Plant protection unmanned aerial vehicle spraying droplet form trajectory capturing method, device, equipment and medium

Abstract

The invention relates to a plant protection unmanned aerial vehicle spraying droplet form trajectory capturing method, device and equipment and a medium, and the method comprises the steps: carrying out the three-dimensional reconstruction of contour mask data corresponding to a droplet group under each view angle according to a preset three-dimensional reconstruction algorithm based on the multi-view angle geometric constraint, so as to generate a three-dimensional contour model of the droplet group; extracting an outer bounding box of a fogdrop group in each frame of fogdrop image from the three-dimensional contour model, and calculating to obtain a spatial size parameter of the outer bounding box of the fogdrop group in a single frame of fogdrop image and a corresponding three-dimensional center coordinate; and performing time correlation analysis on the three-dimensional center coordinates of the fog drop group in the continuous frames of fog drop images to construct a three-dimensional time sequence track of the fog drop group, associating the three-dimensional time sequence track with corresponding spatial size parameters, and outputting the three-dimensional time sequence track in a curve, table or three-dimensional dynamic model form. According to the invention, synchronous quantification of the motion trail and the morphological characteristics of the fog drop group is realized.

Description

Technical Field

[0001] This application relates to the field of agricultural production, and in particular to a method for capturing the morphology and trajectory of spray droplets from a crop protection drone, as well as corresponding devices, electronic equipment, and computer-readable storage media. Background Technology

[0002] Agricultural drones have become central to pesticide spraying in modern agricultural pest and disease control due to their high efficiency, adaptability, and ability to effectively reduce manual labor. However, the morphology, trajectory, and spatial distribution of droplets are key factors directly determining the spraying effect. Therefore, accurately acquiring the dynamic spatiotemporal characteristics of droplets is of significant theoretical and practical value for guiding the practical application of agricultural drones.

[0003] Existing droplet detection technologies suffer from several shortcomings. Some technologies focus on measuring the size and velocity of individual droplets, typically relying on complex laser imaging systems and high-precision algorithms. These technologies are costly and only acquire limited information about localized areas, failing to reflect the overall distribution patterns of the droplet population. Others collect droplet deposition data through traditional field trials. This method is susceptible to interference from environmental factors such as wind speed, temperature, and humidity, resulting in poor data repeatability. Furthermore, it cannot track the complete trajectory of droplets from nozzle ejection to final settling in real time, making it difficult to fully meet the needs of accurate detection and parameter optimization. In addition, some existing laboratory testing devices have limitations in spatial coverage and insufficient temporal sampling frequency, making it difficult to fully record the dynamic changes of the droplet population, thus hindering the in-depth optimization of spraying parameters based on the collected data.

[0004] In summary, the current technology for capturing the morphology and trajectory of spray droplets from agricultural drones relies on complex laser imaging systems and high-precision algorithms, resulting in high equipment costs. Furthermore, traditional field trials are easily affected by environmental factors such as wind speed, temperature, and humidity, leading to poor data repeatability. The applicant has made corresponding explorations to address these issues. Summary of the Invention

[0005] The purpose of this application is to solve the above-mentioned problems by providing a method, device, electronic equipment and computer-readable storage medium for capturing the morphology and trajectory of spray droplets from agricultural drones.

[0006] To achieve the various objectives of this application, the following technical solution is adopted: A method for capturing the morphology and trajectory of droplets sprayed by an agricultural drone, proposed to meet one of the purposes of this application, includes: After the surface light source module is activated and stabilized to emit light, the agricultural drone is controlled to perform spraying operations according to preset parameters. At the same time, the image acquisition module is triggered to continuously acquire multiple frames of droplet images according to the preset image acquisition frame rate to construct a droplet image sequence. The image acquisition module is a multi-vision industrial camera array constructed from multiple industrial cameras, and the droplet image includes one or more droplet groups. A preset image segmentation algorithm is invoked to segment the fog droplet image, and the complete contour region of the fog droplet group in the fog droplet image under a single view is extracted to determine the contour mask data corresponding to the fog droplet group. Based on preset multi-view geometric constraints, the contour mask data corresponding to the fog droplet group under each viewpoint is reconstructed in three dimensions according to the preset three-dimensional reconstruction algorithm to generate a three-dimensional contour model of the fog droplet group. Extract the bounding box of the fog droplet group in each frame of fog droplet image from the three-dimensional contour model, and calculate the spatial size parameters of the bounding box of the fog droplet group in a single frame of fog droplet image and its corresponding three-dimensional center coordinates. Temporal correlation analysis is performed on the three-dimensional center coordinates of the droplet population in continuous frame droplet images to construct the three-dimensional time series trajectory of the droplet population. The three-dimensional time series trajectory is then associated with its corresponding spatial dimension parameters and output in the form of curves, tables, or three-dimensional dynamic models to complete the morphological trajectory capture of the sprayed droplets by the agricultural drone.

[0007] Optionally, the step of calling a preset image segmentation algorithm to segment the fog droplet image and extracting the complete contour region of the fog droplet group in the fog droplet image from a single viewpoint to determine the contour mask data corresponding to the fog droplet group includes: The fog image is input into an image segmentation algorithm that has been trained to convergence. Based on the image segmentation algorithm, the pixel features of the fog droplets and the background in the fog image are classified to separate the fog droplets and the background in the fog image from a single viewpoint. Based on the segmentation results of the image segmentation algorithm, the complete contour region of the fog droplet group is extracted to remove background interference pixels and isolated noise points, and the contour region data corresponding to the fog droplet group under a single view is obtained. The contour region data corresponding to the fog droplet group is binarized, and the contour region data is marked as valid foreground pixels and the background region is marked as invalid pixels, thereby generating and determining the contour mask data corresponding to the fog droplet group.

[0008] Optionally, the step of extracting the bounding box of the fog droplet group in each frame of the fog droplet image from the three-dimensional contour model, and calculating the spatial dimension parameters of the bounding box of the fog droplet group in a single frame of the fog droplet image and its corresponding three-dimensional center coordinates, includes: For a three-dimensional contour model of a group of fog droplets in a single-frame fog droplet image, all three-dimensional pixels in the three-dimensional contour model are traversed to determine the extreme coordinates of the fog droplet group in three-dimensional space. The extreme coordinates include the maximum and minimum coordinates in the horizontal axis direction, the maximum and minimum coordinates in the vertical axis direction, and the maximum and minimum coordinates in the vertical direction. Based on the extreme coordinates, a minimum bounding box is constructed for the fog droplet population so that the minimum bounding box completely encloses the three-dimensional outline of the fog droplet population, and each face of the minimum bounding box is parallel to the coordinate axes of the multi-camera unified three-dimensional coordinate system. Based on the three-dimensional space of the fog droplet group, the first difference between the maximum value of the horizontal axis coordinate and the corresponding minimum value is used as the length parameter in the spatial dimension parameters; the second difference between the maximum value of the vertical axis coordinate and the corresponding minimum value is used as the width parameter in the spatial dimension parameters; and the third difference between the maximum value of the vertical axis coordinate and the corresponding minimum value is used as the height parameter in the spatial dimension parameters. The first average value between the maximum and minimum values ​​of the coordinates in the horizontal direction is calculated and used as the horizontal axis coordinate in the three-dimensional center coordinate system; the second average value between the maximum and minimum values ​​of the coordinates in the vertical direction is calculated and used as the vertical axis coordinate in the three-dimensional center coordinate system; the third average value between the maximum and minimum values ​​of the coordinates in the vertical direction is calculated and used as the vertical axis coordinate in the three-dimensional center coordinate system.

[0009] Optionally, the steps for determining the actual length and actual width of the surface light source module include: The vertical projection range formed by the movement of spray droplets from an agricultural drone within a detection area and the light source spread coefficient are obtained, and the length and width of the vertical projection range are determined. Calculate the first sum between the value one and twice the light source expansion factor, and determine the actual length of the surface light source module based on the first product between the length of the vertical projection range and the first sum. The second sum between the determined value one and twice the light source expansion factor is calculated. Based on the second product between the width of the vertical projection range and the second sum, the actual width of the surface light source module is determined so that the actual coverage of the surface light source module completely covers the vertical projection area of ​​the fog droplet movement.

[0010] Optionally, the number of multi-vision industrial camera arrays in the image acquisition model is determined, wherein the number of industrial cameras required perpendicular to the flight direction of the agricultural drone is expressed as: , The number of industrial cameras required for the flight direction of an agricultural drone is expressed as follows: , in, This represents the number of industrial cameras required to achieve full field-of-view coverage in the horizontal dimension perpendicular to the flight direction of the agricultural drone. The number of industrial cameras required to achieve full field-of-view coverage in the horizontal dimension representing the flight direction of an agricultural drone; The horizontal width of the spraying area represents the maximum coverage width of the droplets in the direction perpendicular to the flight direction of the agricultural drone when it is spraying. The effective field of view of a single camera in the direction perpendicular to the flight direction of the agricultural drone represents the effective physical range in which a single industrial camera can clearly image in the horizontal dimension perpendicular to the flight direction. The effective field of view of a single camera in the flight direction of the agricultural drone represents the effective physical range in which a single industrial camera can clearly image along the horizontal dimension of the agricultural drone's flight direction. is the field-of-view overlap coefficient, which represents the proportion of overlap between the fields of view of adjacent cameras; The symbol indicates rounding up.

[0011] Optionally, a time correlation analysis is performed on the three-dimensional center coordinates of the droplet population in consecutive frame droplet images to construct a three-dimensional time series trajectory of the droplet population. The three-dimensional time series trajectory is then associated with its corresponding spatial dimension parameters and output as a curve, table, or three-dimensional dynamic model to complete the step of capturing the morphological trajectory of the spraying droplets by the agricultural drone. This includes: Temporal correlation analysis is performed on the three-dimensional center coordinate sequence of the fog droplet swarm stored in consecutive frames. The changes in the center position of the same fog droplet swarm are tracked by the inter-frame matching algorithm to construct the three-dimensional time series trajectory of the fog droplet swarm. The three-dimensional time series trajectory is associated and bound with the corresponding spatial dimension parameters, namely length, width, and height, to construct a data set containing spatiotemporal features; The dataset containing spatiotemporal features is output in the form of curves, tables, or three-dimensional dynamic models.

[0012] Optionally, the image segmentation algorithm includes a convolutional neural network or an edge detection algorithm; The three-dimensional reconstruction algorithm includes triangulation algorithm, stereo matching algorithm, or voxel space fusion algorithm.

[0013] A plant protection drone spraying droplet morphology trajectory capture device, provided for another purpose of this application, includes: The droplet image acquisition module is configured to start the surface light source module and make it emit light stably. It controls the agricultural drone to carry out spraying operations according to preset parameters. At the same time, it triggers the image acquisition module to continuously acquire multiple frames of droplet images according to the preset image acquisition frame rate to construct a droplet image sequence. The image acquisition module is a multi-vision industrial camera array constructed by multiple industrial cameras. The droplet image includes one or more droplet groups. The contour mask segmentation module is configured to call a preset image segmentation algorithm to perform image segmentation on the fog droplet image, extract the complete contour region of the fog droplet group in the fog droplet image from a single viewpoint, so as to determine the contour mask data corresponding to the fog droplet group. The three-dimensional contour construction module is configured to perform three-dimensional reconstruction on the contour mask data corresponding to the fog droplet group under each viewpoint according to the preset multi-view geometric constraints and the preset three-dimensional reconstruction algorithm, so as to generate the three-dimensional contour model of the fog droplet group. The spatial dimension determination module is configured to extract the bounding box of the fog droplet group in each frame of fog droplet image from the three-dimensional contour model, and calculate the spatial dimension parameters of the bounding box of the fog droplet group in a single frame of fog droplet image and its corresponding three-dimensional center coordinates. The droplet trajectory capture module is configured to perform time correlation analysis on the three-dimensional center coordinates of droplet groups in continuous frame droplet images to construct the three-dimensional time series trajectory of the droplet group, and output the three-dimensional time series trajectory in the form of curves, tables or three-dimensional dynamic models in association with its corresponding spatial size parameters to complete the morphological trajectory capture of the spraying droplets by the agricultural drone.

[0014] An electronic device provided for another purpose of this application includes a central processing unit and a memory, the central processing unit being configured to invoke and run a computer program stored in the memory to perform the steps of the method for capturing the spray droplet morphology trajectory of an agricultural drone as described in this application.

[0015] A computer-readable storage medium is provided for another purpose of this application, which stores, in the form of computer-readable instructions, a computer program implemented according to the method for capturing the morphology and trajectory of sprayed droplets by an agricultural drone, which, when called by a computer, executes the steps included in the corresponding method.

[0016] Compared to existing technologies, this application addresses the problems of existing agricultural drone spraying droplet morphology trajectory capture relying on complex laser imaging systems and high-precision algorithms, resulting in high equipment costs, and the susceptibility of traditional field trials to interference from environmental factors such as wind speed, temperature, and humidity, leading to poor data repeatability. This application offers the following advantages, including but not limited to: Firstly, this application is based on the collaborative design of a surface light source and a multi-view imaging system, which effectively resists the influence of environmental factors such as changes in natural light and airflow interference in the field. It solves the drawbacks of poor consistency and high uncertainty in measurement results of traditional methods, and can be widely adapted to different field operation scenarios, greatly improving its practicality.

[0017] Secondly, addressing the challenge of traditional feature extraction methods handling complex situations such as small droplet size, mutual occlusion, and dynamic movement, this application employs an image segmentation algorithm to achieve precise separation of droplets from the background, combined with multi-view 3D reconstruction technology to generate a complete droplet contour model. Based on the 3D center coordinates and length, width, and height spatial dimension parameters extracted from extreme coordinates, the simultaneous quantification of droplet group movement trajectory and morphological features is achieved, overcoming the shortcomings of existing technologies that lack detailed droplet movement trajectory and distribution information, and providing comprehensive data support for spraying effect evaluation.

[0018] Thirdly, addressing the issue that droplet motion is influenced by multiple factors such as wind speed and gravity, resulting in complex trajectories and the tendency of existing algorithms to mistrack or miss targets, this application employs an inter-frame matching algorithm to temporally correlate the three-dimensional center coordinates of consecutive frames, ensuring the continuity and uniqueness of the trajectories of the same droplet group. By binding the three-dimensional time-series trajectory with spatial dimensional parameters, a complete spatiotemporal feature dataset is formed, clearly presenting the droplet motion patterns and morphological changes. This provides a reliable basis for analyzing key indicators such as drift distance and diffusion degree, assisting in the dynamic optimization of spraying parameters for agricultural drones, improving pesticide utilization, and aligning with the development trend of precision agriculture.

[0019] Fourth, this application reduces computational power consumption while ensuring accuracy, meeting the real-time processing requirements of continuous frame images. It outputs data in multiple formats, including curves, tables, and 3D dynamic models, facilitating intuitive observation of trajectory and morphological change trends by technicians, while also providing precise quantitative data for scientific research analysis. Adaptable to diverse scenarios such as field operation debugging and laboratory parameter research, it achieves an organic balance between detection efficiency, data accuracy, and application flexibility, providing strong technical support for the quality control of agricultural drone spraying. Attached Figure Description

[0020] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is an exemplary architecture used in the plant protection drone spraying droplet morphology trajectory capture system in the embodiments of this application; Figure 2 This is a flowchart illustrating the method for capturing the morphology and trajectory of sprayed droplets from an agricultural drone in this embodiment of the application. Figure 3 This is a schematic diagram of the principle of the plant protection drone spraying droplet morphology trajectory capture device in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of the computer device in the embodiments of this application. Detailed Implementation

[0021] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0022] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0023] Those skilled in the art will understand that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0024] Those skilled in the art will understand that the terms "client," "terminal," and "terminal device" as used herein include both devices that receive wireless signals, devices that only possess wireless signal receiver capabilities without transmission capabilities, and devices with receiving and transmitting hardware, devices that have receiving and transmitting hardware capable of bidirectional communication over a bidirectional communication link. Such devices may include: cellular or other communication devices such as personal computers or tablets, having single-line displays, multi-line displays, or cellular or other communication devices without multi-line displays; PCS (Personal Communications Service) that can combine voice, data processing, fax, and / or data communication capabilities; PDAs (Personal Digital Assistants) that may include radio frequency receivers, pagers, internet / intranet access, web browsers, notebooks, calendars, and / or GPS (Global Positioning System) receivers; and conventional laptops and / or handheld computers or other devices that have and / or include radio frequency receivers. As used herein, "client," "terminal," and "terminal device" can be portable, transportable, installed in a means of transportation (air, sea, and / or land), or suitable and / or configured to operate locally and / or in a distributed manner, operating in any other location on Earth and / or in space. "Client," "terminal," and "terminal device" as used herein can also be a communication terminal, an internet access terminal, or a music / video playback terminal, such as a PDA, a MID (Mobile Internet Device), and / or a mobile phone with music / video playback capabilities, or a smart TV, set-top box, etc.

[0025] The hardware referred to by the names "server," "client," and "service node" in this application is essentially an electronic device with the equivalent capabilities of a personal computer. It is a hardware device with the necessary components revealed by the von Neumann architecture, such as a central processing unit (including an arithmetic logic unit and a control unit), memory, input devices, and output devices. The computer program is stored in its memory, and the central processing unit loads the program stored in the secondary storage into the main memory to run it, execute the instructions in the program, and interact with the input and output devices to complete specific functions.

[0026] It should be noted that the concept of "server" used in this application can also be extended to the case of server clusters. Based on the network deployment principles understood by those skilled in the art, the servers should be logically divided. Physically, these servers can be independent of each other but accessible through interfaces, or they can be integrated into a single physical computer or a computer cluster. Those skilled in the art should understand this flexibility and should not use it to constrain the implementation of the network deployment method in this application.

[0027] Unless otherwise expressly specified, one or more technical features of this application may be deployed on a server and accessed by a client through remote invocation of the online service interface provided by the server, or they may be directly deployed and run on a client for access.

[0028] Unless otherwise specified, the neural network models referenced or potentially referenced in this application may be deployed on a remote server and invoked remotely on the client, or deployed on a client with the capability to invoke directly. In some embodiments, when running on the client, the corresponding intelligence may be acquired through transfer learning in order to reduce the requirements on the client's hardware resources and avoid excessive consumption of the client's hardware resources.

[0029] Unless otherwise specified, all data involved in this application may be stored remotely on a server or on a local terminal device, as long as it is suitable for use by the technical solution of this application.

[0030] Those skilled in the art will understand that although the various methods in this application are described based on the same concept and thus present commonality among them, they can be performed independently unless otherwise specified. Similarly, the various embodiments disclosed in this application are all based on the same inventive concept; therefore, concepts expressed in the same way, as well as concepts that are appropriately changed for convenience but are expressed differently, should be understood equivalently.

[0031] Unless otherwise expressly stated, the various embodiments disclosed in this application can be combined in a cross-cutting manner to flexibly construct new embodiments, as long as such combination does not depart from the inventive spirit of this application and can meet the needs of the prior art or solve a certain deficiency in the prior art. Those skilled in the art should be aware of such modifications.

[0032] Please see Figure 1 The method for capturing the morphology and trajectory of spray droplets from agricultural drones in this application can be based on, for example... Figure 1The plant protection drone spraying droplet morphology trajectory capture system shown is implemented. The plant protection drone spraying droplet morphology trajectory capture system includes a surface light source module 1, an image acquisition module 2, a power supply module 3, a data transmission module 4, a data processing module, and a plant protection drone, etc. Each module is modularly assembled through standardized interfaces, which has the characteristics of clear connection logic and convenient assembly and disassembly.

[0033] In some embodiments, the surface light source module adopts a flat, integrated design, which can be laid flat on the ground in the area to be detected, with the emitting surface facing upwards. The overall size is designed according to the spraying range of the drone to ensure complete coverage of the vertical projection area of ​​the droplet movement. The light source adopts a high-uniformity light-emitting structure, and the light output is precisely adjusted through constant current drive or dimming circuit.

[0034] The image acquisition module employs a multi-view 3D imaging array composed of multiple industrial cameras to simultaneously acquire the true 3D morphology and 3D motion trajectory of droplet groups within the spraying area. Each industrial camera is adjusted for pitch, roll, and azimuth angles via an adjustable tripod and gimbal to achieve multi-angle, overlapping visual coverage of the spraying area. To ensure temporal consistency of the multi-view images, the industrial camera array achieves frame-level synchronization via a unified clock or hardware trigger signal.

[0035] The data transmission module is used to transmit image data from all cameras to the data processing module in real time, and can select wired (e.g., Ethernet) or wireless transmission schemes according to actual needs.

[0036] The data processing module can be a portable industrial computer, pre-installed with dedicated image processing software, supporting real-time data reception, processing and result display, and remote control of the image acquisition module's parameters and start / stop operations can be achieved through the software interface.

[0037] Please see Figure 2 In one embodiment of the method for capturing the morphology and trajectory of spray droplets from an agricultural drone, this application includes: Step S10: After starting the surface light source module and stabilizing its light emission, control the agricultural drone to perform spraying operations according to preset parameters. At the same time, trigger the image acquisition module to continuously acquire multiple frames of droplet images according to the preset image acquisition frame rate to construct a droplet image sequence. The image acquisition module is a multi-vision industrial camera array constructed from multiple industrial cameras. The droplet image includes one or more droplet groups. After activating the surface light source module in the agricultural drone spraying droplet morphology trajectory capture system and stabilizing its illumination, the system controls the agricultural drone to perform spraying operations according to preset parameters. Simultaneously, the image acquisition module is triggered to continuously acquire multiple frames of droplet images at a preset frame rate to construct a droplet image sequence. This sequence is then transmitted to the data processing module via the data transmission module. Preprocessing operations are performed on the multiple droplet images in the sequence, including grayscale conversion, denoising (e.g., median filtering), and contrast enhancement (e.g., histogram enhancement). The image acquisition module is a multi-vision industrial camera array composed of multiple industrial cameras, and the droplet images include one or more droplet groups. In some embodiments, the actual length of the surface light source module and actual width It is the length of the vertical projection range based on droplet motion. Hekuan It is determined, and the light source spread factor is set. External expansion, wherein the light source expansion coefficient It can take any value between 0.1 and 0.3.

[0038] In a specific embodiment, the steps of determining the actual length and actual width of the surface light source module include: Step S101: Obtain the vertical projection range formed by the spray droplets of the agricultural drone moving within the detection area and the light source spread coefficient, and determine the length and width of the vertical projection range; Step S102: Calculate the first sum between the determined value one and twice the light source expansion coefficient, and determine the actual length of the surface light source module based on the first product between the length of the vertical projection range and the first sum. The formula for calculating the actual length of the surface light source module is as follows: , in, Indicates the actual length of the surface light source module; This represents the light source spread factor, which can take any value between 0.1 and 0.3; This indicates the length of the vertical projection range formed by the movement of spray droplets from an agricultural drone within the detection area.

[0039] Step S103: Calculate the second sum between the determined value one and twice the light source expansion coefficient, and determine the actual width of the surface light source module based on the second product between the width of the vertical projection range and the second sum, so that the actual coverage of the surface light source module completely covers the vertical projection area of ​​the fog droplet movement.

[0040] The formula for calculating the actual width of the surface light source module is as follows: , in, This indicates the actual width of the surface light source module; This indicates the width of the vertical projection range formed by the movement of spray droplets from an agricultural drone within the detection area; This represents the light source expansion factor, which can take any value between 0.1 and 0.3.

[0041] In some embodiments, a multi-view industrial camera array is provided, the multi-view industrial camera array being configured according to the lateral width (perpendicular to the flight direction of the agricultural drone) of the spraying area. Coverage length along the direction of drone flight conduct The arrangement of the multi-vision industrial camera array in the image acquisition model is determined, where the number of industrial cameras required perpendicular to the flight direction of the agricultural drone is expressed as: , The number of industrial cameras required for the flight direction of an agricultural drone is expressed as follows: , in, This represents the number of industrial cameras required to achieve full field-of-view coverage in the horizontal dimension perpendicular to the flight direction of the agricultural drone. The number of industrial cameras required to achieve full field-of-view coverage in the horizontal dimension representing the flight direction of an agricultural drone; The horizontal width of the spraying area represents the maximum coverage width of the droplets in the direction perpendicular to the flight direction of the agricultural drone when it is spraying. The effective field of view of a single camera in the direction perpendicular to the flight direction of the agricultural drone represents the effective physical range in which a single industrial camera can clearly image in the horizontal dimension perpendicular to the flight direction. The effective field of view of a single camera in the flight direction of the agricultural drone represents the effective physical range in which a single industrial camera can clearly image along the horizontal dimension of the agricultural drone's flight direction. is the field-of-view overlap coefficient, which represents the proportion of overlap between the fields of view of adjacent cameras; The symbol indicates rounding up.

[0042] In some embodiments, setting image acquisition parameters requires setting the image acquisition frame rate. The image acquisition frame rate Based on the average velocity of the droplets The actual distance to a single pixel in the image It is determined that the following constraints must be satisfied, which are expressed as follows: , in, Image acquisition frame rate refers to the number of frames of an image captured by an industrial camera per unit of time. The average velocity of the droplets refers to the average speed at which the droplets move in three-dimensional space from the time of spraying to the time of settling. It represents the actual distance corresponding to a single pixel in the image, that is, the conversion factor for converting pixel scale to physical distance.

[0043] At the same time, set the image resolution. The image resolution Based on the actual size of the camera's field of view Actual distance from pixels It is determined that the following relationship must be satisfied: , Image resolution The settings must ensure that the minimum feature size of the fog droplets is represented by several pixels or more. Based on this, spatial calibration is performed. Multi-view images are acquired using a standard calibration board. Through a series of single-camera intrinsic parameter calibration, distortion correction, extrinsic parameter solving, and multi-camera joint calibration algorithms, a unified three-dimensional coordinate system for multiple cameras is established.

[0044] In some embodiments, the image segmentation algorithm includes a convolutional neural network or an edge detection algorithm; the 3D reconstruction algorithm includes a triangulation algorithm, a stereo matching algorithm, or a voxel space fusion algorithm.

[0045] Step S20: Call a preset image segmentation algorithm to segment the fog droplet image, extract the complete contour region of the fog droplet group in the fog droplet image from a single viewpoint, and determine the contour mask data corresponding to the fog droplet group; After activating the surface light source module and stabilizing its illumination, the agricultural drone is controlled to perform spraying operations according to preset parameters. Simultaneously, the image acquisition module is triggered to continuously acquire multiple frames of droplet images at a preset image acquisition frame rate to construct a droplet image sequence. Then, a preset image segmentation algorithm is called to segment the droplet images, extracting the complete contour region of the droplet group in the droplet image from a single viewpoint to determine the contour mask data corresponding to the droplet group. The image segmentation algorithm includes a convolutional neural network or an edge detection algorithm. In some embodiments, the step of calling a preset image segmentation algorithm to segment the fog droplet image and extracting the complete contour region of the fog droplet group in the fog droplet image from a single viewpoint to determine the contour mask data corresponding to the fog droplet group includes: Step S201: Input the fog image into an image segmentation algorithm that has been trained to convergence, and classify the pixel features of the fog droplet group and the background in the fog image based on the image segmentation algorithm, so as to separate the fog droplet group and the background in the fog image from a single viewpoint. Step S202: Based on the segmentation result of the image segmentation algorithm, extract the complete contour region of the fog droplet group to remove background interference pixels and isolated noise points, and obtain the contour region data corresponding to the fog droplet group under a single view. Step S203: Binarize the contour region data corresponding to the fog droplet group, mark the contour region data as valid foreground pixels and mark the background region as invalid pixels, and generate and determine the contour mask data corresponding to the fog droplet group.

[0046] As can be seen from steps S201 to S203 above, the image segmentation algorithm based on training to convergence and based on pixel feature classification can accurately distinguish between fog droplet groups and complex backgrounds, solve the target confusion problem caused by interference such as natural light in the field and equipment noise, and the segmentation accuracy meets the needs of agricultural scenarios.

[0047] By extracting the complete contour region and removing background interference pixels and isolated noise, invalid information is avoided from affecting subsequent processing, while the boundary details of the fog droplet group are preserved, providing high-quality basic data for 3D reconstruction.

[0048] The contour mask data generated by binarization simplifies the data form by clearly marking valid and invalid pixels. This not only adapts to 3D reconstruction algorithms under multi-view geometric constraints (such as triangulation and voxel fusion), but also improves the efficiency and accuracy of subsequent inter-frame matching and coordinate calculation, laying a solid foundation for the accurate acquisition of the 3D trajectory and morphological parameters of the fog droplet population.

[0049] Step S30: Based on preset multi-view geometric constraints, perform three-dimensional reconstruction on the contour mask data corresponding to the fog droplet group under each view according to the preset three-dimensional reconstruction algorithm to generate a three-dimensional contour model of the fog droplet group. A preset image segmentation algorithm is invoked to segment the fog droplet image, extracting the complete contour region of the fog droplet group in the image from a single viewpoint. After determining the contour mask data corresponding to the fog droplet group, based on preset multi-view geometric constraints, a preset 3D reconstruction algorithm is used to perform 3D reconstruction on the contour mask data corresponding to the fog droplet group from each viewpoint to generate a 3D contour model of the fog droplet group. The multi-view geometric constraints refer to a series of geometric rules imposed on the contour data from each viewpoint during the 3D reconstruction of the fog droplet group, based on the spatial positional relationship, imaging principle, and unified coordinate system of multiple industrial cameras. The core purpose is to ensure that the multi-view data can be accurately fused into a realistic 3D contour model. The 3D reconstruction algorithm includes triangulation algorithms, stereo matching algorithms, or voxel space fusion algorithms, etc.

[0050] Step S40: Extract the bounding box of the fog droplet group in each frame of the fog droplet image from the three-dimensional contour model, and calculate the spatial size parameters of the bounding box of the fog droplet group in a single frame of the fog droplet image and its corresponding three-dimensional center coordinates; Based on preset multi-view geometric constraints, the contour mask data corresponding to the fog droplet group under each viewpoint is reconstructed in three dimensions according to a preset three-dimensional reconstruction algorithm to generate a three-dimensional contour model of the fog droplet group. Then, the bounding box of the fog droplet group in each frame of the fog droplet image is extracted from the three-dimensional contour model, and the spatial size parameters of the bounding box of the fog droplet group in a single frame of the fog droplet image and its corresponding three-dimensional center coordinates are calculated. The spatial size parameters include the length a, the width b, and the height c of the bounding box of the fog droplet group. The three-dimensional center coordinates are represented as (x, y, z). The parameters of the fog droplet group in a single frame of the fog droplet image are defined as a six-dimensional spatial data format P=(x, y, z, a, b, c).

[0051] In some embodiments, the step of extracting the bounding box of the fog droplet group in each frame of the fog droplet image from the three-dimensional contour model, and calculating the spatial size parameters of the bounding box of the fog droplet group in a single frame of the fog droplet image and its corresponding three-dimensional center coordinates includes: Step S401: For the three-dimensional contour model of the fog droplet group in a single-frame fog droplet image, traverse all three-dimensional pixels in the three-dimensional contour model to determine the extreme coordinates of the fog droplet group in three-dimensional space. The extreme coordinates include the maximum and minimum coordinates in the horizontal axis direction, the maximum and minimum coordinates in the vertical axis direction, and the maximum and minimum coordinates in the vertical direction. Step S402: Construct the minimum bounding box of the fog droplet group based on the extreme coordinates, so that the minimum bounding box completely encloses the three-dimensional outline of the fog droplet group, and each face of the minimum bounding box is parallel to the coordinate axes of the multi-camera unified three-dimensional coordinate system. Step S403: Based on the three-dimensional space of the fog droplet group, the first difference between the maximum value of the horizontal axis coordinate and the corresponding minimum value is used as the length parameter in the spatial dimension parameters; the second difference between the maximum value of the vertical axis coordinate and the corresponding minimum value is used as the width parameter in the spatial dimension parameters; and the third difference between the maximum value of the vertical axis coordinate and the corresponding minimum value is used as the height parameter in the spatial dimension parameters. Step S404: Calculate and determine the first average value between the maximum value and the corresponding minimum value of the coordinate in the horizontal axis direction, and use it as the horizontal axis coordinate in the three-dimensional center coordinate system; calculate and determine the second average value between the maximum value and the corresponding minimum value of the coordinate in the vertical axis direction, and use it as the vertical axis coordinate in the three-dimensional center coordinate system; calculate and determine the third average value between the maximum value and the corresponding minimum value of the coordinate in the vertical axis direction, and use it as the vertical axis coordinate in the three-dimensional center coordinate system.

[0052] As can be seen from steps S201 to S203 above, by traversing the three-dimensional pixels to obtain the three-dimensional extreme coordinates, it is ensured that the smallest bounding box can completely enclose the fog droplet group, and that each face is parallel to a unified coordinate system, thus avoiding coordinate deviation. The size parameters and center coordinates calculated based on the extreme value difference and the average value are directly related to the real physical space, resulting in high data accuracy and closely matching the actual spatial state of the fog droplet group.

[0053] Constructing bounding boxes and deriving parameters based on extreme coordinates eliminates the need for complex iterative calculations, significantly reducing computational complexity and adapting to the real-time processing requirements of continuous frame images. Simultaneously, the unified computational logic ensures standardized parameter extraction, avoiding computational biases caused by irregular droplet morphology.

[0054] The output length, width, and height dimensions, along with the three-dimensional center coordinates, are core quantitative indicators of the spatiotemporal characteristics of the droplet population. The spatial dimensions directly reflect the degree of droplet population diffusion, while the center coordinates provide a stable tracking target for inter-frame temporal correlation analysis. Together, they provide crucial data support for the construction of three-dimensional time series trajectories and the optimization of spraying parameters.

[0055] Step S50: Perform time correlation analysis on the three-dimensional center coordinates of the droplet group in the continuous frame droplet images to construct the three-dimensional time series trajectory of the droplet group, and output the three-dimensional time series trajectory in the form of curves, tables or three-dimensional dynamic models in association with its corresponding spatial size parameters to complete the morphological trajectory capture of the spraying droplets by the agricultural drone.

[0056] After extracting the bounding box of the droplet group in each frame of the droplet image from the three-dimensional contour model, calculating the spatial size parameters of the bounding box of the droplet group in a single frame of the droplet image and its corresponding three-dimensional center coordinates, performing time correlation analysis on the three-dimensional center coordinates of the droplet group in consecutive frames of droplet images to construct the three-dimensional time series trajectory of the droplet group, and outputting the three-dimensional time series trajectory in association with its corresponding spatial size parameters in the form of curves, tables or three-dimensional dynamic models to complete the morphological trajectory capture of the spraying droplets by the agricultural drone.

[0057] In some embodiments, a time correlation analysis is performed on the three-dimensional center coordinates of the droplet population in consecutive frame droplet images to construct a three-dimensional time series trajectory of the droplet population. The three-dimensional time series trajectory is then associated with its corresponding spatial dimension parameters and output as a curve, table, or three-dimensional dynamic model to complete the step of capturing the morphological trajectory of the sprayed droplets by the agricultural drone. This includes: Step S5001: Perform time correlation analysis on the three-dimensional center coordinate sequence of the fog droplet group stored in consecutive frames, track the changes in the center position of the same fog droplet group through the inter-frame matching algorithm, and construct the three-dimensional time series trajectory of the fog droplet group; Step S5002: Associate and bind the three-dimensional time series trajectory with the corresponding spatial dimension parameters, including the length, width, and height parameters, to construct a data set containing spatiotemporal features; Step S5003: Output the dataset containing spatiotemporal features in the form of curves, tables, or three-dimensional dynamic models.

[0058] As can be seen from steps S5001 to S5003 above, by using the inter-frame matching algorithm to correlate the three-dimensional center coordinates of consecutive frames in time, the motion trajectory of the same group of fog droplets can be accurately locked, avoiding confusion among multiple groups or mismatch between frames. This solves the problems of trajectory breakage and inaccurate tracking in the existing technology, and ensures the continuity and authenticity of the three-dimensional time series trajectory.

[0059] By binding the three-dimensional trajectory with spatial dimensional parameters of length, width, and height, a dataset containing spatiotemporal characteristics is constructed, which not only reflects the dynamic movement law of the fog droplet population but also reflects its morphological change characteristics. This makes up for the limitations of single trajectory or morphological data and achieves a comprehensive characterization of the fog droplet population.

[0060] Data is output in multiple formats, including curves, tables, and 3D dynamic models. Curves can intuitively show the trajectory and shape change trend, tables facilitate quantitative analysis and data archiving, and 3D dynamic models can three-dimensionally reproduce the motion process. It is suitable for different scenarios such as field operation debugging of agricultural drones and laboratory parameter research, improving the practicality and operability of the data and providing direct data support for the evaluation of the spraying effect and parameter optimization of agricultural drones.

[0061] In some embodiments, a 2.2m × 1.2m surface light source module is first laid flat on the ground of the work area. This size is set based on the vertical projection range of 1.6m × 1.0m for the drone spraying, and the length and width dimensions are expanded outwards according to a light source expansion factor n = 0.15. Next, a three-view imaging array is formed using three industrial cameras. Each camera is mounted on a tripod at a height of 1.8m, and the pitch and azimuth angles of the cameras are adjusted via a gimbal. The three cameras are arranged horizontally at a spacing of 0.6m and connected to a data processing module via a synchronization trigger line to achieve frame-level synchronous acquisition. In the data processing module, the image acquisition frame rate is further set to n = 120fps, and the image resolution is set to m = 2048 × 2048 pixels. Based on this, a standard calibration board is used to sequentially complete the single-camera calibration, distortion correction, extrinsic parameter calibration, and multi-camera joint calibration of the three cameras to establish a unified three-dimensional reconstruction coordinate system.

[0062] After the device is deployed and calibrated, the agricultural drone is controlled to fly over the surface light source area at a height of 2.5m and start the spraying module. After the surface light source is started and emitting stable light, the data processing module triggers the three cameras to simultaneously acquire 1500 frames of droplet image sequence. Subsequently, the three acquired images are imported into the data processing module and preprocessed sequentially, including grayscale conversion, median filtering for noise reduction, and histogram stretching enhancement. Further, the outline of the droplet group is extracted using a method combining edge detection and morphological closing operation, and the outline mask is saved according to frame number. Based on the extracted outline, a three-view voxel space fusion method is used to generate an instantaneous three-dimensional voxel model for each frame. Then, voxel filtering is used to process noise, and the bounding box size (a,b,c) and three-dimensional center coordinates (x,y,z) of the droplet group are calculated based on voxel points. The data of each frame is saved in vector format P=(x,y,z,a,b,c). Finally, a time-series correlation was performed on the center coordinates of the 1500 consecutive frames to obtain the three-dimensional motion trajectory T(t) of the droplet swarm. The trajectory file and parameter table were output in CSV format and stored on an external storage device.

[0063] In some embodiments, a fixed spraying test platform was constructed in a greenhouse environment. This embodiment uses an imaging array composed of four-view high-resolution phased-field arrays to acquire the three-dimensional morphology and motion trajectory of droplets.

[0064] In this embodiment, a surface light source module with dimensions of 1.8m × 1.8m is first placed on a hard surface. The light source is powered by a constant current drive and its brightness is adjusted by a dimming circuit to achieve the best backlight effect. Next, a four-view array is formed using four 5-megapixel industrial cameras, each mounted on a 2.0m high bracket. Each camera is equipped with a 12mm fixed-focus lens, and its tilt angle is adjusted by a gimbal. The four cameras are synchronized via a unified clock signal and connected to a data processing module via a wireless data transmission terminal. Furthermore, the acquisition frame rate is set to n = 200fps, and the image resolution is set to m = 2592 × 2592 pixels. The intrinsic parameter calibration, distortion correction, and extrinsic parameter calculation of the four cameras are performed using a calibration board, and a unified three-dimensional coordinate system is established using a multi-camera joint calibration method.

[0065] Subsequently, the agricultural drone was fixed on the spraying test frame and set to constant spray volume mode. The four cameras simultaneously started synchronous acquisition, obtaining a total of 2000 image frames. After inputting the four images, grayscale conversion, Gaussian filtering, and CLAHE local contrast enhancement were performed sequentially. Further, a convolutional neural network segmentation algorithm was used to obtain the droplet population contour, supplemented by morphological opening operations to remove isolated noise points. A dense voxel model was formed using the four contour input voxel space fusion module. The bounding box size (a, b, c) and three-dimensional center coordinates (x, y, z) of the droplet population were calculated. All frames were stored in the vector format P=(x, y, z, a, b, c). Finally, the center coordinates of the 2000 frames were correlated in chronological order to generate a three-dimensional trajectory T(t). The trajectory file and the three-dimensional dynamic visualization animation were stored in a structured format on an external storage device.

[0066] As can be seen from the above embodiments, compared with the prior art, the present application addresses the problems of existing technologies, such as the reliance on complex laser imaging systems and high-precision algorithms for capturing the morphology and trajectory of spray droplets from agricultural drones, resulting in high equipment costs, and the susceptibility of traditional field trials to interference from environmental factors such as wind speed, temperature, and humidity, leading to poor data repeatability. The present application provides the following beneficial effects, including but not limited to: Firstly, this application is based on the collaborative design of a surface light source and a multi-view imaging system, which effectively resists the influence of environmental factors such as changes in natural light and airflow interference in the field. It solves the drawbacks of poor consistency and high uncertainty in measurement results of traditional methods, and can be widely adapted to different field operation scenarios, greatly improving its practicality.

[0067] Secondly, addressing the challenge of traditional feature extraction methods handling complex situations such as small droplet size, mutual occlusion, and dynamic movement, this application employs an image segmentation algorithm to achieve precise separation of droplets from the background, combined with multi-view 3D reconstruction technology to generate a complete droplet contour model. Based on the 3D center coordinates and length, width, and height spatial dimension parameters extracted from extreme coordinates, the simultaneous quantification of droplet group movement trajectory and morphological features is achieved, overcoming the shortcomings of existing technologies that lack detailed droplet movement trajectory and distribution information, and providing comprehensive data support for spraying effect evaluation.

[0068] Thirdly, addressing the issue that droplet motion is influenced by multiple factors such as wind speed and gravity, resulting in complex trajectories and the tendency of existing algorithms to mistrack or miss targets, this application employs an inter-frame matching algorithm to temporally correlate the three-dimensional center coordinates of consecutive frames, ensuring the continuity and uniqueness of the trajectories of the same droplet group. By binding the three-dimensional time-series trajectory with spatial dimensional parameters, a complete spatiotemporal feature dataset is formed, clearly presenting the droplet motion patterns and morphological changes. This provides a reliable basis for analyzing key indicators such as drift distance and diffusion degree, assisting in the dynamic optimization of spraying parameters for agricultural drones, improving pesticide utilization, and aligning with the development trend of precision agriculture.

[0069] Fourth, this application reduces computational power consumption while ensuring accuracy, meeting the real-time processing requirements of continuous frame images. It outputs data in multiple formats, including curves, tables, and 3D dynamic models, facilitating intuitive observation of trajectory and morphological change trends by technicians, while also providing precise quantitative data for scientific research analysis. Adaptable to diverse scenarios such as field operation debugging and laboratory parameter research, it achieves an organic balance between detection efficiency, data accuracy, and application flexibility, providing strong technical support for the quality control of agricultural drone spraying.

[0070] Please see Figure 3This application provides a device for capturing the morphology and trajectory of spraying droplets from an agricultural drone, comprising a droplet image acquisition module 1100, a contour mask segmentation module 1200, a 3D contour construction module 1300, a spatial size determination module 1400, and a droplet trajectory capture module 1500. The droplet image acquisition module 1100 is configured to activate a surface light source module and stabilize its illumination, then control the agricultural drone to perform spraying operations according to preset parameters. Simultaneously, it triggers the image acquisition module to continuously acquire multiple frames of droplet images at a preset image acquisition frame rate to construct a droplet image sequence. The image acquisition module is a multi-view industrial camera array composed of multiple industrial cameras, and the droplet images include one or more droplet groups. The contour mask segmentation module 1200 is configured to call a preset image segmentation algorithm to segment the droplet images, extracting the complete contour region of the droplet group in the droplet image from a single viewpoint to determine the contour mask data corresponding to the droplet group. The 3D contour construction module 1300 is configured to, based on preset multi-view geometric constraints, construct a 3D contour based on preset 3D contours. The algorithm performs 3D reconstruction on the contour mask data corresponding to the droplet group from various perspectives to generate a 3D contour model of the droplet group. The spatial size determination module 1400 is configured to extract the bounding box of the droplet group in each frame of the droplet image from the 3D contour model, and calculate the spatial size parameters of the bounding box of the droplet group in a single frame of the droplet image and its corresponding 3D center coordinates. The droplet trajectory capture module 1500 is configured to perform temporal correlation analysis on the 3D center coordinates of the droplet group in consecutive frames of droplet images to construct the 3D time series trajectory of the droplet group, and output the 3D time series trajectory in association with its corresponding spatial size parameters in the form of curves, tables or 3D dynamic models to complete the morphological trajectory capture of the sprayed droplets by the agricultural drone.

[0071] Based on any embodiment of this application, please refer to Figure 4 Another embodiment of this application also provides an electronic device, which can be implemented by a computer device, such as... Figure 4The diagram shows the internal structure of a computer device. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected via a system bus. The computer-readable storage medium stores an operating system, a database, and computer-readable instructions. The database may store control information sequences. When the computer-readable instructions are executed by the processor, the processor can implement a method for capturing the morphology and trajectory of spraying droplets from an agricultural drone. The processor of the computer device provides computing and control capabilities, supporting the operation of the entire computer device. The memory of the computer device may store computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor can execute the method for capturing the morphology and trajectory of spraying droplets from an agricultural drone according to this application. The network interface of the computer device is used for communication with a terminal. Those skilled in the art will understand that… Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0072] In this embodiment, the processor is used to execute... Figure 3 The memory stores the specific functions of each module, and stores the program code and various data required to execute the above modules. The network interface is used for data transmission between the user terminal and the server. In this embodiment, the memory stores the program code and data required to execute all modules in the agricultural drone spraying droplet morphology trajectory capturing device of this application, and the server can call the server's program code and data to execute the functions of all modules.

[0073] This application also provides a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the method for capturing the morphology and trajectory of spraying droplets by an agricultural drone as described in any embodiment of this application.

[0074] This application also provides a computer program product, including a computer program / instructions that, when executed by one or more processors, implement the steps of the method for capturing the morphology and trajectory of spraying droplets from an agricultural drone as described in any embodiment of this application.

[0075] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. This computer program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0076] The above description is only a partial embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for capturing the morphology and trajectory of spray droplets from an agricultural protection drone, characterized in that, include: After the surface light source module is activated and stabilized to emit light, the agricultural drone is controlled to perform spraying operations according to preset parameters. At the same time, the image acquisition module is triggered to continuously acquire multiple frames of droplet images according to the preset image acquisition frame rate to construct a droplet image sequence. The image acquisition module is a multi-vision industrial camera array constructed from multiple industrial cameras, and the droplet image includes one or more droplet groups. A preset image segmentation algorithm is invoked to segment the fog droplet image, and the complete contour region of the fog droplet group in the fog droplet image under a single view is extracted to determine the contour mask data corresponding to the fog droplet group. Based on preset multi-view geometric constraints, the contour mask data corresponding to the fog droplet group under each viewpoint is reconstructed in three dimensions according to the preset three-dimensional reconstruction algorithm to generate a three-dimensional contour model of the fog droplet group. Extract the bounding box of the fog droplet group in each frame of the fog droplet image from the three-dimensional contour model, and calculate the spatial size parameters of the bounding box of the fog droplet group in a single frame of the fog droplet image and its corresponding three-dimensional center coordinates. Temporal correlation analysis is performed on the three-dimensional center coordinates of the droplet population in consecutive frame droplet images to construct the three-dimensional time series trajectory of the droplet population. The three-dimensional time series trajectory is then associated with its corresponding spatial dimension parameters and output in the form of curves, tables, or three-dimensional dynamic models to complete the morphological trajectory capture of the sprayed droplets by the agricultural drone.

2. The method for capturing the morphology and trajectory of spray droplets from an agricultural drone according to claim 1, characterized in that, The steps of using a preset image segmentation algorithm to segment the fog droplet image and extracting the complete contour region of the fog droplet group in the fog droplet image from a single viewpoint to determine the contour mask data corresponding to the fog droplet group include: The fog image is input into an image segmentation algorithm that has been trained to convergence. Based on the image segmentation algorithm, the pixel features of the fog droplets and the background in the fog image are classified to separate the fog droplets and the background in the fog image from a single viewpoint. Based on the segmentation results of the image segmentation algorithm, the complete contour region of the fog droplet group is extracted to remove background interference pixels and isolated noise points, and the contour region data corresponding to the fog droplet group under a single view is obtained. The contour region data corresponding to the fog droplet group is binarized, and the contour region data is marked as valid foreground pixels and the background region is marked as invalid pixels, thereby generating and determining the contour mask data corresponding to the fog droplet group.

3. The method for capturing the morphology and trajectory of spray droplets from an agricultural drone according to claim 1, characterized in that, The steps of extracting the bounding box of the fog droplet group in each frame of the fog droplet image from the three-dimensional contour model, and calculating the spatial size parameters of the bounding box of the fog droplet group in a single frame of the fog droplet image and its corresponding three-dimensional center coordinates, include: For a three-dimensional contour model of a group of fog droplets in a single-frame fog droplet image, all three-dimensional pixels in the three-dimensional contour model are traversed to determine the extreme coordinates of the fog droplet group in three-dimensional space. The extreme coordinates include the maximum and minimum coordinates in the horizontal axis direction, the maximum and minimum coordinates in the vertical axis direction, and the maximum and minimum coordinates in the vertical direction. Based on the extreme coordinates, a minimum bounding box is constructed for the fog droplet population so that the minimum bounding box completely encloses the three-dimensional outline of the fog droplet population, and each face of the minimum bounding box is parallel to the coordinate axes of the multi-camera unified three-dimensional coordinate system. Based on the three-dimensional space of the fog droplet group, the first difference between the maximum value of the horizontal axis coordinate and the corresponding minimum value is used as the length parameter in the spatial dimension parameters; the second difference between the maximum value of the vertical axis coordinate and the corresponding minimum value is used as the width parameter in the spatial dimension parameters; and the third difference between the maximum value of the vertical axis coordinate and the corresponding minimum value is used as the height parameter in the spatial dimension parameters. The first average value between the maximum and minimum values ​​of the coordinates in the horizontal direction is calculated and used as the horizontal axis coordinate in the three-dimensional center coordinate system; the second average value between the maximum and minimum values ​​of the coordinates in the vertical direction is calculated and used as the vertical axis coordinate in the three-dimensional center coordinate system; the third average value between the maximum and minimum values ​​of the coordinates in the vertical direction is calculated and used as the vertical axis coordinate in the three-dimensional center coordinate system.

4. The method for capturing the morphology and trajectory of spray droplets from an agricultural drone according to claim 1, characterized in that, The steps for determining the actual length and actual width of the surface light source module include: The vertical projection range formed by the movement of spray droplets from an agricultural drone within a detection area and the light source spread coefficient are obtained, and the length and width of the vertical projection range are determined. Calculate the first sum between the value one and twice the light source expansion factor, and determine the actual length of the surface light source module based on the first product between the length of the vertical projection range and the first sum. The second sum between the determined value one and twice the light source expansion factor is calculated. Based on the second product between the width of the vertical projection range and the second sum, the actual width of the surface light source module is determined so that the actual coverage of the surface light source module completely covers the vertical projection area of ​​the fog droplet movement.

5. The method for capturing the morphology and trajectory of spray droplets from an agricultural drone according to claim 1, characterized in that, The number of industrial cameras arranged in the multi-vision industrial camera array in the image acquisition model is determined, where the number of industrial cameras required perpendicular to the flight direction of the agricultural drone is expressed as: ， The number of industrial cameras required for the flight direction of an agricultural drone is expressed as follows: ， in, This represents the number of industrial cameras required to achieve full field-of-view coverage in the horizontal dimension perpendicular to the flight direction of the agricultural drone. The number of industrial cameras required to achieve full field-of-view coverage in the horizontal dimension representing the flight direction of an agricultural drone; The horizontal width of the spraying area represents the maximum coverage width of the droplets in the direction perpendicular to the flight direction of the agricultural drone when it is spraying. The effective field of view of a single camera in the direction perpendicular to the flight direction of the agricultural drone represents the effective physical range in which a single industrial camera can clearly image in the horizontal dimension perpendicular to the flight direction. The effective field of view of a single camera in the flight direction of the agricultural drone represents the effective physical range in which a single industrial camera can clearly image along the horizontal dimension of the agricultural drone's flight direction. is the field-of-view overlap coefficient, which represents the proportion of overlap between the fields of view of adjacent cameras; The symbol indicates rounding up.

6. The method for capturing the morphology and trajectory of spray droplets from an agricultural drone according to claim 1, characterized in that, The process involves performing temporal correlation analysis on the three-dimensional center coordinates of droplet ensembles in consecutive frame droplet images to construct a three-dimensional time-series trajectory of the droplet ensemble. This trajectory is then associated with its corresponding spatial dimension parameters and output as a curve, table, or three-dimensional dynamic model. This completes the step of capturing the morphological trajectory of droplets sprayed by an agricultural drone, including: Temporal correlation analysis is performed on the three-dimensional center coordinate sequence of the fog droplet swarm stored in consecutive frames. The changes in the center position of the same fog droplet swarm are tracked by the inter-frame matching algorithm to construct the three-dimensional time series trajectory of the fog droplet swarm. The three-dimensional time series trajectory is associated and bound with the corresponding spatial dimension parameters, namely length, width, and height, to construct a dataset containing spatiotemporal features; The dataset containing spatiotemporal features is output in the form of curves, tables, or three-dimensional dynamic models.

7. The method for capturing the morphology and trajectory of spray droplets from an agricultural drone according to any one of claims 1 to 6, characterized in that, The image segmentation algorithm includes a convolutional neural network or an edge detection algorithm; The three-dimensional reconstruction algorithm includes triangulation algorithm, stereo matching algorithm, or voxel space fusion algorithm.

8. A device for capturing the morphology and trajectory of spray droplets from an agricultural drone, characterized in that, include: The droplet image acquisition module is configured to start the surface light source module and make it emit light stably. It controls the agricultural drone to carry out spraying operations according to preset parameters. At the same time, it triggers the image acquisition module to continuously acquire multiple frames of droplet images according to the preset image acquisition frame rate to construct a droplet image sequence. The image acquisition module is a multi-vision industrial camera array constructed by multiple industrial cameras. The droplet image includes one or more droplet groups. The contour mask segmentation module is configured to call a preset image segmentation algorithm to perform image segmentation on the fog droplet image, extract the complete contour region of the fog droplet group in the fog droplet image from a single viewpoint, so as to determine the contour mask data corresponding to the fog droplet group. The three-dimensional contour construction module is configured to perform three-dimensional reconstruction on the contour mask data corresponding to the fog droplet group under each viewpoint according to the preset multi-view geometric constraints and the preset three-dimensional reconstruction algorithm, so as to generate the three-dimensional contour model of the fog droplet group. The spatial dimension determination module is configured to extract the bounding box of the fog droplet group in each frame of fog droplet image from the three-dimensional contour model, and calculate the spatial dimension parameters of the bounding box of the fog droplet group in a single frame of fog droplet image and its corresponding three-dimensional center coordinates. The droplet trajectory capture module is configured to perform time correlation analysis on the three-dimensional center coordinates of droplet groups in continuous frame droplet images to construct the three-dimensional time series trajectory of the droplet group, and output the three-dimensional time series trajectory in the form of curves, tables or three-dimensional dynamic models in association with its corresponding spatial size parameters to complete the morphological trajectory capture of the spraying droplets by the agricultural drone.

9. An electronic device comprising a central processing unit and a memory, characterized in that, The central processing unit is used to invoke and run a computer program stored in the memory to perform the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores, in the form of computer-readable instructions, a computer program implemented according to any one of claims 1 to 7, which, when invoked by a computer, executes the steps included in the corresponding method.

Images