Pesticide variable spraying system and method

Abstract

The invention discloses a pesticide variable spraying system and method, and relates to the technical field of plant protection intellectualization, the system comprises an unmanned aerial vehicle-mounted multispectral imager, a ground control center and a variable spraying device; the unmanned aerial vehicle-mounted multispectral imager obtains remote sensing image data of a to-be-sprayed area; the ground control center is used for segmenting the trees in the to-be-sprayed area based on the remote sensing image data by using a segmentation model to obtain segmented images; determining a ground sampling distance of the unmanned aerial vehicle-mounted multispectral imager; for any tree, determining the actual geographic position based on the bounding box, the width and height of the segmented image, the ground sampling distance, the longitude, the latitude and the latitude radian value of the central pixel point of the segmented image; based on the mask, the ground sampling distance, the image data of the red band and the image data of the near-infrared band, the pesticide spraying amount is determined; and the variable spraying device sprays pesticide to the corresponding trees based on the actual geographic positions and the pesticide spraying amounts of the trees. According to the invention, precise variable spraying of pesticide is realized.

Description

Technical Field

[0001] The present application relates to the field of intelligent plant protection technology, and in particular to a pesticide variable spraying system and method. Background Art

[0002] In modern agriculture, plant protection operations in orchards play a vital role in the prevention and control of pests and diseases. Traditional plant protection operations have problems such as large pesticide usage, uneven application, and low operation efficiency, resulting in waste of resources and environmental pollution. Especially in the process of pesticide spraying, traditional spraying methods cannot accurately apply pesticides according to the actual needs of crops. Uneven spraying or excessive use of pesticides often leads to waste of liquid medicine and environmental pollution. Harmful chemicals in pesticides not only have toxic effects on crops, but also enter the air and water through evaporation and runoff, endangering the ecological environment and human health.

[0003] Therefore, the efficiency and accuracy of pesticide spraying has become an agricultural technology problem that needs to be solved urgently. At present, there are mainly GPS navigation-assisted spraying systems and drone spraying systems to solve the problems of uneven pesticide spraying, excessive use and low spraying efficiency. However, these methods have shortcomings such as limited coverage and insufficient real-time response capabilities, and cannot achieve efficient and accurate spraying of large-scale orchards.

[0004] A Chinese patent document with publication number CN116142462A discloses a variable-variable pesticide spraying drone, comprising the following steps: step 1, a plurality of first spray pipes are rotatably arranged at the lower ends of a plurality of liquid guide pipes; step 2, a plurality of second spray pipes are fixedly connected to the bottom of a medicine box; step 3, two landing rods are respectively located at the front and rear sides of the medicine box, and the landing rods are connected to the lower end of the drone shell through two driving rod assemblies; step 4, a plurality of connecting ropes are respectively slid through positioning rings, and the two ends of the connecting ropes are respectively fixedly connected to the first spray pipes and the driving rod assemblies at corresponding positions, and the driving rod assemblies are used to pull the connecting ropes when the landing rods are in contact with the ground, so that the connecting ropes pull the first spray pipes to rotate; step 5, when the landing rods leave the ground, the first spray pipes rotate to a horizontal position under the action of their own gravity and pull the connecting ropes.

[0005] Among the above methods of variable spraying of pesticides by drones, the variable spraying of a single drone involves relatively high requirements for variables, and requires precise control of the spraying amount and range. Such a design requires a trade-off between the spraying effect and the complexity of the structure. In addition, although the design of the landing rod and the connecting rope can realize the rotation of the spraying tube, it lacks refined control of variable spraying in actual operation, and it is difficult to adjust the spraying parameters according to different crops and environmental conditions, thus failing to achieve truly precise variable spraying. Due to these limitations, the system complexity increases, mechanical failures are prone to occur, and maintenance is difficult, which affects the overall reliability and operational stability of the drone. Summary of the invention

[0006] The purpose of this application is to provide a pesticide variable spraying system and method to solve the problem that related technologies cannot achieve precise variable spraying of pesticides.

[0007] To achieve the above objectives, this application provides the following solutions:

[0008] In a first aspect, the present application provides a pesticide variable spraying system, comprising: an unmanned aerial vehicle-mounted multispectral imager, a ground control center, and a variable spraying device; the unmanned aerial vehicle-mounted multispectral imager and the variable spraying device are wirelessly connected to the ground control center respectively;

[0009] The multispectral imager carried by the unmanned aerial vehicle is used to obtain remote sensing image data of the area to be sprayed; the remote sensing image data includes: image data of the green band, image data of the red band, image data of the red edge band, and image data of the near infrared band;

[0010] The ground control center is used to:

[0011] Using a segmentation model, each tree in the area to be sprayed is segmented based on the remote sensing image data to obtain a segmented image; the segmented image is an image containing a mask and a bounding box of each tree; the segmentation model is obtained by training an improved Yolov8 neural network;

[0012] Acquire imaging parameters and position parameters of the multispectral imager carried by the unmanned aerial vehicle; the imaging parameters include: pixel size and true focal length; the position parameters include: longitude, latitude and relative height to the ground;

[0013] Determining a ground sampling distance of the multispectral imager on board the unmanned aerial vehicle based on the imaging parameters and the relative height;

[0014] Determine any tree as the current tree;

[0015] Determining a pixel position offset of the current tree based on a bounding box of the current tree in the segmented image and a width and a height of the segmented image;

[0016] Determine the geographical location offset of the current tree based on the pixel position offset of the current tree and the ground sampling distance;

[0017] Determine the actual geographical location of the current tree based on the geographical location offset of the current tree, the longitude, the latitude and the latitude arc value of the central pixel point of the segmented image;

[0018] Determining a crown area of ​​a current tree based on a mask of the current tree in the segmented image and the ground sampling distance;

[0019] Determine the spraying amount of the current tree based on the crown area of ​​the current tree, the image data of the red band and the image data of the near infrared band;

[0020] The variable spraying device is used to spray pesticides on corresponding trees based on the actual geographical location and spraying amount of each tree.

[0021] Optionally, the wavelength range of the green band is 560nm±16nm, the wavelength range of the red band is 650nm±16nm, the wavelength range of the red edge band is 730nm±16nm, and the wavelength range of the near infrared band is 860nm±26nm.

[0022] Optionally, the process of determining the segmentation model includes:

[0023] Acquire a training set; the training set includes: remote sensing image data of multiple training areas and corresponding segmented images;

[0024] An improved Yolov8 neural network is constructed; the improved Yolov8 neural network includes: a backbone network, a feature fusion network and a detection network; the backbone network includes: 5 GhostConv modules, 2 RepVGG modules, 4 C2f modules and 1 SPPF module, the feature fusion network includes: 6 feature splicing modules, 3 upsampling modules, 6 C2f modules and 3 GhostConv modules, and the detection network includes: 4 detection heads;

[0025] The improved Yolov8 neural network is trained with the remote sensing image data of each training area as input and the corresponding segmented image as output to obtain the segmentation model.

[0026] Optionally, determining a ground sampling distance of the unmanned aerial vehicle-mounted multispectral imager based on the imaging parameter and the relative height includes:

[0027] The ground sampling distance calculation formula is used to calculate the ground sampling distance according to the imaging parameters and the relative height; the ground sampling distance calculation formula includes:

[0028]

[0029] Among them, H is the ground sampling distance; h1 is the relative height; s is the pixel size; and f is the real focal length.

[0030] Optionally, determining the pixel position offset of the current tree based on the bounding box of the current tree in the segmented image and the width and height of the segmented image includes:

[0031] The pixel position offset calculation formula is used to calculate the pixel position offset of the current tree according to the boundary box of the current tree in the segmented image and the width and height of the segmented image; the pixel position offset calculation formula includes:

[0032]

[0033] Among them, P x is the x-axis component of the pixel position offset; N x is the x-axis component of the position of the center pixel of the bounding box; w is the width of the segmented image; P y is the y-axis component of the pixel position offset; h is the height of the segmented image; N y The y-axis component of the position of the center pixel of the bounding box.

[0034] Optionally, determining the geographical location offset of the current tree based on the pixel position offset of the current tree and the ground sampling distance includes:

[0035] The geographic location offset calculation formula is used to calculate the geographic location offset of the current tree according to the pixel position offset of the current tree and the ground sampling distance; the geographic location offset calculation formula includes:

[0036] G x =P x H;

[0037] G y =P y H;

[0038] Among them, G x is the x-axis component of the geographic location offset; G y The y-axis component of the geographic location offset.

[0039] Optionally, determining the actual geographical location of the current tree based on the geographical location offset of the current tree, the longitude, the latitude and the latitude arc value of the central pixel point of the segmented image includes:

[0040] The actual geographical location calculation formula is used to calculate the actual geographical location of the current tree according to the geographical location offset of the current tree, the longitude, the latitude and the latitude arc value of the central pixel point of the segmented image; the actual geographical location calculation formula includes:

[0041]

[0042] Among them, α2 is the longitude in the actual geographical location; α1 is the longitude in the location parameter; γ is the latitude arc value of the central pixel point of the segmented image; β2 is the latitude in the actual geographical location; β1 is the latitude in the location parameter.

[0043] Optionally, determining the crown area of ​​the current tree based on the mask of the current tree in the segmented image and the ground sampling distance includes:

[0044] The crown area calculation formula is used to calculate the crown area of ​​the current tree according to the mask of the current tree in the segmented image and the ground sampling distance; the crown area calculation formula includes:

[0045] A=M·H 2 ;

[0046] Among them, A is the crown area; M is the total number of pixels in the tree mask.

[0047] Optionally, based on the crown area of ​​the current tree, the image data of the red band and the image data of the near infrared band, the spraying amount of the current tree is determined, including:

[0048] The spraying amount calculation formula is used to calculate the spraying amount of the current tree according to the crown area of ​​the current tree, the image data of the red band and the image data of the near infrared band; the spraying amount calculation formula includes:

[0049]

[0050] Among them, Q is the spraying amount; k is the correction coefficient; NDVI is the normalized vegetation index; NIR is the image data of the near-infrared band; and R is the image data of the red band.

[0051] In a second aspect, the present application provides a method for variable spraying of pesticides, which is implemented using a variable spraying system for pesticides. The method comprises:

[0052] Acquire remote sensing image data of the area to be sprayed by using a multispectral imager carried by an unmanned aerial vehicle; the remote sensing image data includes: image data of a green band, image data of a red band, image data of a red edge band, and image data of a near infrared band;

[0053] Using a segmentation model, each tree in the area to be sprayed is segmented based on the remote sensing image data to obtain a segmented image; the segmented image is an image containing a mask and a bounding box of each tree; the segmentation model is obtained by training an improved Yolov8 neural network;

[0054] Acquire imaging parameters and position parameters of the multispectral imager carried by the unmanned aerial vehicle; the imaging parameters include: pixel size and true focal length; the position parameters include: longitude, latitude and relative height to the ground;

[0055] Determining a ground sampling distance of the multispectral imager on board the unmanned aerial vehicle based on the imaging parameters and the relative height;

[0056] Determine any tree as the current tree;

[0057] Determine a pixel position offset of a bounding box of a current tree based on a bounding box of a current tree in the segmented image and a width and a height of the segmented image;

[0058] Determine the geographic location offset of the current tree based on the pixel position offset of the bounding box of the current tree and the ground sampling distance;

[0059] Determine the actual geographical location of the current tree based on the geographical location offset of the current tree, the longitude, the latitude and the latitude arc value of the central pixel point of the segmented image;

[0060] Determining a crown area of ​​a current tree based on a mask of the current tree in the segmented image and the ground sampling distance;

[0061] Determine the spraying amount of the current tree based on the crown area of ​​the current tree, the image data of the red band and the image data of the near infrared band;

[0062] Pesticides are sprayed on the corresponding trees based on their actual geographical locations and spraying amounts.

[0063] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0064] The present application discloses a pesticide variable spraying system and method, the system comprises: an unmanned aerial vehicle multispectral imager, a ground control center and a variable spraying device; the unmanned aerial vehicle multispectral imager is used to obtain remote sensing image data of a spraying area; the ground control center is used to: use a segmentation model to segment each tree in the spraying area based on the remote sensing image data to obtain a segmented image; obtain imaging parameters and position parameters of the unmanned aerial vehicle multispectral imager; the imaging parameters include: pixel size and true focal length; the position parameters include: longitude, latitude and relative height to the ground; based on the imaging parameters and relative height, determine the ground sampling distance of the unmanned aerial vehicle multispectral imager; determine any tree as the current tree; based on the segmentation The pixel position offset of the current tree is determined based on the bounding box of the current tree in the image and the width and height of the segmented image; the geographical location offset of the current tree is determined based on the pixel position offset of the current tree and the ground sampling distance; the actual geographical location of the current tree is determined based on the geographical location offset, longitude, latitude and the latitude arc value of the central pixel of the segmented image; the crown area of ​​the current tree is determined based on the mask of the current tree in the segmented image and the ground sampling distance; the spraying amount of the current tree is determined based on the crown area of ​​the current tree, the image data of the red band and the image data of the near-infrared band; the variable spraying device is used to spray pesticides on the corresponding trees based on the actual geographical location and spraying amount of each tree. This application is based on the joint cooperation of the multi-spectral imager carried by the unmanned aerial vehicle, the ground control center and the variable spraying device, and can realize the coordinated operation of the variable spraying devices carried by multiple unmanned aerial vehicles, accurately locate the trees and calculate the spraying amount, improve the utilization rate of pesticides, reduce environmental pollution, and improve the operation efficiency and coverage. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0066] Figure 1 A schematic diagram of the structure of a variable-rate pesticide spraying system provided in one embodiment of the present application;

[0067] Figure 2 This is a schematic diagram of the pesticide variable spraying system architecture;

[0068] Figure 3 To segment the image;

[0069] Figure 4 This is a schematic diagram of the improved Yolov8 neural network structure;

[0070] Figure 5 This is a schematic diagram of the spraying process of the variable spraying device. DETAILED DESCRIPTION

[0071] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0072] The purpose of this application is to provide a pesticide variable spraying system and method, aiming to achieve precise variable spraying of pesticides.

[0073] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0074] In an exemplary embodiment, Figure 1 and Figure 2 As shown, the variable pesticide spraying system in this embodiment includes: an unmanned aerial vehicle-mounted multi-spectral imager, a ground control center and a variable spraying device; the unmanned aerial vehicle-mounted multi-spectral imager and the variable spraying device are wirelessly connected to the ground control center respectively.

[0075] Specifically, the multispectral imager on the drone uses the DJI M3M-multispectral camera, which has a maximum flight time of 43 minutes and a maximum range of 32 km. The ground control center uses a high-performance computer for real-time data processing and analysis. The hardware requirements are a processor of i7-7700HQ or above, a memory of 4G or above, and a hard disk of 50G or above; the software requirements are Windows 10 64-bit system and Python 3.8 or above.

[0076] The multispectral imager carried by the unmanned aerial vehicle is used to obtain remote sensing image data of the area to be sprayed; the remote sensing image data includes: image data of the green band, image data of the red band, image data of the red edge band, and image data of the near infrared band.

[0077] As an optional implementation, the wavelength range of the green band is 560nm±16nm, the wavelength range of the red band is 650nm±16nm, the wavelength range of the red edge band is 730nm±16nm, and the wavelength range of the near infrared band is 860nm±26nm.

[0078] Ground Control Centre for:

[0079] Step 01: Use the segmentation model to segment the trees in the spraying area based on the remote sensing image data to obtain a segmented image; the segmented image is an image containing the mask and bounding box of each tree; the segmentation model is obtained by training the improved Yolov8 neural network.

[0080] like Figure 3 As shown, each red rectangle in the segmented image is the bounding box of each tree, and the red covered area in each bounding box is the mask of each tree.

[0081] As an optional implementation, the process of determining the segmentation model includes:

[0082] Step 011: Obtain a training set; the training set includes: remote sensing image data of multiple training areas and corresponding segmented images.

[0083] Step 012: Construct the improved Yolov8 neural network. Figure 4 As shown in the figure, the improved Yolov8 neural network includes: a backbone network, a feature fusion network and a detection network; the backbone network includes: 5 GhostConv modules, 2 RepVGG modules, 4 C2f modules and 1 SPPF module, the feature fusion network includes: 6 feature splicing modules, 3 upsampling modules, 6 C2f modules and 3 GhostConv modules, and the detection network includes: 4 detection heads.

[0084] Among them, the GhostConv module is a module used to reduce the computational cost of convolutional neural networks. The GhostConv module generates a small number of intrinsic feature maps and applies simple linear transformations to them to produce multiple virtual feature maps to reduce the need for convolution operations. Specifically, the GhostConv module first generates a small number of intrinsic feature maps through primary convolution, and then applies linear operations to each intrinsic feature map to generate virtual feature maps, thereby reducing the filters and the amount of computation while maintaining the network's expressiveness. The RepVGG module is a VGG-style convolutional network structure that uses a five-stage 3×3 convolutional layer arrangement, and the first layer of each stage is downsampled by a convolution with a stride of 2. In order to reduce the amount of computation and the number of parameters, the RepVGG module inserts grouped convolutions in some convolutional layers and adjusts the channel width of each stage by scaling factors, with a larger number of channels in the last stage to enhance the feature expression capability. In addition, the module uses global average pooling and fully connected layers as the head structure in image classification tasks.

[0085] Step 013: Using the remote sensing image data of each training area as input and the corresponding segmented image as output, the improved Yolov8 neural network is trained to obtain a segmentation model.

[0086] Step 02: Obtain the imaging parameters and position parameters of the multispectral imager on the drone; the imaging parameters include: pixel size and true focal length; the position parameters include: longitude, latitude and relative height to the ground.

[0087] Specifically, the imaging parameters and position parameters of the multispectral imager carried by the UAV are obtained by using the EXIF ​​data file storing the additional information of the remote sensing image data.

[0088] Step 03: Determine the ground sampling distance of the UAV-mounted multispectral imager based on the imaging parameters and relative altitude.

[0089] As an optional implementation, based on the imaging parameters and the relative altitude, determining the ground sampling distance of the multispectral imager on the unmanned aerial vehicle includes:

[0090] The ground sampling distance calculation formula is used to calculate the ground sampling distance according to the imaging parameters and relative height; the ground sampling distance calculation formula includes:

[0091]

[0092] Among them, H is the ground sampling distance; h1 is the relative height; s is the pixel size; and f is the real focal length.

[0093] Step 04: Identify any tree as the current tree.

[0094] Step 05: Determine the pixel position offset of the current tree based on the bounding box of the current tree in the segmented image and the width and height of the segmented image.

[0095] As an optional implementation, determining the pixel position offset of the current tree based on the bounding box of the current tree in the segmented image and the width and height of the segmented image includes:

[0096] The pixel position offset calculation formula is used to calculate the pixel position offset of the current tree according to the boundary box of the current tree in the segmented image and the width and height of the segmented image; the pixel position offset calculation formula includes:

[0097]

[0098] Among them, P x is the x-axis component of the pixel position offset; N x is the x-axis component of the position of the center pixel of the bounding box; w is the width of the segmented image; P y is the y-axis component of the pixel position offset; h is the height of the segmented image; N y The y-axis component of the position of the center pixel of the bounding box.

[0099] Step 06: Determine the geographic location offset of the current tree based on the pixel position offset of the current tree and the ground sampling distance.

[0100] As an optional implementation, determining the geographical location offset of the current tree based on the pixel position offset of the current tree and the ground sampling distance includes:

[0101] The geographic location offset calculation formula is used to calculate the geographic location offset of the current tree according to the pixel position offset of the current tree and the ground sampling distance; the geographic location offset calculation formula includes:

[0102] G x =P x ·H.

[0103] G y =P y ·H.

[0104] Among them, G x is the x-axis component of the geographic location offset; G y The y-axis component of the geographic location offset.

[0105] Step 07: Determine the actual geographic location of the current tree based on the geographic location offset, longitude, latitude and the latitude arc value of the central pixel point of the segmented image.

[0106] As an optional implementation, the actual geographical location of the current tree is determined based on the geographical location offset, longitude, latitude and the latitude arc value of the central pixel point of the segmented image, including:

[0107] The actual geographic location calculation formula is used to calculate the actual geographic location of the current tree according to the geographic location offset, longitude, latitude and the latitude arc value of the central pixel point of the segmented image; the actual geographic location calculation formula includes:

[0108]

[0109] Among them, α2 is the longitude in the actual geographical location; α1 is the longitude in the location parameter; γ is the latitude arc value of the central pixel point of the segmented image; β2 is the latitude in the actual geographical location; β1 is the latitude in the location parameter.

[0110] Step 08: Determine the crown area of ​​the current tree based on the mask and ground sampling distance of the current tree in the segmented image.

[0111] As an optional implementation, determining the crown area of ​​the current tree based on the mask of the current tree in the segmented image and the ground sampling distance includes:

[0112] The crown area calculation formula is used to calculate the crown area of ​​the current tree according to the mask of the current tree in the segmented image and the ground sampling distance; the crown area calculation formula includes:

[0113] A=M·H 2 .

[0114] Among them, A is the crown area; M is the total number of pixels in the tree mask.

[0115] Step 09: Determine the spraying amount for the current tree based on the crown area of ​​the current tree, the image data of the red band, and the image data of the near-infrared band.

[0116] As an optional implementation, the spraying amount of the current tree is determined based on the crown area of ​​the current tree, the image data of the red band, and the image data of the near infrared band, including:

[0117] The spraying amount calculation formula is used to calculate the spraying amount of the current tree according to the crown area of ​​the current tree, the image data of the red band and the image data of the near-infrared band; the spraying amount calculation formula includes:

[0118]

[0119] Among them, Q is the spraying amount; k is the correction coefficient; NDVI is the normalized vegetation index; NIR is the image data of the near-infrared band; and R is the image data of the red band.

[0120] Specifically, the correction coefficient is related to the types of trees and pests in the area to be sprayed. After the area to be sprayed is determined, the correction coefficient is also determined.

[0121] Step 10: The variable spraying device is used to spray pesticides on the corresponding trees based on the actual geographical location and spraying amount of each tree.

[0122] Specifically, the hardware structure of the variable spraying device includes: a spray rod, a water pump, an electronic control device, a water tank and its pipelines. Among them, the quadcopter equipped with the variable spraying device uses a quick-detachable lateral spray rod. The water pump is powered by an electronic speed regulator and can control the water pump pressure according to the PWM signal provided by the flight control system on the quadcopter. The electrical control part is mainly composed of a power supply and an electronic speed regulator. The power supply uses a lithium battery. The input end of the electronic speed regulator is connected to the battery, the signal line is connected to the flight control, and the output end is connected to the water pump. The water tank and pipelines use a special water tank and high-pressure water pipe system for agricultural plant protection.

[0123] like Figure 5As shown in the figure, when there is a need for spraying operations, the actual geographical location and spraying amount of each tree input externally are used to spray pesticides on the corresponding trees to adjust the pulse width of the PWM signal output by the flight control board to control the pressure of the water pump, thereby achieving variable spraying of the amount of medicine. For a quad-rotor drone, the spray rod of the spraying system uses a vertical spray rod and a long-tube fan-shaped high-pressure nozzle, which is connected to the brushless water pump and water tank through a high-pressure water pipe. The spray rod is fixed on the four arms and is located directly below the electronic speed regulator. The water tank is fixed on the bottom plate of the frame, and the water pump is fixed on the water tank.

[0124] In an exemplary embodiment, a method for variable-rate spraying of pesticides is provided. The method for variable-rate spraying of pesticides is implemented by using a variable-rate spraying system for pesticides. The method for variable-rate spraying of pesticides includes:

[0125] The remote sensing image data of the area to be sprayed is obtained by a multi-spectral imager carried by an unmanned aerial vehicle; the remote sensing image data includes: image data of the green band, image data of the red band, image data of the red edge band, and image data of the near infrared band.

[0126] Using the segmentation model, the trees in the spraying area are segmented based on the remote sensing image data to obtain a segmented image; the segmented image is an image containing the mask and bounding box of each tree; the segmentation model is obtained by training the improved Yolov8 neural network.

[0127] Obtain imaging parameters and position parameters of the multispectral imager onboard the UAV; imaging parameters include: pixel size and true focal length; position parameters include: longitude, latitude and relative height to the ground.

[0128] Based on the imaging parameters and relative altitude, the ground sampling distance of the UAV-mounted multispectral imager is determined.

[0129] Identify any tree as the current tree.

[0130] Based on the bounding box of the current tree in the segmented image and the width and height of the segmented image, the pixel position offset of the bounding box of the current tree is determined.

[0131] Determine the geographic location offset of the current tree based on the pixel position offset of the current tree's bounding box and the ground sampling distance.

[0132] The actual geographic location of the current tree is determined based on the geographic location offset, longitude, latitude and the latitude arc value of the central pixel point of the segmented image.

[0133] Determine the crown area of ​​the current tree based on the mask of the current tree in the segmented image and the ground sampling distance.

[0134] The spraying amount of the current tree is determined based on the crown area of ​​the current tree, the image data of the red band, and the image data of the near-infrared band.

[0135] Pesticides are sprayed on the corresponding trees based on their actual geographical locations and spraying amounts.

[0136] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0137] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the system, method and core ideas of this application; at the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A pesticide variable spraying system, characterized in that: The variable-rate pesticide spraying system comprises: an unmanned aerial vehicle-mounted multispectral imager, a ground control center and a variable-rate spraying device; the unmanned aerial vehicle-mounted multispectral imager and the variable-rate spraying device are wirelessly connected to the ground control center respectively; The multispectral imager carried by the unmanned aerial vehicle is used to obtain remote sensing image data of the area to be sprayed; the remote sensing image data includes: image data of the green band, image data of the red band, image data of the red edge band, and image data of the near infrared band; The ground control center is used to: Using a segmentation model, each tree in the area to be sprayed is segmented based on the remote sensing image data to obtain a segmented image; the segmented image is an image containing a mask and a bounding box of each tree; the segmentation model is obtained by training an improved Yolov8 neural network; Acquire imaging parameters and position parameters of the multispectral imager carried by the unmanned aerial vehicle; the imaging parameters include: pixel size and true focal length; the position parameters include: longitude, latitude and relative height to the ground; Determining a ground sampling distance of the multispectral imager on board the unmanned aerial vehicle based on the imaging parameters and the relative height; Determine any tree as the current tree; Determining a pixel position offset of the current tree based on a bounding box of the current tree in the segmented image and a width and a height of the segmented image; Determine the geographical location offset of the current tree based on the pixel position offset of the current tree and the ground sampling distance; Determine the actual geographical location of the current tree based on the geographical location offset of the current tree, the longitude, the latitude and the latitude arc value of the central pixel point of the segmented image; Determining a crown area of ​​a current tree based on a mask of the current tree in the segmented image and the ground sampling distance; Determine the spraying amount of the current tree based on the crown area of ​​the current tree, the image data of the red band and the image data of the near infrared band; The variable spraying device is used to spray pesticides on corresponding trees based on the actual geographical location and spraying amount of each tree.

2. The variable-rate pesticide spraying system according to claim 1, characterized in that: The wavelength range of the green band is 560nm±16nm, the wavelength range of the red band is 650nm±16nm, the wavelength range of the red edge band is 730nm±16nm, and the wavelength range of the near infrared band is 860nm±26nm.

3. The variable-rate pesticide spraying system according to claim 2, characterized in that: The process of determining the segmentation model includes: Acquire a training set; the training set includes: remote sensing image data of multiple training areas and corresponding segmented images; An improved Yolov8 neural network is constructed; the improved Yolov8 neural network includes: a backbone network, a feature fusion network and a detection network; the backbone network includes: 5 GhostConv modules, 2 RepVGG modules, 4 C2f modules and 1 SPPF module, the feature fusion network includes: 6 feature splicing modules, 3 upsampling modules, 6 C2f modules and 3 GhostConv modules, and the detection network includes: 4 detection heads; The improved Yolov8 neural network is trained with the remote sensing image data of each training area as input and the corresponding segmented image as output to obtain the segmentation model.

4. The pesticide variable spraying system according to claim 1, characterized in that: Determining a ground sampling distance of the multispectral imager on the unmanned aerial vehicle based on the imaging parameters and the relative height includes: The ground sampling distance calculation formula is used to calculate the ground sampling distance according to the imaging parameters and the relative height; the ground sampling distance calculation formula includes: Among them, H is the ground sampling distance; h1 is the relative height; s is the pixel size; and f is the real focal length.

5. The variable-rate pesticide spraying system according to claim 4, characterized in that: Determining a pixel position offset of the current tree based on a boundary box of the current tree in the segmented image and a width and a height of the segmented image includes: The pixel position offset calculation formula is used to calculate the pixel position offset of the current tree according to the boundary box of the current tree in the segmented image and the width and height of the segmented image; the pixel position offset calculation formula includes: Among them, P x is the x-axis component of the pixel position offset; N x is the x-axis component of the position of the center pixel of the bounding box; w is the width of the segmented image; P y is the y-axis component of the pixel position offset; h is the height of the segmented image; N y The y-axis component of the position of the center pixel of the bounding box.

6. The variable-rate pesticide spraying system according to claim 5, characterized in that: Based on the pixel position offset of the current tree and the ground sampling distance, the geographical position offset of the current tree is determined, including: The geographic location offset calculation formula is used to calculate the geographic location offset of the current tree according to the pixel position offset of the current tree and the ground sampling distance; the geographic location offset calculation formula includes: G x =P x ·H; G y =P y ·H; Among them, G x is the x-axis component of the geographic location offset; G y The y-axis component of the geographic location offset.

7. The pesticide variable spraying system according to claim 6, characterized in that: Determining the actual geographical location of the current tree based on the geographical location offset of the current tree, the longitude, the latitude, and the latitude arc value of the central pixel point of the segmented image includes: The actual geographical location calculation formula is used to calculate the actual geographical location of the current tree according to the geographical location offset of the current tree, the longitude, the latitude and the latitude arc value of the central pixel point of the segmented image; the actual geographical location calculation formula includes: Among them, α2 is the longitude in the actual geographical location; α1 is the longitude in the location parameter; γ is the latitude arc value of the central pixel point of the segmented image; β2 is the latitude in the actual geographical location; β1 is the latitude in the location parameter.

8. The variable-rate pesticide spraying system according to claim 7, characterized in that: Determining a crown area of ​​a current tree based on a mask of the current tree in the segmented image and the ground sampling distance includes: The crown area calculation formula is used to calculate the crown area of ​​the current tree according to the mask of the current tree in the segmented image and the ground sampling distance; the crown area calculation formula includes: A=M·H 2 ; Among them, A is the crown area; M is the total number of pixels in the tree mask.

9. The variable-rate pesticide spraying system according to claim 8, characterized in that: Based on the crown area of ​​the current tree, the image data of the red band and the image data of the near infrared band, the spraying amount of the current tree is determined, including: The spraying amount calculation formula is used to calculate the spraying amount of the current tree according to the crown area of ​​the current tree, the image data of the red band and the image data of the near infrared band; the spraying amount calculation formula includes: Among them, Q is the spraying amount; k is the correction coefficient; NDVI is the normalized vegetation index; NIR is the image data of the near-infrared band; and R is the image data of the red band.

10. A method for variable-rate spraying of pesticides, the method for variable-rate spraying of pesticides being implemented by using the variable-rate spraying system of pesticides as claimed in any one of claims 1 to 9, characterized in that: The variable-rate pesticide spraying method comprises: Acquire remote sensing image data of the area to be sprayed by using a multispectral imager carried by an unmanned aerial vehicle; the remote sensing image data includes: image data of a green band, image data of a red band, image data of a red edge band, and image data of a near infrared band; Using a segmentation model, each tree in the area to be sprayed is segmented based on the remote sensing image data to obtain a segmented image; the segmented image is an image containing a mask and a bounding box of each tree; the segmentation model is obtained by training an improved Yolov8 neural network; Acquire imaging parameters and position parameters of the multispectral imager carried by the unmanned aerial vehicle; the imaging parameters include: pixel size and true focal length; the position parameters include: longitude, latitude and relative height to the ground; Determining a ground sampling distance of the multispectral imager on board the unmanned aerial vehicle based on the imaging parameters and the relative height; Determine any tree as the current tree; Determine a pixel position offset of a bounding box of a current tree based on a bounding box of a current tree in the segmented image and a width and a height of the segmented image; Determine the geographic location offset of the current tree based on the pixel position offset of the bounding box of the current tree and the ground sampling distance; Determine the actual geographical location of the current tree based on the geographical location offset of the current tree, the longitude, the latitude and the latitude arc value of the central pixel point of the segmented image; Determining a crown area of ​​a current tree based on a mask of the current tree in the segmented image and the ground sampling distance; Determine the spraying amount of the current tree based on the crown area of ​​the current tree, the image data of the red band and the image data of the near infrared band; Pesticides are sprayed on the corresponding trees based on their actual geographical locations and spraying amounts.

Images