A method for optimizing the fishing position of larvae and juveniles

Abstract

The invention relates to the technical field of fry and juvenile fishing position optimization, and discloses a fry and juvenile fishing position optimization processing method. A fishing area is demarcated in a natural water area to be studied, a fishing area map based on a GIS system is generated according to surveying and mapping data, and according to the characteristic that fry and juvenile fish generally prefer the area around aquatic plants, an aquatic plant area in the demarcated fishing water area is identified by shooting with an unmanned aerial vehicle and based on an aquatic plant image detection model, static fishing is carried out in the aquatic plant area by arranging fixed stake net cages, and in order to carry out fishing more fully, a fishing boat is used to fish in the part outside the aquatic plant area in the demarcated fishing area, so that the fishing work of fry and juvenile fish in the demarcated fishing area can be completed more efficiently and more fully, and it is helpful to more accurately evaluate the development prospect of local fish resources.

Description

Technical Field

[0001] The invention belongs to the technical field of fishing position optimization, and in particular relates to a method for optimizing the fishing position of fry and juvenile fish. Background Art

[0002] In fishery research, the health and reproduction of fish populations can be evaluated by catching fish fry, which helps to evaluate the development prospects of fish resources. At present, there are two main ways to catch fish fry. One is static fishing with fixed stakes and cages. For example, Chinese patent (CN116831091A) discloses a fish fry collection device for the main stream of the Yellow River, which is fixed stake fishing. The other is dynamic fishing using a fishing boat. For example, Chinese patent (CN117775202B) discloses an automatic fishing boat for fishery resource surveys. There is no method specifically optimized for fry fry fishing, such as how to better perform static fixed stake fishing of fish fry based on the conventional characteristics of fish fry, how to better drive the fishing boat to catch fish fry, or how to cooperate with the fixed stakes and the fishing boat to catch fish fry. Summary of the invention

[0003] In order to overcome the technical problems described in the above-mentioned background technology, the present invention provides a method for optimizing the fishing position of fry and juvenile fish. The fishing area is demarcated in the natural waters to be studied, and a fishing area map based on the GIS system is generated according to the surveying and mapping data. According to the characteristics that fry and juvenile fish generally prefer the area around aquatic plants, the aquatic plant area in the demarcated fishing waters is identified by taking pictures by an unmanned aerial vehicle and based on an aquatic plant image detection model. Static fishing is carried out in the aquatic plant area by arranging fixed stake cages. At the same time, in order to carry out fishing more fully, fishing boats are used to fish in the part outside the aquatic plant area in the demarcated fishing area. The fishing work of fry and juvenile fish in the demarcated fishing area can be completed more efficiently and more fully, which is helpful to more accurately evaluate the development prospects of local fish resources.

[0004] The technical solution of the present invention is: a method for optimizing the fishing position of fry and juvenile fish, comprising the following steps:

[0005] S1. Delineate the fishing area for fry and juvenile fish in natural waters, obtain the surveying and mapping data of the fishing area for fry and juvenile fish from the natural resources management department, and import it into the GIS system;

[0006] S2. In the GIS system, a serpentine UAV flight path is generated using a grid method, the endpoints of each line segment in the flight path are converted into corresponding satellite positioning coordinates, and the UAV flight trajectory coordinate data based on the satellite positioning coordinates are output;

[0007] S3, drone flight track coordinate data are imported into drone control system, drone is according to predetermined flight altitude h and flies with flight speed v, drone camera takes pictures with shooting frequency f simultaneously, and satellite positioning coordinates and the heading angle of drone are written in the metadata of picture when shooting, obtain a picture data set that delimits fry and juvenile fish catching area;

[0008] S4, using the pictures in the picture data set demarcating the fry and juvenile fish fishing area, successively detecting pictures with water plants through the water plant picture detection model, obtaining all detected water plant pictures, and the water plant area and the minimum circumscribed rectangle coordinates of the water plant closed area in each detected water plant picture, and extracting the satellite positioning coordinates of the corresponding picture center read out from the EXIF ​​metadata in the picture, converting the diagonal point coordinates of the minimum circumscribed rectangle of the water plant closed area into satellite positioning coordinates, and outputting the water plant detection data containing the area field of the water plant closed area in the picture and the satellite positioning coordinate field of the diagonal points of the minimum circumscribed rectangle of the water plant closed area;

[0009] S5. Import the waterweed detection data into the GIS system and highlight it, generate a waterweed distribution map, and count the area of ​​each continuous waterweed area in the GIS system , according to the coverage capacity of the cages, calculate the number of cages to be placed in each continuous aquatic plant area ;

[0010] S6. The number of cages placed in each continuous aquatic plant area , in the GIS system, each continuous aquatic plant area is divided into The output contains the satellite positioning coordinates of the cage after the copy. The satellite positioning coordinate data of each cage placement is recorded as the attribute of the corresponding continuous aquatic plant area, and the satellite positioning coordinates of all cage placements are displayed on the aquatic plant distribution map in the form of highlighted dots;

[0011] S7. The satellite positioning coordinate data of the cage placement output from the GIS system is imported into a portable satellite positioning device, and the corresponding geographical location is found in the designated fry and juvenile fish catching area to arrange the cage.

[0012] Furthermore, in step S3, the drone used is a drone with a fixed-altitude flight capability.

[0013] Further, in step S3, the formula for the shooting frequency f of the drone camera with the lens vertically facing is:

[0014] ,

[0015] Among them, f is the shooting frequency of the drone camera, v is the flight speed of the drone, is the camera's field of view. The larger the camera's field of view, the wider the coverage of a single shot. h is the drone's flight altitude, which determines the size of the shooting area. The higher the flight altitude, the larger the shooting field of view. o is the overlap, which is used to ensure that adjacent photos overlap to avoid blind spots.

[0016] Further, in step S5, the number of cages placed in the continuous aquatic plant area is The calculation formula is: ,in represents the area of ​​the i-th continuous aquatic plant area, s represents the coverage area of ​​the cage, This means rounding up to ensure that the entire area is covered.

[0017] Furthermore, the method further comprises the following steps:

[0018] S8, in the GIS system, the aquatic plant area is set as an obstacle, an adaptive segmentation algorithm is used to plan the running track of the fry fishing boat, and a fishing route with satellite positioning nodes is generated, and satellite positioning coordinate data of the fishing route nodes are generated;

[0019] S9, importing the satellite positioning coordinate data of the fishing route nodes into the controller of the fry fishing vessel, starting the fry fishing vessel according to the fishing route in the designated fry fishing area, and performing dynamic fishing in the designated fry fishing area.

[0020] Furthermore, the operation of generating a fishing route in step S8 includes the following steps:

[0021] S801. According to the boundaries of the fishing area for fry and juvenile fish Generate a regular grid structure , ,in Represents the index of the grid cell. The initialized grid size is the same as the fishing width w;

[0022] S802: For each grid unit , based on the demarcation of the boundaries of the larvae and juvenile fishing areas and the boundary of the aquatic plant area The relationship is adaptively adjusted if a grid cell Contains both water and shallow water boundaries, or is located at the edge of an irregular area, that is, if and , then refine the grid cell The refinement process uses the quartering method to subdivide the grid cells into smaller grid cells to adapt to complex areas, and in order to avoid over-refinement, the maximum subdivision depth is set , if the refinement depth exceeds , then no further refinement;

[0023] S803, for the grid structure after adaptive subdivision , parallel routes are generated in each grid unit, and the route interval is the fishing width w, but in the refined grid area, the direction and distribution of the route are dynamically adjusted according to the shape of the grid, that is, in the regular grid area, the route is a parallel line, and in the irregular grid area, the direction of the route is adaptively adjusted according to the grid shape to ensure the coverage, so that the fishing route calculation formula is: , where i is the index number of the route, is the position coordinate of the i-th fishing route, is the position coordinate of the starting fishing route, w is the fishing width;

[0024] S804, Use the boundary of the aquatic plant area For the generated route segment Perform Boolean operations to remove overlapping parts and obtain effective fishing route segments , that is ;

[0025] S805, using the nearest neighbor algorithm to optimize the effective fishing route Perform global optimization to calculate the shortest path between routes;

[0026] S806, dynamically adjust the speed of the fishing boat according to the shape of the water area and the complexity of the route , maintain a higher speed in regular areas and slow down in irregular areas to increase fishing accuracy, The formula is: ,in Indicates the location of the larvae fishing boat in the grid cell The speed inside, is the maximum fishing speed of the fry fishing vessel, is the complexity adjustment factor used to control the impact of complexity on speed, is the i,jth grid cell The higher the complexity, the more irregular the grid area is or the more obstacles there are, and the fishing boat needs to slow down.

[0027] Furthermore, in step S4, the detection process of the waterweed image detection model includes the following steps:

[0028] S4101, the image to be detected Input the trained model , get the prediction result R, including the bounding box of the aquatic plant area and its confidence, and filter out valid detections according to the confidence threshold;

[0029] S4102, according to the upper left corner coordinates of the identified aquatic plant area boundary box ( , ) and the lower right corner coordinates ( , ), calculate the pixel area of ​​the water grass area in the image , ;

[0030] S4103, based on the ground sampling distance GSD and the pixel area of ​​the aquatic plant area in the image , calculate the actual area corresponding to the aquatic plant area in the figure , ;

[0031] S4104, according to the longitude and latitude coordinates of the satellite positioning coordinates corresponding to the center of the picture read from the EXIF ​​metadata in the picture ( , ), heading angle and pictures Size , and calculate the coordinates of the upper left corner of the rectangular box of the aquatic plant area in the figure by converting pixels into actual longitude and latitude coordinates ( , ) and the lower right corner coordinates ( , ) corresponds to the satellite positioning longitude and latitude coordinates .

[0032] Furthermore, in step S4103, the ground sampling distance GSD is calculated as: , where h is the flight altitude of the drone, is the camera's field of view. The larger the camera's field of view, the wider the coverage of a single shot. is the number of pixels of the drone camera image sensor in the corresponding length direction.

[0033] Furthermore, in step S4104, the calculation of the formula for converting pixels into actual longitude and latitude coordinates includes the following steps:

[0034] S1041. For any point in the figure ( , ), calculate the relative to the image center ( , ) and , the formula is and ;

[0035] S1042, the offset and Convert to geographic distance offset respectively and , and the formula is and ;

[0036] S1043, Use the rotation matrix Adjusting geographic distance offset and , and the formula is;

[0037] ;

[0038] S1044, adding the adjusted geographic offset to the longitude and latitude coordinates of the satellite positioning coordinates of the center of the image ( , ) to get the actual longitude coordinate of the pixel point and latitude coordinates , and the formula is:

[0039] and , where R is the radius of the earth, which is 6371000m.

[0040] The present invention has the following beneficial effects due to the adoption of the above technology.

[0041] 1. The present invention demarcates fishing areas in natural waters to be studied, generates fishing area maps based on a GIS system according to surveying and mapping data, and completes identification of aquatic plant areas in the demarcated fishing waters based on aquatic plant image detection model by taking pictures with an unmanned aerial vehicle according to the characteristics that fry and juvenile fish generally prefer areas around aquatic plants. Static fishing is carried out in the aquatic plant area by arranging fixed stake net cages. At the same time, in order to fish more fully, fishing boats are used to fish in the part outside the aquatic plant area in the demarcated fishing area. The fishing work of fry and juvenile fish in the demarcated fishing area can be completed more efficiently and more fully, which is helpful to more accurately evaluate the development prospects of local fish resources.

[0042] 2. By introducing a combination of drone photography, satellite positioning and GIS systems, the present invention can more accurately delineate the fishing area for fry and juvenile fish compared to manually randomly selecting the fishing location for fry and juvenile fish, and realizes automatic detection of aquatic plant areas and optimal calculation of cage placement positions, thereby reducing human errors, ensuring that the fishing cages are accurately placed in suitable locations where fish gather, and improving fishing efficiency and accuracy.

[0043] 3. The present invention generates aquatic plant distribution maps and highlighted cage fishing locations through the GIS system, which can help fishery management personnel more intuitively grasp the spatial distribution of aquatic plants, so as to effectively arrange fishing cages and better manage water resources. This visual management method facilitates subsequent data analysis and decision-making.

[0044] 4. The present invention reduces interference with the ecological environment such as aquatic plants by detecting the distribution area of ​​aquatic plants in advance and automatically calculating the placement position of the cages, and can achieve more accurate demarcation of fishing areas and more reasonable allocation of cages. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a flow chart of the first embodiment of the present invention.

[0046] Figure 2 It is a flow chart of the second embodiment of the present invention.

[0047] Figure 3 It is a schematic diagram of the process of generating a fishing route in step S8 of the second embodiment of the present invention.

[0048] Figure 4 It is a schematic diagram of the use process of the aquatic plant image detection model of the present invention.

[0049] Figure 5 It is a schematic diagram of the process of converting pixels in a waterweed picture into actual longitude and latitude coordinates of the present invention. DETAILED DESCRIPTION

[0050] Embodiment 1: Figure 1 As shown, the present invention provides a method for optimizing the fishing position of fry and juvenile fish, comprising the following steps:

[0051] S1. Delineate the fishing area for fry and juvenile fish in natural waters. The delineated fishing area for fry and juvenile fish may be irregular in shape. Obtain the surveying and mapping data for delineating the fishing area for fry and juvenile fish from the natural resources management department. Usually, it is a vector format file, such as Shapefile or GeoJSON format. Import it into the GIS system. The main step is to open the GIS software and create a new project. Select the "Import Data" function, import the surveying and mapping data file obtained from the management department into the project, and generate the corresponding map. Pay attention to check whether the imported data contains the correct projection coordinate system, and perform necessary coordinate conversion, and generate a regional boundary map in the GIS interface to ensure that the fishing area accurately reflects the irregular shape of the actual waters;

[0052] S2. In the GIS system, a serpentine UAV flight path is generated using a grid method, the endpoints of each line segment in the flight path are converted into corresponding satellite positioning coordinates, and the UAV flight trajectory coordinate data based on the satellite positioning coordinates are output;

[0053] The specific operation steps are as follows: S201, select the grid generation tool, and set the appropriate grid spacing according to the boundary and area of ​​the fishing area; S202, the generated grid divides the fishing area into several small areas, and uses the serpentine path planning algorithm to generate the UAV flight path in each grid; S203, ensure that the path covers the fishing area and the turning points of the path meet the flight control requirements; S204, convert the endpoint coordinates of each line segment in the path into corresponding satellite positioning coordinates, such as the longitude and latitude coordinates in the WGS84 coordinate system; S205, export the satellite positioning coordinate data of the UAV flight trajectory, and ensure that the output data format is compatible with the navigation system of the UAV, such as the CSV format;

[0054] S3, drone flight track coordinate data are imported into drone control system, drone is according to predetermined flight altitude h and flies with flight speed v, drone camera takes pictures with shooting frequency f simultaneously, and satellite positioning coordinates and the heading angle of drone are written in the metadata of picture when shooting, obtain a picture data set that delimits fry and juvenile fish catching area;

[0055] S4, using the pictures in the picture data set demarcating the fry and juvenile fish fishing area, successively detecting pictures with water plants through the water plant picture detection model, obtaining all detected water plant pictures, and the water plant area and the minimum circumscribed rectangle coordinates of the water plant closed area in each detected water plant picture, and extracting the satellite positioning coordinates of the corresponding picture center read out from the EXIF ​​metadata in the picture, converting the diagonal point coordinates of the minimum circumscribed rectangle of the water plant closed area into satellite positioning coordinates, and outputting the water plant detection data containing the area field of the water plant closed area in the picture and the satellite positioning coordinate field of the diagonal points of the minimum circumscribed rectangle of the water plant closed area;

[0056] S5. Import the waterweed detection data into the GIS system and highlight it, generate a waterweed distribution map, and count the area of ​​each continuous waterweed area in the GIS system , according to the coverage capacity of the cages, by dividing and rounding up, calculate the number of cages required for each continuous aquatic plant area ;

[0057] S6. The number of cages placed in each continuous aquatic plant area , in the GIS system, each continuous aquatic plant area is divided into The output contains the satellite positioning coordinates of the cage after the copy. The satellite positioning coordinate data of each cage placement is recorded as the attribute of the corresponding continuous aquatic plant area, and the satellite positioning coordinates of all cage placements are displayed on the aquatic plant distribution map in the form of highlighted dots;

[0058] S7. Export the cage placement satellite positioning coordinate data output from the GIS system to a file in a data format compatible with a portable satellite positioning device, such as GPX, KML or CSV format. Then transfer the exported file to the portable satellite positioning device via a data cable or wireless transmission. Carry the portable satellite positioning device to the designated fry and juvenile fish catching area, and find the precise geographical location of each cage placement point based on the navigation function provided by the device. Arrange staked cages at each coordinate point to ensure that the cage layout meets the requirements of the fry and juvenile fish catching plan.

[0059] In step S3, the drone used can accurately fly at a fixed altitude to ensure that the angle and resolution of the captured images are consistent, which is suitable for subsequent aquatic plant detection and coordinate calculation. The drone camera needs to have the function of recording the heading angle. The camera system receives the heading angle data transmitted by the drone system and writes the heading angle together with the satellite positioning information into the EXIF ​​metadata of the image.

[0060] In step S3, the formula for the shooting frequency f of the drone camera with the lens vertically facing is:

[0061] ,

[0062] Among them, f is the shooting frequency of the drone camera, v is the flight speed of the drone, is the camera's field of view. The larger the camera's field of view, the wider the coverage of a single shot. h is the drone's flight altitude, which determines the size of the shooting area. The higher the flight altitude, the larger the shooting field of view. o is the overlap, which is used to ensure that adjacent photos overlap to avoid blind spots.

[0063] In step S4, the generation of the waterweed image detection model includes the following steps:

[0064] S4001. Collect and organize a dataset D of images containing and not containing aquatic plants. ,in represents the i-th picture, Indicates the corresponding label. If the image contains water plants, is 1, if not included is 0, M is the image index in the dataset D;

[0065] S4002, for the picture containing aquatic plants, use annotation software to mark the boundary box or segmented area of ​​the aquatic plants to obtain coordinate information of the aquatic plant area;

[0066] S4003, adjusting the size of all images in the data set D to a uniform size, normalizing the pixel values, and dividing the data set D into a training set and a validation set according to a quantity ratio of 8:2;

[0067] S4004. Use a deep learning model suitable for target detection and define the model structure , where θ is the model parameter;

[0068] S4005. Define the objective function , optimized model parameters The formula is:

[0069] ,

[0070] in is the number of training samples, is the preprocessed training sample, is the true label, indicating whether the image contains water plants and the rectangular bounding box of water plants. is the loss function, and the loss function L calculates the prediction result of the model With the true label The difference between Represents the search for a loss function The parameter θ that reaches the minimum value;

[0071] S4006. Run the model code, use the gradient descent method, Adam or SGD and other optimization algorithms to update the parameter θ, complete the training and evaluate the model performance on the validation set, and then adjust the learning rate and other hyperparameters for further optimization, and finally get a trained model. .

[0072] In step S4, the detection process of the waterweed image detection model includes the following steps:

[0073] S4101, the image to be detected Input the trained model , get the prediction result R, including the bounding box of the aquatic plant area and its confidence, and filter out valid detections according to the confidence threshold;

[0074] S4102, according to the upper left corner coordinates of the identified aquatic plant area boundary box ( , ) and the lower right corner coordinates ( , ), calculate the pixel area of ​​the water grass area in the image , ;

[0075] S4103, based on the ground sampling distance GSD and the pixel area of ​​the aquatic plant area in the image , calculate the actual area corresponding to the aquatic plant area in the figure , ;

[0076] S4104, according to the longitude and latitude coordinates of the satellite positioning coordinates corresponding to the center of the picture read from the EXIF ​​metadata in the picture ( , ), heading angle and pictures Size , and calculate the coordinates of the upper left corner of the rectangular box of the aquatic plant area in the figure by converting pixels into actual longitude and latitude coordinates ( , ) and the lower right corner coordinates ( , ) corresponds to the satellite positioning longitude and latitude coordinates .

[0077] In step S4103, the ground sampling distance GSD is calculated as: , where h is the flight altitude of the drone, is the camera's field of view. The larger the camera's field of view, the wider the coverage of a single shot. is the number of pixels of the drone camera image sensor in the corresponding length direction.

[0078] In step S4104, the calculation of the formula for converting pixels into actual longitude and latitude coordinates includes the following steps:

[0079] S1041. For any point in the figure ( , ), calculate the relative to the image center ( , ) and , the formula is and ;

[0080] S1042, the offset and Convert to geographic distance offset respectively and , and the formula is and ;

[0081] S1043, Use the rotation matrix Adjusting geographic distance offset and , , then the formula is ,

[0082] Right now ,

[0083] ;

[0084] S1044, adding the adjusted geographic offset to the longitude and latitude coordinates of the satellite positioning coordinates of the center of the image ( , ) to get the actual longitude coordinate of the pixel point and the actual latitude coordinates , and the formula is:

[0085] , , where R is the radius of the earth, which is 6371000m. Finally, the actual longitude coordinate of the pixel point is calculated. and the actual latitude coordinates The unit is angle;

[0086] S1045、( , )and( , ) respectively replace the steps S1041 to S1044 ( , ), you can get the corresponding actual longitude coordinates and actual latitude coordinates.

[0087] In step S5, the number of cages placed in the continuous aquatic plant area The calculation formula is: ,in represents the area of ​​the i-th continuous aquatic plant area, s represents the coverage area of ​​the cage, This means rounding up to ensure that the entire area is covered.

[0088] Embodiment 2: Figure 2 As shown, based on the first embodiment, the following steps are also included:

[0089] S8. Use the "Region Editing" tool in the GIS system to set the aquatic weed distribution area as an impassable obstacle area. Adopt an adaptive segmentation algorithm to dynamically adjust the segmentation size of the fishing grid according to the boundary of the fishing area and the distribution of aquatic weed obstacles to adapt to the complex shape of the water area. Then, plan the optimal operation trajectory of the fishing boat in the segmented grid to ensure that it avoids the aquatic weed area and covers the fishing area to the maximum extent. Then use the path planning tool to generate a fishing route and set multiple satellite positioning nodes on the route. Each node contains precise latitude and longitude coordinates. Export the generated fishing route node satellite positioning coordinate data, usually in KML or CSV format;

[0090] S9, transmitting the satellite positioning coordinate data file to the controller of the fish fry fishing boat through a USB or wireless network, starting the fish fry fishing boat according to the fishing route in the designated fish fry fishing area, and performing dynamic fishing in the designated fish fry fishing area.

[0091] Among them, in step S9, the fry fishing vessel should be equipped with a satellite positioning module to ensure that the fishing vessel can adjust its course according to real-time satellite positioning data during operation, so as to accurately move along the fishing route while avoiding aquatic plant areas and maximizing coverage of the designated fry fishing area.

[0092] like Figure 3 As shown, the operation of generating a fishing route in step S8 includes the following steps:

[0093] S801. According to the boundaries of the fishing area for fry and juvenile fish Generate a regular grid structure , ,in Represents the index of the grid cell. The initialized grid size is the same as the fishing width w;

[0094] S802: For each grid unit , based on the demarcation of the boundaries of the larvae and juvenile fishing areas and the boundary of the aquatic plant area The relationship is adaptively adjusted if a grid cell Contains both water and shallow water boundaries, or is located at the edge of an irregular area, that is, if and , then refine the grid cell The refinement process uses the quartering method to subdivide the grid cells into smaller grid cells to adapt to complex areas, and in order to avoid over-refinement, the maximum subdivision depth is set , if the refinement depth exceeds , then no further refinement;

[0095] S803, for the grid structure after adaptive subdivision , parallel routes are generated in each grid unit, and the route interval is the fishing width w, but in the refined grid area, the direction and distribution of the route are dynamically adjusted according to the shape of the grid, that is, in the regular grid area, the route is a parallel line, and in the irregular grid area, the direction of the route is adaptively adjusted according to the grid shape to ensure the coverage, so that the fishing route calculation formula is:

[0096] ,

[0097] Where i is the index number of the route. is the position coordinate of the i-th fishing route, is the position coordinate of the starting fishing route, w is the fishing width;

[0098] S804, Use the boundary of the aquatic plant area For the generated route segment Perform Boolean operations to remove overlapping parts and obtain effective fishing route segments , that is ;

[0099] S805, using the nearest neighbor algorithm to optimize the effective fishing route Perform global optimization to calculate the shortest path between routes;

[0100] S806, dynamically adjust the speed of the fishing boat according to the shape of the water area and the complexity of the route , maintain a higher speed in regular areas and slow down in irregular areas to increase fishing accuracy, The formula is:

[0101] ,

[0102] in Indicates the location of the larvae fishing boat in the grid cell The speed inside, is the maximum fishing speed of the fry fishing vessel, is the complexity adjustment factor used to control the impact of complexity on speed, is the i,jth grid cell The higher the complexity, the more irregular the grid area is or the more obstacles there are, and the fishing boat needs to slow down.

Claims

1. A method for optimizing the fishing position of fry and juvenile fish, characterized in that: The steps include: S1. Delineate the fishing area for fry and juvenile fish in natural waters, obtain the surveying and mapping data of the fishing area for fry and juvenile fish from the natural resources management department, and import it into the GIS system; S2. In the GIS system, a serpentine UAV flight path is generated using a grid method, the endpoints of each line segment in the flight path are converted into corresponding satellite positioning coordinates, and the UAV flight trajectory coordinate data based on the satellite positioning coordinates are output; S3, drone flight track coordinate data are imported into drone control system, drone is according to predetermined flight altitude h and flies with flight speed v, drone camera takes pictures with shooting frequency f simultaneously, and satellite positioning coordinates and the heading angle of drone are written in the metadata of picture when shooting, obtain a picture data set that delimits fry and juvenile fish catching area; S4, using the pictures in the picture data set demarcating the fry and juvenile fish fishing area, successively detecting pictures with water plants through the water plant picture detection model, obtaining all detected water plant pictures, and the water plant area and the minimum circumscribed rectangle coordinates of the water plant closed area in each detected water plant picture, and extracting the satellite positioning coordinates of the corresponding picture center read out from the EXIF ​​metadata in the picture, converting the diagonal point coordinates of the minimum circumscribed rectangle of the water plant closed area into satellite positioning coordinates, and outputting the water plant detection data containing the area field of the water plant closed area in the picture and the satellite positioning coordinate field of the diagonal points of the minimum circumscribed rectangle of the water plant closed area; S5. Import the waterweed detection data into the GIS system and highlight it, generate a waterweed distribution map, and count the area of ​​each continuous waterweed area in the GIS system , according to the coverage capacity of the cages, calculate the number of cages to be placed in each continuous aquatic plant area ; S6. The number of cages placed in each continuous aquatic plant area , in the GIS system, each continuous aquatic plant area is divided into The output contains the satellite positioning coordinates of the cage after the copy. The satellite positioning coordinate data of each cage placement is recorded as the attribute of the corresponding continuous aquatic plant area, and the satellite positioning coordinates of all cage placements are displayed on the aquatic plant distribution map in the form of highlighted dots; S7. The satellite positioning coordinate data of the cage placement output from the GIS system is imported into a portable satellite positioning device, and the cages are arranged at corresponding geographical locations in the designated fry and juvenile fish catching area.

2. A method for optimizing the catching position of fry according to claim 1, characterized in that: In step S3, the drone used is a drone with a fixed-altitude flight capability.

3. A method for optimizing the catching position of fry according to claim 1, characterized in that: In step S3, the formula for the shooting frequency f of the drone camera with the lens vertically facing is: ， Among them, f is the shooting frequency of the drone camera, v is the flight speed of the drone, is the camera's field of view. The larger the camera's field of view, the wider the coverage of a single shot. h is the drone's flight altitude, which determines the size of the shooting area. The higher the flight altitude, the larger the shooting field of view. o is the overlap, which is used to ensure that adjacent photos overlap to avoid blind spots.

4. A method for optimizing the catching position of fry according to claim 3, characterized in that: In step S5, the number of cages placed in the continuous aquatic plant area The calculation formula is: ， in represents the area of ​​the i-th continuous aquatic plant area, s represents the coverage area of ​​the cage, This means rounding up to ensure that the entire area is covered.

5. A method for optimizing the catching position of fry according to any one of claims 1 to 4, characterized in that: The following steps are also included: S8, in the GIS system, the aquatic plant area is set as an obstacle, an adaptive segmentation algorithm is used to plan the running track of the fry fishing boat, and a fishing route with satellite positioning nodes is generated, and satellite positioning coordinate data of the fishing route nodes are generated; S9, importing the satellite positioning coordinate data of the fishing route nodes into the controller of the fry fishing vessel, starting the fry fishing vessel according to the fishing route in the designated fry fishing area, and performing dynamic fishing in the designated fry fishing area.

6. A method for optimizing the catching position of fry according to claim 5, characterized in that: The operation of generating a fishing route in step S8 includes the following steps: S801. According to the boundaries of the fishing area for fry and juvenile fish Generate a regular grid structure , ,in Represents the index of the grid cell. The initialized grid size is the same as the fishing width w; S802: For each grid unit , based on the demarcation of the boundaries of the larvae and juvenile fishing areas and the boundary of the aquatic plant area The relationship is adaptively adjusted if a grid cell Contains both water and shallow water boundaries, or is located at the edge of an irregular area, that is, if and , then refine the grid cell The refinement process uses the quartering method to subdivide the grid cells into smaller grid cells to adapt to complex areas, and in order to avoid over-refinement, the maximum subdivision depth is set , if the refinement depth exceeds , then no further refinement; S803, for the grid structure after adaptive subdivision , parallel routes are generated in each grid unit, and the route interval is the fishing width w, but in the refined grid area, the direction and distribution of the route are dynamically adjusted according to the shape of the grid, that is, in the regular grid area, the route is a parallel line, and in the irregular grid area, the direction of the route is adaptively adjusted according to the grid shape to ensure the coverage, so that the fishing route calculation formula is: ， Where i is the index number of the route. is the position coordinate of the i-th fishing route, is the position coordinate of the starting fishing route, w is the fishing width; S804, Use the boundary of the aquatic plant area For the generated route segment Perform Boolean operations to remove overlapping parts and obtain effective fishing route segments , that is ; S805, using the nearest neighbor algorithm to optimize the effective fishing route Perform global optimization to calculate the shortest path between routes; S806, dynamically adjust the speed of the fishing boat according to the shape of the water area and the complexity of the route , maintain a higher speed in regular areas and slow down in irregular areas to increase fishing accuracy, The formula is: ， in Indicates the location of the larvae fishing boat in the grid cell The speed inside, is the maximum fishing speed of the fry fishing vessel, is the complexity adjustment factor used to control the impact of complexity on speed, is the i,jth grid cell The higher the complexity, the more irregular the grid area is or the more obstacles there are, and the fishing boat needs to slow down.

7. A method for optimizing the catching position of fry according to claim 1, characterized in that: In step S4, the detection process of the waterweed image detection model includes the following steps: S4101, the image to be detected Input the trained model , get the prediction result R, including the bounding box of the aquatic plant area and its confidence, and filter out valid detections according to the confidence threshold; S4102, according to the upper left corner coordinates of the identified aquatic plant area boundary box ( , ) and the lower right corner coordinates ( , ), calculate the pixel area of ​​the water grass area in the image , ; S4103, based on the ground sampling distance GSD and the pixel area of ​​the aquatic plant area in the image , calculate the actual area corresponding to the aquatic plant area in the figure , ; S4104, according to the longitude and latitude coordinates of the satellite positioning coordinates corresponding to the center of the picture read from the EXIF ​​metadata in the picture ( , ), heading angle and pictures Size , and calculate the coordinates of the upper left corner of the rectangular box of the aquatic plant area in the figure by converting pixels into actual longitude and latitude coordinates ( , ) and the lower right corner coordinates ( , ) corresponds to the satellite positioning longitude and latitude coordinates .

8. A method for optimizing the catching position of fry according to claim 7, characterized in that: In step S4103, the ground sampling distance GSD is calculated as: ， Where h is the flight altitude of the drone, is the camera's field of view. The larger the camera's field of view, the wider the coverage of a single shot. is the number of pixels of the drone camera image sensor in the corresponding length direction.

9. A method for optimizing the catching position of fry according to claim 7, characterized in that: In step S4104, the calculation of the formula for converting pixels into actual longitude and latitude coordinates includes the following steps: S1041. For any point in the figure ( , ), calculate the relative to the image center ( , ) and , the formula is and ; S1042, the offset and Convert to geographic distance offset respectively and , and the formula is and ; S1043, Use the rotation matrix Adjusting geographic distance offset and , and the formula is; ； S1044, adding the adjusted geographic offset to the longitude and latitude coordinates of the satellite positioning coordinates of the center of the image ( , ) to get the actual longitude coordinate of the pixel point and latitude coordinates , and the formula is: , , where R is the radius of the earth, which is 6371000m.

Images