency of sailing supervision; by deeply integrating image recognition and artificial intelligence technologies, it realizes a fully automated process for ship overloading determination, greatly reducing the burden of manual monitoring, while reducing the subjectivity and error rate of human judgment, and improving the objectivity and reliability of the determination results; the technical solution of the present invention is not limited by ship types, scales, and sailing environments, and can be flexibly applied to various complex and changeable sailing scenarios, showing strong versatility and adaptability; by integrating infrared camera technology, the system can work stably even at night or under bad weather conditions, realizing all-weather and uninterrupted overloading monitoring of passing ships, ensuring no dead angle in sailing safety; the system architecture is simple, and only by deploying infrared and visible light cameras, an edge computing module, and remote transmission equipment, a complete overloading determination system can be quickly built, which is not only easy to install, but also has low maintenance costs, facilitating wide promotion and use in various ports, waterways and other areas.

[0111] Through a preset image segmentation model, the present invention can accurately segment a target ship, obstacles, and the current water level height from consecutive frame images, providing an accurate data basis for subsequent analysis and early warning. According to the change range of the water level height in the current water area, it intelligently selects a dynamic or static electronic fence mode. The dynamic mode can adapt to water level changes, improving the accuracy and timeliness of early warning; the static mode is applicable to situations where the water level changes little, reducing the calculation amount. Considering multiple-dimensional factors, the current speed, course, distance from obstacles and the dynamic / static electronic fence are input into a preset ship navigation early warning model to obtain a risk assessment result. According to the assessment result, corresponding reminder or alarm instructions are generated to timely notify the ship driver or relevant departments to take corresponding measures, effectively preventing the occurrence of navigation accidents.

[0112] The present invention formulates early warning strategies at different levels according to different ranges of the risk assessment result. For relatively low risks, reminder instructions are generated; for high risks, alarm instructions are generated and sent to a preset department for processing. Using this multi-level early warning strategy can more effectively cope with different levels of navigation risks. By comprehensively applying various technical means such as image processing, water area models, dynamic analysis, and risk assessment, it realizes the comprehensive monitoring and early warning of ship navigation safety, and has high practical value and broad application prospects.

[0113] Figure 2 FIG. is a schematic structural diagram of a ship navigation early warning system based on an electronic fence provided by the present invention. The ship navigation early warning system based on an electronic fence includes a first input unit 1. The first input unit is used to input any frame image in consecutive frame images into a preset image segmentation model to obtain the target ship, obstacles, and the current water level height output by the preset image segmentation model. The working principle of the first input unit 1 can refer to the foregoing step 101 and will not be elaborated here.

[0114] The ship navigation early warning system based on an electronic fence further includes a first determination unit 2. The first determination unit is used to determine the target mode of the electronic fence according to the current water area where the target ship is located. When it is determined that the target mode is the dynamic mode, the current water level height is input into a preset water area model to obtain the dynamic electronic fence corresponding to the current water level height output by the preset water area model. The target mode includes a dynamic mode and a static mode. The working principle of the first determination unit 2 can refer to the foregoing step 102 and will not be elaborated here.

[0115] The electronic fence-based ship navigation warning system also includes a second determination unit 3, which is used to determine the current speed of the target ship according to the position information of the target ship in each frame image at different timestamps, determine the current heading change speed of the target ship according to the motion trajectory of the target ship in each frame image at different timestamps, determine the first distance between the target ship and the obstacle, and determine the second distance between the target ship and the dynamic electronic fence. The working principle of the second determination unit 3 can refer to the aforementioned step 103 and will not be repeated here.

[0116] The electronic fence-based ship navigation warning system also includes a second input unit 4, which is used to input the current speed, the current heading change speed, the first distance and the second distance into a preset ship navigation warning model to obtain a risk assessment result output by the preset ship navigation warning model, and determine the heading warning strategy of the target ship based on the risk assessment result. The working principle of the second input unit 4 can refer to the aforementioned step 104 and will not be repeated here.

[0117] The present invention proposes a ship navigation aid identification alarm system, including camera AI recognition, target positioning and electronic fence technology, which enhances the safety and efficiency of water transportation, integrates high-precision cameras for AI intelligent recognition, can capture and analyze images of ships in sailing in real time, and accurately judge the type, size and navigation status of ships through deep learning algorithms. At the same time, the system uses advanced target positioning technologies, such as Beidou positioning and radar data fusion, to achieve accurate tracking of ship positions and ensure the real-time and accuracy of information. On this basis, the system also incorporates electronic fence technology to set virtual boundaries in key waters, waterway intersections or prohibited navigation areas in advance. Once a ship enters these preset electronic fence areas, the system immediately triggers an alarm and sends an emergency notice to the ship's bridge, port management center and even relevant maritime departments, effectively preventing the occurrence of safety incidents such as ship collisions and illegal navigation. In addition, the system can also combine historical navigation data with real-time environmental information to provide ships with optimal route planning suggestions, further improving navigation efficiency and safety.

[0118] Through a preset image segmentation model, the present invention can accurately segment the target ship, obstacles, and the current water level height from consecutive frame images, providing an accurate data basis for subsequent analysis and early warning. According to the change range of the water level height in the current water area, it intelligently selects the dynamic or static electronic fence mode. The dynamic mode can adapt to water level changes and improve the accuracy and timeliness of early warning; the static mode is suitable for situations where the water level changes little, reducing the computational amount. By comprehensively considering multi-dimensional factors, the current speed, heading, and distances from obstacles and the dynamic / static electronic fence are input into a preset ship navigation early warning model to obtain a risk assessment result. According to the assessment result, corresponding reminder or alarm instructions are generated to timely notify the ship driver or relevant departments to take corresponding measures, effectively preventing the occurrence of navigation accidents;

[0119] According to different ranges of the risk assessment result, the present invention formulates early warning strategies at different levels. For lower risks, reminder instructions are generated; for high risks, alarm instructions are generated and sent to a preset department for processing. Using this multi-level early warning strategy can more effectively cope with different levels of navigation risks. By comprehensively applying various technical means such as image processing, water area models, dynamic analysis, and risk assessment, it realizes the comprehensive monitoring and early warning of ship navigation safety, and has high practical value and broad application prospects.

[0120] Figure 3 It is a schematic structural diagram of the electronic device provided by the present invention. As Figure 3As shown in the figure, the electronic device may include: a processor 310, a communication interface 320, a memory 330, and a communication bus 340. Among them, the processor 310, the communication interface 320, and the memory 330 complete communication with each other through the communication bus 340. The processor 310 may call the logical instructions in the memory 330 to execute a ship navigation warning method based on an electronic fence. The method includes: for any frame image in a series of consecutive frame images, input the image into a preset image segmentation model to obtain the target ship, obstacles, and the current water level height output by the preset image segmentation model; according to the current water area where the target ship is located, determine the target mode of the electronic fence. When it is determined that the target mode is the dynamic mode, input the current water level height into a preset water area model to obtain the dynamic electronic fence corresponding to the current water level height output by the preset water area model. The target mode includes the dynamic mode and the static mode; determine the current speed of the target ship according to the position information of the target ship in each frame image at different timestamps, determine the current heading change speed of the target ship according to the movement trajectory of the target ship in each frame image at different timestamps, determine the first distance between the target ship and the obstacles, and determine the second distance between the target ship and the dynamic electronic fence; input the current speed, the current heading change speed, the first distance, and the second distance into a preset ship navigation warning model to obtain the risk assessment result output by the preset ship navigation warning model. According to the risk assessment result, determine the heading warning strategy of the target ship.

[0121] In addition, when the logical instructions in the above-mentioned memory 330 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0122] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute a ship navigation warning method based on an electronic fence provided by each of the above methods. The method includes: for any frame image in a continuous frame of images, input the image into a preset image segmentation model to obtain the target ship, obstacles, and the current water level height output by the preset image segmentation model; determine the target mode of the electronic fence according to the current water area where the target ship is located. When it is determined that the target mode is the dynamic mode, input the current water level height into a preset water area model to obtain the dynamic electronic fence corresponding to the current water level height output by the preset water area model. The target mode includes a dynamic mode and a static mode; determine the current speed of the target ship according to the position information of the target ship in each frame of image at different timestamps, determine the current heading change speed of the target ship according to the movement trajectory of the target ship in each frame of image at different timestamps, determine the first distance between the target ship and the obstacle, and determine the second distance between the target ship and the dynamic electronic fence; input the current speed, the current heading change speed, the first distance, and the second distance into a preset ship navigation warning model to obtain a risk assessment result output by the preset ship navigation warning model. According to the risk assessment result, determine the heading warning strategy of the target ship.

[0123] In another aspect, the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the ship navigation early warning method based on an electronic fence provided by the above-mentioned various methods. The method includes: for any frame image in a series of consecutive frame images, input the image into a preset image segmentation model to obtain the target ship, obstacles, and the current water level height output by the preset image segmentation model; according to the current water area where the target ship is located, determine the target mode of the electronic fence. When it is determined that the target mode is the dynamic mode, input the current water level height into a preset water area model to obtain the dynamic electronic fence corresponding to the current water level height output by the preset water area model. The target mode includes the dynamic mode and the static mode; determine the current speed of the target ship according to the position information of the target ship in each frame image at different timestamps, determine the current heading change speed of the target ship according to the movement trajectory of the target ship in each frame image at different timestamps, determine the first distance between the target ship and the obstacles, and determine the second distance between the target ship and the dynamic electronic fence; input the current speed, the current heading change speed, the first distance, and the second distance into a preset ship navigation early warning model to obtain the risk assessment result output by the preset ship navigation early warning model. According to the risk assessment result, determine the heading early warning strategy of the target ship.

[0124] The device embodiments described above are merely illustrative. 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 may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0125] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disc, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A ship navigation warning method based on electronic fence, characterized in that: include: For any frame image in the continuous frame images, the image is input into a preset image segmentation model to obtain the target ship, obstacle and current water level output by the preset image segmentation model; Determine the target mode of the electronic fence according to the current waters where the target ship is located. When it is determined that the target mode is a dynamic mode, input the current water level height into a preset waters model to obtain a dynamic electronic fence corresponding to the current water level height output by the preset waters model. The target mode includes a dynamic mode and a static mode. Determine the current speed of the target ship according to the position information of the target ship in each frame image at different timestamps, determine the current course change speed of the target ship according to the motion trajectory of the target ship in each frame image at different timestamps, determine the first distance between the target ship and the obstacle, and determine the second distance between the target ship and the dynamic electronic fence; Inputting the current speed, the current heading change speed, the first distance, and the second distance into a preset ship navigation warning model, obtaining a risk assessment result output by the preset ship navigation warning model, and determining a heading warning strategy for the target ship according to the risk assessment result; Determining the target mode of the electronic fence according to the current waters where the target ship is located includes: When the change amplitude of the water level height corresponding to the current water area is less than a preset amplitude, determining that the target mode is a static mode; When the change amplitude of the water level height corresponding to the current water area is greater than or equal to the preset amplitude, determining that the target mode is a dynamic mode; After determining the target mode of the electronic fence, the method further includes: In the case where it is determined that the target mode is a static mode, according to the current water level, from the correspondence between the preset water level and the preset electronic fence, a static electronic fence corresponding to the current water level is determined, the static electronic fence is in use within a preset time period after the determination, and according to the target ship and the static electronic fence in each frame image at different timestamps, a second distance between the target ship and the static electronic fence is determined; Before inputting the current water level to the preset water area model to obtain the dynamic electronic fence corresponding to the current water level output by the preset water area model, the method further includes: Get all sample images at different water level heights; For each sample image at each water level, gridding is performed on the waterway waters in the sample image, and for each grid in the sample image, a depth value of the grid is determined according to the water level depth corresponding to the grid, all grids are traversed to form a grid array, and an electronic fence is marked at the boundary of the grid array using a preset value, wherein the depth value is a value greater than zero, and the preset value is a value less than or equal to zero, and all sample images are traversed to obtain all sample images after gridding; The preset initial model is trained according to all sample images after gridding processing and the water level height corresponding to each sample image to obtain the preset water area model.

2. The ship navigation warning method based on electronic fence according to claim 1 is characterized in that: Before inputting the image into a preset image segmentation model, the method further includes: When the current brightness is greater than or equal to the preset brightness, using a visible light camera to collect the continuous frame images; When the current brightness is less than the preset brightness, continuous initial images are collected using an infrared camera, the continuous initial images are processed using super-resolution to obtain continuous processed images, and noise suppression and image enhancement are performed on the continuous processed images to obtain the continuous frame images.

3. The ship navigation warning method based on electronic fence according to claim 1 is characterized in that: The step of inputting the image into a preset image segmentation model to obtain the target ship, obstacle and current water level output by the preset image segmentation model includes: Inputting the image into a preset image segmentation model to obtain target ships and obstacles output by the preset image segmentation model; Using a preset edge detection algorithm to identify the image, obtain a water surface segmentation line, and determine the current water level height according to the water surface segmentation line and a preset reference line; The preset image segmentation model is determined after training based on all image samples and the annotation information on each image sample, wherein the annotation information includes ship information and obstacle information.

4. The ship navigation warning method based on electronic fence according to claim 1 is characterized in that: The determining the current speed of the target ship according to the position information of the target ship in each frame of image at different time stamps includes: The position information of the target ship in each frame of the image is converted into actual position coordinates using longitude and latitude; Determining the displacement and time stamp difference of the target ship in at least two frames of images; The current speed of the target ship is determined according to the displacement and the timestamp difference.

5. The ship navigation warning method based on electronic fence according to claim 1 is characterized in that: The step of determining the current course change speed of the target ship according to the motion trajectory of the target ship in each frame of the image at different time stamps includes: Constructing a position sequence according to the position information of the target ship under a preset continuous time length; Fitting the position sequence using the least squares method to obtain the motion trajectory of the target ship; Determine, according to the motion trajectory, a first direction vector of the target ship at a previous position and a second direction vector of the target ship at a current position; The current course change speed of the target ship is determined according to the first direction vector and the second direction vector.

6. The ship navigation warning method based on electronic fence according to claim 1 is characterized in that: Determining the heading warning strategy of the target ship according to the risk assessment result includes: When the risk assessment result is greater than a first threshold value and less than a second threshold value, generating a reminder instruction, wherein the reminder instruction is used to instruct the ship to return to the route; When the risk assessment result is greater than or equal to the second threshold, an alarm instruction is generated, the alarm instruction is used to issue an emergency alarm, and the alarm instruction is sent to a preset department for processing.

7. A ship navigation warning system based on electronic fence, characterized in that: include: A first input unit, the first input unit is used to input any frame image in the continuous frame image into a preset image segmentation model to obtain the target ship, obstacle and current water level output by the preset image segmentation model; A first determining unit, the first determining unit is used to determine a target mode of the electronic fence according to the current waters where the target ship is located, and when it is determined that the target mode is a dynamic mode, input the current water level height into a preset waters model to obtain a dynamic electronic fence corresponding to the current water level height output by the preset waters model, wherein the target mode includes a dynamic mode and a static mode; a second determination unit, the second determination unit being used to determine a current speed of the target ship according to the position information of the target ship in each frame image at different timestamps, determine a current course change speed of the target ship according to the motion trajectory of the target ship in each frame image at different timestamps, determine a first distance between the target ship and the obstacle, and determine a second distance between the target ship and the dynamic electronic fence; a second input unit, the second input unit being used to input the current speed, the current heading change speed, the first distance and the second distance into a preset ship navigation warning model, to obtain a risk assessment result output by the preset ship navigation warning model, and to determine a heading warning strategy for the target ship according to the risk assessment result; Determining the target mode of the electronic fence according to the current waters where the target ship is located includes: When the change amplitude of the water level height corresponding to the current water area is less than a preset amplitude, determining that the target mode is a static mode; When the change amplitude of the water level height corresponding to the current water area is greater than or equal to the preset amplitude, determining that the target mode is a dynamic mode; After determining the target mode of the electronic fence, it also includes: In the case where it is determined that the target mode is a static mode, according to the current water level, from the correspondence between the preset water level and the preset electronic fence, a static electronic fence corresponding to the current water level is determined, the static electronic fence is in use within a preset time period after the determination, and according to the target ship and the static electronic fence in each frame image at different timestamps, a second distance between the target ship and the static electronic fence is determined; Before inputting the current water level to the preset water area model to obtain the dynamic electronic fence corresponding to the current water level output by the preset water area model, the method further includes: Get all sample images at different water level heights; For each sample image at each water level, gridding is performed on the waterway waters in the sample image, and for each grid in the sample image, a depth value of the grid is determined according to the water level depth corresponding to the grid, all grids are traversed to form a grid array, and an electronic fence is marked at the boundary of the grid array using a preset value, wherein the depth value is a value greater than zero, and the preset value is a value less than or equal to zero, and all sample images are traversed to obtain all sample images after gridding; The preset initial model is trained according to all sample images after gridding processing and the water level height corresponding to each sample image to obtain the preset water area model.

Images