Intelligent control method for pesticide spraying of crops

Abstract

This invention relates to the field of pesticide spraying technology, specifically to an intelligent control method for pesticide spraying of crops. The method includes: acquiring crop images within a pesticide spraying range collected by a drone at multiple sampling times; for each sampling time, determining the crop's growth characteristics and pest / disease impact characteristics based on the crop images, whereby the growth characteristics characterize the degree of crop structural occlusion within the corresponding area of ​​the crop image, and the pest / disease impact characteristics characterize the severity of pest / disease impact on the crop within the corresponding area of ​​the crop image; for each sampling time, determining the drone's flight control parameters based on the growth characteristics and pest / disease impact characteristics; and controlling the drone to fly along a preset flight path based on the flight control parameters. This invention solves the problem of insufficient pesticide spraying by drones in areas with severe occlusion or severe pest / disease damage.

Description

Technical Field

[0001] This invention relates to the field of pesticide spraying technology, and more specifically to an intelligent control method for pesticide spraying of crops. Background Technology

[0002] In agricultural production, pesticide spraying is an important means of preventing and controlling pests and diseases and ensuring the healthy growth of crops. Traditional spraying methods mainly rely on manual labor or mechanized equipment, which have problems such as low efficiency, uneven coverage, and high labor costs.

[0003] With the development of drone technology, using drones for pesticide spraying has become an important direction in modern agriculture. It has advantages such as high operating efficiency, adaptability to complex terrain, and large-scale operation, which can effectively improve spraying efficiency and pesticide utilization rate, and alleviate the labor pressure in agricultural plant protection.

[0004] However, existing drone spraying technologies mostly adopt a uniform spraying mode, which does not fully consider the structural differences in crop growth and the uneven distribution of pests and diseases. This results in insufficient spraying in areas that are severely covered or heavily affected by pests and diseases, thus affecting the control effect. Summary of the Invention

[0005] To address the technical problem of insufficient pesticide spraying by drones in areas with severe obstruction or heavy pest and disease infestation, the present invention aims to provide an intelligent control method for pesticide spraying of crops. The specific technical solution adopted is as follows: Firstly, a method for intelligent control of pesticide spraying on crops is provided. This method includes: acquiring crop images within a pesticide spraying range collected by a drone at multiple sampling times; for each sampling time, determining crop growth characteristics and pest / disease impact characteristics based on the crop images, whereby the growth characteristics characterize the degree of crop structure occlusion within the corresponding area of ​​the crop image, and the pest / disease impact characteristics characterize the severity of pest / disease impact on the crop within the corresponding area of ​​the crop image; for each sampling time, determining drone flight control parameters based on the growth characteristics and pest / disease impact characteristics, whereby the flight control parameters indicate the duration for which the drone's pesticide spraying range covers the location corresponding to the sampling time; and controlling the drone to fly along a preset flight path based on the flight control parameters.

[0006] In one possible design, the above-mentioned determination of crop growth characteristics based on crop images includes: identifying multiple closed regions in the crop image, where each closed region is a region of the crop in the crop image; determining the structural density of the crop in the crop image based on the number of multiple closed regions and the area of ​​the crop region in the crop image; determining the area variance of the multiple closed regions, where the area variance is used to characterize the growth of the crop in the corresponding region of the crop image; and determining the growth characteristics based on the structural density and the area variance.

[0007] In one possible design, the determination of crop pest and disease impact characteristics based on crop images includes: identifying multiple closed regions in the crop image, where each closed region represents a region of the crop within the image; for each closed region, determining its area and average gray value, where the average gray value characterizes the degree of pest and disease impact on the corresponding crop; determining pesticide demand based on the area and average gray value, where the pesticide demand characterizes the degree of pesticide demand for the closed region; determining the relative position of each closed region within the pesticide spraying range, where the relative position characterizes the position of the closed region relative to the drone's flight direction; determining a positional demand weight based on the relative position and the diameter of the pesticide spraying range, where the positional demand weight characterizes the degree of demand for a reduction in the drone's flight speed for the closed region; and determining the pest and disease impact characteristics based on the pesticide demand and positional demand weight for each closed region among the multiple closed regions.

[0008] In one possible design, the above-mentioned determination of the UAV's flight control parameters based on growth characteristics and pest and disease impact characteristics includes: for each sampling time, determining the dwell time weight based on growth characteristics and pest and disease impact characteristics; and determining the flight control parameters corresponding to each sampling time based on the dwell time weights of multiple sampling times and the preset dwell time.

[0009] In one possible design, controlling the UAV to fly along a preset flight path according to flight control parameters includes: determining the target flight speed corresponding to each position of the UAV on the preset flight path according to the flight control parameters; and controlling the UAV to fly at the target flight speed corresponding to each position on the preset flight path.

[0010] In one possible design, determining multiple closed regions in a crop image includes: processing the crop image based on a preset edge detection algorithm to determine the edge image of each crop in the crop image; and fitting the edge image of each crop to obtain multiple closed regions.

[0011] In one possible design, determining the relative position of each closed region within the pesticide spraying range includes: for each closed region, determining the projected position of the centroid of the closed region in the direction of the drone's flight, and the projected distance between the projected position and the drone; if the projected position is in front of the drone in the direction of the drone's flight, determining the relative position as the sum of the radius of the pesticide spraying range and the projected distance; if the projected position is behind the drone in the direction of the drone's flight, determining the relative position as the difference between the radius of the pesticide spraying range and the projected distance.

[0012] In one possible design, determining the flight control parameters corresponding to each sampling moment based on the dwell time weights of multiple sampling moments and a preset dwell time includes: determining a weight reference value based on the dwell time weights of multiple sampling moments, wherein the weight reference value is the average of the dwell time weights of multiple sampling moments, or the average of the maximum and minimum values ​​among the dwell time weights of multiple sampling moments; and determining the flight control parameters corresponding to each sampling moment based on the ratio of the dwell time weight to the weight reference value and the preset dwell time.

[0013] In one possible design, the drone flies at a fixed altitude, and the amount of pesticide sprayed by the drone remains constant.

[0014] In one possible design, the interval between any two adjacent sampling times is the same across multiple sampling times.

[0015] The present invention has the following beneficial effects: In the intelligent control method for pesticide spraying of crops provided by this invention, by automatically analyzing crop images at each sampling moment, it is possible to simultaneously perceive the structural complexity of crops represented by growth characteristics and the health status represented by pest and disease impact characteristics. This realizes the direct association between abstract visual characteristics of crops (such as density, color depth, and positional relationship) and specific flight speed control, ensuring that areas with lush growth or severe pest and disease can obtain longer spraying time and more pesticide dosage. This effectively solves the problem of blind spots or insufficient spraying caused by shading or uneven distribution of pests and diseases, and ensures the control effect. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the hardware architecture of a drone provided in one embodiment of the present invention; Figure 2 This is a flowchart illustrating an intelligent control method for pesticide spraying on crops, provided in one embodiment of the present invention. Figure 3 This is a schematic diagram of a crop image captured by a drone according to an embodiment of the present invention; Figure 4 A schematic diagram of a dwell time weighting curve provided in one embodiment of the present invention; Figure 5 This is a flowchart illustrating an intelligent control method for pesticide spraying on crops, provided in one embodiment of the present invention. Figure 6 This is a flowchart illustrating an intelligent control method for pesticide spraying on crops, provided in one embodiment of the present invention. Figure 7 This is a schematic diagram showing the relative positions of a closed region according to an embodiment of the present invention; Figure 8 This is a schematic flowchart of an intelligent control method for pesticide spraying on crops, provided as an embodiment of the present invention. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an intelligent control method for pesticide spraying on crops proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] In embodiments of the present invention, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" or "for example" in embodiments of the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0020] In the description of this invention, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Furthermore, "at least one" and "more than one" refer to two or more. The terms "first," "second," etc., do not limit the quantity or order of execution, and "first," "second," etc., do not necessarily imply differences.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0022] The following description, in conjunction with the accompanying drawings, details a specific scheme for an intelligent control method for pesticide spraying on crops provided by the present invention.

[0023] Please see Figure 1 The diagram illustrates the hardware architecture of a drone provided in an embodiment of the present invention, wherein the drone 10 includes an image acquisition module 11, a pesticide spraying module 12, a control module 13, and a communication module 14.

[0024] The image acquisition module 11 is used to acquire visual information of crops in real time during the flight of the UAV 10. The image acquisition module 11 includes a camera unit 111 and a preprocessing unit 112.

[0025] The camera unit 111 can be a high-resolution visible light camera, a multispectral camera, etc., and is installed on the drone 10 to collect images of crops within the pesticide spraying range at multiple sampling times.

[0026] Optionally, the camera unit 111 is fixedly mounted on the drone 10 via a gimbal. The gimbal can enhance the stability of the camera unit 111 to counteract the attitude changes of the drone 10 during flight and ensure image quality.

[0027] The preprocessing unit 112 can be integrated into the onboard computing device of the UAV 10 to perform preliminary processing on the raw images acquired by the camera unit 111.

[0028] Optionally, the preliminary processing flow includes using Gaussian denoising to eliminate noise interference and using a preset image segmentation algorithm to effectively separate the crop areas in the image from the background such as soil and shadows, providing high-quality image data for subsequent feature analysis.

[0029] In some embodiments, the crop images processed by the image acquisition module 11 can be transmitted to the control module 13 via the communication module 14 as the original basis for determining the crop growth and pest and disease status.

[0030] The pesticide spraying module 12 is used to complete the pesticide spraying action according to control commands. The pesticide spraying module 12 includes a pesticide liquid storage and delivery unit 121 and a nozzle array unit 122.

[0031] The pesticide storage and delivery unit 121 consists of a pesticide tank, pump, pipeline and control valve, and is used to store pesticides and provide stable delivery pressure.

[0032] The nozzle array unit 122 consists of multiple atomizing nozzles that can be controlled independently or in groups to ensure that the drone 10 can form a pesticide spraying range of the expected shape and extent at a fixed flight altitude.

[0033] The pesticide spraying module 12 executes the pesticide spraying action in response to the control command of the control module 13.

[0034] In some embodiments, by keeping the flight altitude and pump pressure of the drone 10 constant, the instantaneous spraying volume of the pesticide spraying module 12 can be kept constant, thereby transforming the control of pesticide coverage into the control of the dwell time of the drone 10 in the area, that is, by controlling the flight speed of the drone 10 to achieve variable spraying of pesticides.

[0035] The control module 13 is used for data processing, decision generation, and flight control. The control module 13 can be implemented by the onboard computer of the UAV 10 or a high-performance embedded module. The control module 13 includes a path planning unit 131, a feature analysis unit 132, and a decision and flight control unit 133.

[0036] The path planning unit 131 is used to pre-calculate and store a preset flight path that covers the entire area, based on the regional geographic information of the crop to be sprayed and the pesticide spraying range of the pesticide spraying module 12, before the pesticide spraying operation begins.

[0037] The feature analysis unit 132 is used to receive the image processed by the image acquisition module 11 and analyze the crop features in the image. Specifically, the feature analysis unit 132 first identifies multiple closed regions in the image using a preset edge detection algorithm, and then calculates growth features to characterize the degree of crop structural occlusion, and pest and disease impact features to characterize the severity of crop damage caused by pests and diseases.

[0038] The decision and flight control unit 133 is used to calculate the flight control parameters of the UAV based on the growth characteristics and pest and disease impact characteristics output by the feature analysis unit 132, and then generate control commands to control the flight speed of the UAV, so as to control the UAV 10 to fly at a dynamic speed along a preset flight path.

[0039] In some embodiments, the speed control commands output by the decision and flight control unit 133 of the control module 13 are ultimately sent to the flight controller of the UAV 10 via the communication module 14 to drive its execution. Simultaneously, the control module 13 also sends pesticide spraying control commands to the pesticide spraying module 12 via the communication module 14 to ensure that pesticide spraying is synchronized with the flight status.

[0040] The communication module 14 is used to ensure communication between the various modules within the UAV 10, as well as data transmission between the UAV 10 and a remote control terminal (such as a ground station). Optionally, the communication module 14 integrates wireless communication devices such as a high-speed data radio and a 4G / 5G cellular network module.

[0041] Inside the UAV 10, the communication module 14 is used to forward the crop images processed by the image acquisition module 11 to the control module 13, and send the control commands generated by the control module 13 to the flight controller and pesticide spraying module 12 of the UAV 10 respectively.

[0042] In external communication, the communication module 14 can be used to receive remote start and stop commands, report the status information of the UAV 10 (such as location, battery level, operation progress, etc.), and transmit key process images to realize the visualization monitoring and traceability of the operation process.

[0043] Please see Figure 2 The diagram illustrates a method flowchart for an intelligent control method for pesticide spraying on crops according to an embodiment of the present invention, including steps S201-S204.

[0044] S201. Acquire crop images within the pesticide spraying range collected by the drone at multiple sampling times.

[0045] One possible approach is to control the drone to fly along a preset flight path and collect images at multiple sampling times based on a set interval. Then, the images collected at each sampling time are processed to obtain crop images.

[0046] It should be noted that the drone flies at a fixed altitude during the operation, and the amount of pesticide sprayed by the drone remains constant. The preset flight path refers to a fixed spatial trajectory determined before the pesticide spraying operation begins, based on the boundary of the crop planting area and the pesticide spraying range of the drone at a fixed altitude. This preset flight path remains unchanged to ensure that the drone can traverse the entire crop planting area and achieve full coverage of the crop planting area by the pesticide spraying operation.

[0047] In some embodiments, the interval between any two adjacent sampling times in the multiple sampling times is the same, for example, 0.5 seconds, 1 second, 2 seconds, etc., which can be determined according to the flight speed of the drone and the pesticide spraying range of the drone.

[0048] It should be noted that the spraying range of pesticide spraying modules installed on drones mainly depends on the drone's flight altitude. The higher the flight altitude, the larger the coverage area, but the actual spray content for individual crops will decrease. Therefore, this invention aims to eliminate the influence of drone flight altitude on the pesticide spraying content of crops, ensuring that the drone maintains a consistent flight altitude throughout operation, facilitating subsequent processing.

[0049] With the drone flying at a fixed altitude, the actual pesticide spraying range during the drone's spraying process can be determined, such as... Figure 3 As shown, the pesticide spraying range is smaller than the size of the acquired image range, thus allowing the acquisition of crop images corresponding to the pesticide spraying process by the drone at each acquisition time from the acquired crop images.

[0050] In some embodiments, the images acquired by the UAV at each sampling time are processed, including: eliminating the influence of noise through denoising processing (such as using a Gaussian filtering method), and separating the crop planting area from the background through an image segmentation algorithm (such as using a semantic segmentation algorithm), thereby obtaining a crop image corresponding to each sampling time.

[0051] Optionally, median filtering can be used for noise reduction, or a deep learning-based instance segmentation method can be used to separate the crop planting area from the background. This embodiment of the invention does not specifically limit the specific methods used.

[0052] In some embodiments, the drone can also transmit the images collected at each sampling moment to a remote control terminal via a communication module (Internet of Things), where the remote control terminal processes the images (denoising and image segmentation, etc.) and then receives crop images returned by the remote control terminal.

[0053] In some embodiments, the drone also performs grayscale processing on the images collected at each sampling time, using a grayscale algorithm to convert the images into grayscale images, which facilitates subsequent analysis of the impact characteristics of crop diseases and pests.

[0054] S202. For each sampling time, determine the growth characteristics and pest and disease impact characteristics of the crop based on the crop image.

[0055] Among them, the growth feature is used to characterize the degree of occlusion of crop structure in the corresponding area of ​​the crop image, and the pest and disease impact feature is used to characterize the severity of the impact of pests and diseases on crops in the corresponding area of ​​the crop image.

[0056] As one possible approach, for the crop growth characteristics at each sampling time, edge detection can be performed on the crops in the crop image. After fitting, a closed region corresponding to each crop in the crop image can be obtained. Combined with the crop planting area in the crop image, the structural density of the crop in the corresponding region and the area variance that can characterize the dispersion of the closed region area can be determined. The two are combined to form a comprehensive growth characteristic. The larger the value of the growth characteristic, the more vigorous the crop growth and the more serious the structural occlusion. It is more necessary to slow down the flight speed or stop when passing through the crop region.

[0057] For the pest and disease impact characteristics of crops at each sampling time, the area of ​​each closed region is determined, and its average gray value is determined based on the corresponding grayscale image. Then, the pesticide demand is determined based on the area and average gray value. A lower average gray value indicates a darker closed region, suggesting a potentially more severe impact from pests and diseases. Simultaneously, a positional demand weight is determined for each closed region based on its relative position within the pesticide spraying area. This positional demand weight characterizes the degree to which the closed region requires a reduction in the drone's flight speed. Finally, by combining the pesticide demand and positional demand weights of all closed regions, the pest and disease impact characteristics of crops in the crop image corresponding to that sampling time are obtained. A higher value for this pest and disease impact characteristic indicates a more urgent need for pesticide spraying in the current area.

[0058] It should be noted that the crop growth characteristics at each sampling time can be referred to the description of steps S301-S304 in the subsequent embodiments of the present invention, and the crop disease and pest impact characteristics at each sampling time can be referred to the description of steps S401-S406 in the subsequent embodiments of the present invention, which will not be repeated here.

[0059] In some embodiments, for the crop growth characteristics at each sampling time, after determining the closed regions, the ratio of the perimeter to the area of ​​each closed region can be calculated to obtain the shape complexity of the closed region. Crops with good growth have more structural branches, resulting in a more complex shape of the closed region, and thus a higher shape complexity. Simultaneously, the vertical number of layers in the closed region of the crop image is determined; the more severe the occlusion, the more layers there are. The mean of the shape complexity is then weighted and summed with the number of layers to obtain the growth characteristics. A higher growth characteristic value indicates better crop growth in that region.

[0060] In some embodiments, for the characteristics of crop pest and disease impact at each sampling time, after determining the closed region, the variance of the gray value of each closed region can be calculated. The larger the variance, the more uneven the distribution of pests and diseases within the region and the more severe the local damage. Simultaneously, the vertical distance from the centroid of the closed region to the UAV's flight direction is calculated. The greater the distance, the shorter the time the UAV takes to pass through the region, and the stronger the need for speed reduction. Further, the gray value variance and the vertical distance are weighted and summed, and then normalized to obtain the pest and disease impact characteristics. The larger this characteristic value, the more severe the pest and disease impact and the stronger the need for speed reduction.

[0061] In some embodiments, deep learning methods can be used to directly classify image regions into levels, thereby obtaining the growth level corresponding to the growth characteristics and the pest and disease impact level corresponding to the pest and disease impact characteristics.

[0062] S203. For each sampling time, determine the flight control parameters of the UAV based on the growth characteristics and the impact characteristics of pests and diseases.

[0063] Among them, the flight control parameters are used to indicate the duration for which the pesticide spraying range of the UAV covers the location corresponding to the sampling time.

[0064] As one possible approach, for each sampling time, the growth characteristics are associated with the pest and disease impact characteristics to obtain the dwell time weight corresponding to that sampling time. This dwell time weight is used to characterize the comprehensive weight that needs to be adjusted for dwell time under the combined effect of crop growth and pests and diseases at that sampling time.

[0065] Furthermore, by iterating through all sampling times, with the sampling time order as the horizontal axis and the dwell time weight as the vertical axis, the dwell time weights of adjacent sampling times are connected by curves. The connected curves are then smoothed to eliminate the influence of instantaneous weight fluctuations on dwell time, resulting in a curve of "dwell time weight - sampling time" covering the entire area. This curve is used to characterize the differences in dwell time requirements at different sampling times and corresponding locations.

[0066] Furthermore, based on the smoothed weight change curve, the maximum and minimum values ​​of the dwell time weights at all sampling times are extracted, and the average of the two is calculated as the weight reference value. The weight reference value is associated with the preset dwell time to ensure that the dwell time corresponding to the weight is within a reasonable range. This ensures that the dwell time is not too short due to too small a weight, resulting in insufficient pesticide spraying, nor too long due to too large a weight, resulting in low efficiency.

[0067] Furthermore, for each sampling time, the actual dwell time at the location corresponding to that sampling time is calculated based on the ratio of the dwell time weight to the weight reference value, combined with the preset dwell time. This is used as the flight control parameter of the drone. The dwell time weight is positively correlated with the dwell time. The larger the dwell time weight, the longer the drone's pesticide spraying range covers the location corresponding to that sampling time, ensuring that the coverage time at that location matches the needs of the crops.

[0068] It should be noted that the preset dwell time can be set to 8 seconds, 10 seconds, 12 seconds, etc. based on experience. The specific setting can be based on historical experience or farmland conditions. This embodiment of the invention does not impose specific limitations on this.

[0069] In some embodiments, the smoothed weight change curve is as follows: Figure 4 As shown, the horizontal axis represents the sampling time sequence, and the vertical axis represents the dwell time weight. The preset dwell time is removed from the unit and used as the straight line corresponding to the average of the maximum and minimum values ​​of the curve. Based on the difference between the dwell time weight value at each sampling time position and the straight line, as well as the value of the preset dwell time, the dwell time at each sampling time position can be obtained.

[0070] In some embodiments, after determining the dwell time weight for each sampling moment, the dwell time weight corresponding to the sampling moment can be further classified into levels, such as low weight, medium weight, and high weight. Then, based on the weight level corresponding to the dwell time weight at each sampling moment, a corresponding dwell time range is matched. For example, a low weight level corresponds to a dwell time of 8-10 seconds, a medium weight level corresponds to a dwell time of 10-12 seconds, and a high weight level corresponds to a dwell time of 12-15 seconds. Based on the dwell time range corresponding to the dwell time weight, any value within the range is taken as the actual dwell time at the location corresponding to the sampling moment, i.e., the flight control parameter.

[0071] S204. Control the UAV to fly along the preset flight path according to the flight control parameters.

[0072] One possible implementation involves controlling the drone to fly along a preset flight path. When the drone's pesticide spraying range passes the location corresponding to the sampling time, the flight speed is controlled according to the flight control parameters at that sampling time, ensuring that pesticide is sprayed for the appropriate duration at that location. During this process, the drone maintains a fixed flight altitude, and the pesticide spraying module maintains a constant spraying rate, thus ensuring that changes in flight speed directly and accurately reflect differences in pesticide coverage.

[0073] In some embodiments, controlling the UAV to fly along a preset flight path according to flight control parameters further includes first determining the target flight speed of the UAV at each position on the preset flight path according to the flight control parameters.

[0074] Specifically, for each sampling moment, the location corresponding to that sampling moment is determined. This includes the moments when the drone enters and leaves the pesticide spraying range along the preset flight path. Since the drone flies at a constant speed while acquiring images along the preset flight path, the distance the pesticide spraying range covers to the location corresponding to that sampling moment can be obtained by using the time interval between these two moments and the constant speed. Furthermore, based on the duration of the pesticide spraying range covering the location corresponding to that sampling moment as indicated by the flight control parameters, and the physical relationship between distance, duration, and speed, the flight speed at which the pesticide spraying range covers the location corresponding to that sampling moment can be obtained.

[0075] Based on this, the flight speed corresponding to each sampling time can be obtained. Furthermore, at any position on the preset flight path (including the position corresponding to each sampling time during the acquisition of crop images, as well as the position corresponding to non-sampling times), if there is only one or more identical flight speeds, then that flight speed is used as the target flight speed corresponding to that position; if there are multiple different flight speeds, then the minimum speed among the multiple flight speeds is used as the target flight speed corresponding to that position, so as to meet the pesticide spraying needs of crops.

[0076] Furthermore, the drone is controlled to fly at the target speed corresponding to each position along a preset flight path.

[0077] After obtaining the target flight speed along the complete preset flight path, the control module generates corresponding flight control commands based on the target flight speed at each location, driving the drone to fly along the preset flight path at the calculated target flight speed. During this process, the drone maintains a fixed flight altitude, and the pesticide spraying module maintains a constant spraying rate.

[0078] Understandably, in the intelligent control method for pesticide spraying of crops provided by this invention, by automatically analyzing crop images at each sampling moment, it is possible to simultaneously perceive the structural complexity of crops characterized by growth characteristics and the health status characterized by pest and disease impact characteristics. This realizes the direct association between abstract visual characteristics of crops (such as density, color depth, and positional relationship) and specific flight speed control, ensuring that areas with lush growth or severe damage can obtain longer spraying time and more pesticide dosage. This effectively solves the problem of blind spots or insufficient spraying caused by shading or uneven distribution of pests and diseases, and ensures the control effect.

[0079] In a design, such as Figure 5 As shown, this embodiment of the invention also provides a specific implementation method for determining the growth characteristics of crops based on crop images, including steps S301-S304.

[0080] S301. Identify multiple closed regions in the crop image.

[0081] The closed region is the area of ​​the crop in the crop image.

[0082] One possible implementation involves processing crop images based on a pre-defined edge detection algorithm to determine the edge image of each crop in the image. Further, the edge image of each crop is fitted to obtain multiple closed regions.

[0083] Optionally, the Canny edge detection algorithm can be used to process the crop image to identify the pixel boundaries of local structures such as leaf edges and branch outlines, ultimately obtaining an edge image containing only the edge pixels of the crop. In this edge image, discrete edge pixels correspond to the contours of different structures of the crop. Further, contour fitting is performed on the discrete edge pixels in the edge image, that is, connecting adjacent edge pixels belonging to the same local structure of the crop to form a complete, closed contour line, ultimately forming multiple independent local closed regions.

[0084] In some embodiments, a deep learning-based instance segmentation model can also be used to directly output the precise closed region of each crop instance in the crop image.

[0085] S302. Determine the structural density of crops in the crop image based on the number of multiple closed regions and the area of ​​the crop region in the crop image.

[0086] As one possible implementation, after determining multiple closed regions in the crop image based on step S301 above, the total number of closed regions is determined, and the total area of ​​the entire crop region in the image is calculated. Furthermore, the structural density of the crop in the crop image is determined based on the ratio of the total number of closed regions to the total area of ​​the crop region.

[0087] Understandably, structure density reflects the number of independent structures contained within a crop region in a crop image. When crops grow better and have denser foliage, the number of independent closed regions that can be identified within the same area of ​​the image is greater, and the corresponding structure density value is higher. This effectively quantifies the complexity of the crop canopy and the density of the branches and leaves, providing a data basis for assessing the structural occlusion caused by vigorous growth.

[0088] In some embodiments, the formula for calculating the structure density at each sampling time is as follows: In the formula, Indicates the first Each sampling time, For the first Each sampling time corresponds to the structural density of crops in the crop image. For the first Each sampling time corresponds to the number of closed regions in the crop image. For the first Each sampling time point corresponds to the area of ​​the crop region in the crop image, where... >0, if If the value is 0, then this region is ignored.

[0089] In the formula, for the same crop coverage area The number of locally closed regions The higher the density, the greater the structural density of crops in the region. The higher the height, the better the overall growth.

[0090] S303. Determine the area variance of multiple closed regions.

[0091] Among them, the area variance is used to characterize the crop growth in the corresponding area of ​​the crop image.

[0092] One possible approach is to calculate the area of ​​each closed region, and then calculate the area variance within the corresponding region of the crop image.

[0093] Understandably, area variance reflects the degree of difference in area size among different closed regions. When crops are growing well and have rich structural layers, there will be a significant area difference between the upper and lower leaves. The upper leaves have a larger area, while the lower, shaded leaves have a smaller area, leading to increased dispersion in the area values ​​of each closed region. Consequently, the area variance increases accordingly. Therefore, area variance can objectively reflect the complexity of crop structure and the richness of its hierarchical distribution.

[0094] S304. Determine the growth characteristics based on structural density and area variance.

[0095] As one possible approach, growth features are used to characterize the degree of crop structure occlusion within the corresponding region of a crop image. They are positively correlated with both structure density and area variance. Therefore, the product of structure density and area variance can be used as the growth feature.

[0096] In some embodiments, the formula for calculating the growth characteristics of crops in crop images is as follows: In the formula, Indicates the first Each sampling time, For the first Each sampling time point corresponds to the growth characteristics of crops in the crop image. For the first Each sampling time corresponds to the structural density of crops in the crop image. For the first Each sampling time corresponds to the area variance of multiple closed regions in the crop image. () is the normalization function. In this embodiment, the normalization method can be the maximum and minimum value normalization.

[0097] Among them, growth characteristics The larger the value, the more likely it is to be the first. The better the actual growth of crops in the crop image corresponding to each sampling time, the more occlusions are generated, and the greater the need to reduce the flight speed of the drone.

[0098] It should be noted that eigenvalues It is a dimensionless indicator used to comprehensively evaluate the growth of crops.

[0099] Understandably, based on the method for determining crop growth characteristics shown in steps S301-S304 above, the morphological characteristics of crops in images can be transformed into quantifiable growth evaluation indicators. By analyzing the distribution characteristics of closed regions in the image, an evaluation system is constructed from two dimensions: "quantity density" and "size distribution." Structural density reflects the number of independent structures of crops per unit area, while area variance characterizes the degree of difference in size among these structures. This two-dimensional evaluation mechanism can comprehensively capture the spatial structural characteristics of crops, considering both the density of branches and leaves and reflecting the differences in area distribution caused by upper shading.

[0100] In a design, such as Figure 6 As shown, this embodiment of the invention also provides a specific implementation method for determining the characteristics of crop pest and disease impacts based on crop images, including steps S401-S406.

[0101] S401. Identify multiple closed regions in the crop image.

[0102] The closed region is the area of ​​the crop in the crop image.

[0103] It should be noted that the specific implementation of step S401 can be referred to the description of step S301 in the above embodiments of the present invention, and will not be repeated here.

[0104] S402. For each closed region, determine the area and average gray value of the closed region.

[0105] The average gray value is used to characterize the degree of pest and disease impact on crops corresponding to the closed area.

[0106] As one possible implementation, the area of ​​each closed region in the determined crop image is determined, and the average gray value of the closed region is determined based on the gray values ​​of each pixel in the closed region in the corresponding grayscale image of the crop image.

[0107] Based on this, we can obtain any closed region in the crop image corresponding to that sampling time. area and the closed region Average gray value .

[0108] S403. Determine the pesticide requirement based on the area and average gray value.

[0109] Among them, pesticide demand is used to characterize the degree of pesticide demand in a closed area.

[0110] It should be noted that because crops have a certain height, and pests and diseases can affect crops at different heights, the size of the enclosed area reflects the height of the crop. When crops are taller, the corresponding enclosed area is less obstructed, resulting in a larger enclosed area. Conversely, when crops are shorter, they are obstructed by taller crops, leading to a smaller enclosed area. For taller crops affected by pests and diseases, pesticide spraying can be done more frequently, while for shorter crops affected by pests and diseases, longer spraying time is required. Therefore, the pesticide requirement for crops with enclosed areas is negatively correlated with the area of ​​the enclosed area.

[0111] It should be noted that, because the lighting angle of crop images is fixed, under fixed lighting conditions, darkening of crop areas often reflects visual symptoms of pests and diseases such as chlorophyll degradation and tissue necrosis. Therefore, the average gray value can be used to characterize the degree of pest and disease impact on crops. A lower average gray value in a closed area indicates a higher degree of pest and disease impact, requiring a longer pesticide spraying time. Conversely, a higher average gray value in a closed area indicates a lower degree of pest and disease impact, allowing for shorter pesticide spraying times. Therefore, the pesticide requirement for crops corresponding to closed areas is negatively correlated with the average gray value of those areas.

[0112] Based on this, the formula for calculating the pesticide requirement for any closed region is as follows: In the formula, This indicates that the sampling time corresponds to any closed region in the crop image. For any closed region The demand for pesticides, For any closed region area, For any closed region Average gray value, The denominator should be a very small positive number to avoid being 0 in the formula.

[0113] Among them, when any closed region Area size The smaller the value, the lower the mean grayscale value. The smaller the value, the greater the susceptibility to pests and diseases. Therefore, by combining the two parameters and expressing them as a fraction, the pesticide requirement can be obtained. The larger the pesticide requirement value, the more likely any closed region will be affected. The higher the demand for pesticide spraying.

[0114] S404. Determine the relative position of each closed area within the pesticide spraying range.

[0115] The relative position is used to characterize the position of the closed region relative to the flight direction of the UAV.

[0116] As one possible implementation, such as Figure 7 As shown, a coordinate system is established with the drone's flight direction as the vertical axis and the location at the edge of the pesticide spraying range in the opposite direction of the drone's flight direction as the origin. The data size of the vertical axis is calibrated by the diameter of the pesticide spraying range. Furthermore, for any closed region in the crop image, the centroid of the closed region is obtained, and the vertical projection of the centroid onto the vertical axis is determined. The vertical axis reading of this projection represents the relative position of the closed region within the pesticide spraying range. The smaller the value of this relative position, the farther the closed region is from the drone's flight direction, and the higher the requirement for reducing the drone's flight speed.

[0117] In some embodiments, to determine the relative position of each closed region within the pesticide spraying range, the following method can also be used, wherein, for each closed region, the projected position of the centroid of the closed region in the direction of the UAV's flight and the projected distance between the centroid of the closed region and the UAV in the direction of flight are first determined, and the projected distance represents the distance between the centroid of the closed region and the UAV in the direction of flight.

[0118] Furthermore, when the projection position is located in front of the drone's direction, the sum of the radius of the pesticide spraying range and the projection distance is determined as the relative position of the closed area within the pesticide spraying range.

[0119] When the projection position is behind the drone in the direction of the drone, the difference between the radius of the pesticide spraying range and the projection distance is determined as the relative position of the closed area within the pesticide spraying range.

[0120] Understandably, the smaller the relative position of the closed area, the shorter the time that the closed area will be covered by the pesticide spraying range, and the higher the need for the closed area to slow down the drone's flight speed. Conversely, the larger the relative position of the closed area, the longer the time that the closed area will be covered by the pesticide spraying range, and the lower the need for the closed area to slow down the drone's flight speed.

[0121] S405. Determine the location requirement weight based on the relative position and the diameter of the pesticide spraying range.

[0122] Among them, the location demand weight is used to characterize the degree of demand for a reduction in the flight speed of the drone within a closed area.

[0123] In some embodiments, the formula for calculating the location requirement weight for any closed region is as follows: In the formula, This indicates that the sampling time corresponds to any closed region in the crop image. For any closed region Location demand weight, For any closed region The relative position, This refers to the diameter of the area from which the pesticide is sprayed.

[0124] This involves calculating the relative positions of closed regions. Diameter of the pesticide spraying range The ratio of these values ​​represents the relative longitudinal position of the closed region within the pesticide spraying area. At the same time, combined with relative position Weight of location demand The negative correlation can be used to calculate the location demand weight. The higher the position demand weight value, the greater the need to reduce the flight speed of the drone.

[0125] S406. Determine the characteristics of pest and disease impact based on the pesticide demand and location weight of each closed region in multiple closed regions.

[0126] As one possible approach, for the crop image at the current sampling time, the product of the pesticide demand and location demand weights of all closed regions in the crop image is accumulated and normalized to obtain the pest and disease impact characteristics of the crop image at the current sampling time. The larger the value of the pest and disease impact characteristics, the greater the demand for pesticide spraying and the higher the degree of need to reduce the flight speed of the drone.

[0127] In some embodiments, the calculation formula for the impact characteristics of pests and diseases is as follows: In the formula, Indicates the first Each sampling time, This indicates that the sampling time corresponds to any closed region in the crop image. For the first Each sampling time point corresponds to the characteristics of crop pest and disease impacts in the crop image. For the first Each sampling time corresponds to the number of closed regions in the crop image. For any closed region Location demand weight, For any closed region The demand for pesticides.

[0128] Among them, by accumulating the first The product of pesticide demand and location demand weights for all closed regions in the crop image at each sampling time point, after normalization, yields the pest and disease impact characteristics of the crop in the crop image at the i-th sampling time point. .

[0129] Understandably, based on the method for determining the impact characteristics of crop diseases and pests shown in steps S401-S406 above, the pesticide demand and location demand weight of each closed region (crop) are used to determine the degree of pesticide spraying demand of closed regions from the severity of the impact of diseases and pests and their spatial distribution characteristics. Then, by combining the pesticide spraying demand of all closed regions in the crop image, the impact characteristics of crop diseases and pests in the crop image are obtained, thereby reflecting the demand of crops for pesticide spraying in the crop image.

[0130] In a design, such as Figure 8 As shown, this embodiment of the invention also provides a specific implementation method for determining the flight control parameters of the UAV based on the growth characteristics and the pest and disease impact characteristics, including steps S501-S502.

[0131] S501. For each sampling time, determine the weight of the dwell time based on the growth characteristics and the impact characteristics of pests and diseases.

[0132] As one possible approach, growth characteristics are positively correlated with crop demand for pesticide spraying, and pest and disease impact characteristics are also positively correlated with crop demand for pesticide spraying. Based on this, the product of growth characteristics and pest and disease impact characteristics can be calculated as the weight of dwell time.

[0133] In some embodiments, the formula for calculating the dwell time weight corresponding to the sampling time is as follows: In the formula, Indicates the first Each sampling time, For the first The weight of the dwell time corresponding to each sampling time. For the first Each sampling time point corresponds to the characteristics of crop pest and disease impacts in the crop image. For the first Each sampling time point corresponds to the growth characteristics of crops in the crop image.

[0134] Among them, for the first At the sampling time point, if the weight of the dwell time is larger, it indicates that the th sampling time is... The higher the demand for pesticide spraying in the crop images corresponding to each sampling time, the longer the drone stays at that location should be increased to meet the crop's pesticide spraying needs.

[0135] It should be noted that, since the drone flies at a fixed altitude and along a fixed path during continuous spraying, and the instantaneous amount of pesticide sprayed at each location within the pesticide spraying area is fixed, the actual pesticide coverage of crops within the current pesticide spraying area depends on the duration of the drone's stay at the current location.

[0136] S502. Based on the dwell time weights of multiple sampling times and the preset dwell time, determine the flight control parameters corresponding to each sampling time.

[0137] As one possible implementation, a weight reference value is first determined based on the dwell time weights at multiple sampling times. This weight reference value is either the average of the dwell time weights at multiple sampling times, or the average of the maximum and minimum dwell time weights at multiple sampling times. Further, for each sampling time, the dwell time corresponding to each sampling time is calculated based on the ratio of the dwell time weight at that sampling time to the weight reference value, and a preset dwell time. This calculated dwell time is then used as the flight control parameter for that sampling time.

[0138] Understandably, based on the ratio of the dwell time weight to the weight reference value at the sampling time, it can be determined whether the dwell time should be increased or decreased for that sampling time. Furthermore, based on the preset dwell time, the dwell time can be avoided to prevent it from being too short or too long, so as to ensure the pesticide spraying effect on crops.

[0139] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0140] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for intelligent control of pesticide spraying on crops, characterized in that, The method includes: Acquire crop images within the pesticide spraying area collected by the drone at multiple sampling times; For each sampling time, the growth characteristics and pest and disease impact characteristics of the crop are determined based on the crop image. The growth characteristics are used to characterize the degree of crop structure occlusion in the corresponding area of ​​the crop image, and the pest and disease impact characteristics are used to characterize the severity of the impact of pests and diseases on the crop in the corresponding area of ​​the crop image. For each sampling time, the flight control parameters of the UAV are determined based on the growth characteristics and the pest and disease impact characteristics. The flight control parameters are used to indicate the duration for which the pesticide spraying range of the UAV covers the location corresponding to the sampling time. Based on the flight control parameters, the UAV is controlled to fly along a preset flight path; Determining the characteristics of crop pest and disease impacts based on the crop images includes: Identify multiple closed regions in the crop image, where each closed region corresponds to a crop in the crop image. For each closed region, the area and average gray value of the closed region are determined, and the average gray value is used to characterize the degree of pest and disease impact on the corresponding crop of the closed region; The pesticide demand is determined based on the area and the average gray value, and the pesticide demand is used to characterize the degree of pesticide demand of the closed area. Determine the relative position of each of the closed regions within the pesticide spraying range, the relative position being used to characterize the position of the closed region relative to the flight direction of the drone; Based on the relative position and the diameter of the pesticide spraying range, a position demand weight is determined, which is used to characterize the degree of need for the closed area to reduce the flight speed of the UAV. The pest and disease impact characteristics are determined based on the pesticide demand and location demand weight of each of the multiple closed regions.

2. The intelligent control method for pesticide spraying on crops according to claim 1, characterized in that, Determining crop growth characteristics based on the crop images includes: The structural density of crops in the crop image is determined based on the number of multiple closed regions and the area of ​​the crop region in the crop image; Determine the area variance of multiple closed regions, the area variance being used to characterize the growth of crops within the corresponding regions of the crop image; The growth characteristics are determined based on the structural density and the area variance.

3. The intelligent control method for pesticide spraying on crops according to claim 1, characterized in that, The step of determining the flight control parameters of the UAV based on the growth characteristics and the pest and disease impact characteristics includes: For each sampling time, the dwell time weight is determined based on the growth characteristics and the pest and disease impact characteristics; The flight control parameters corresponding to each sampling time are determined based on the dwell time weights of multiple sampling times and the preset dwell time.

4. The intelligent control method for pesticide spraying on crops according to claim 1, characterized in that, Controlling the UAV to fly along a preset flight path according to the flight control parameters includes: Based on the flight control parameters, the target flight speed of the UAV at each position on the preset flight path is determined; The drone is controlled to fly at the target flight speed corresponding to each position along the preset flight path.

5. The intelligent control method for pesticide spraying on crops according to claim 1, characterized in that, Determining multiple closed regions in the crop image includes: The crop image is processed based on a preset edge detection algorithm to determine the edge image of each crop in the crop image; The edge image of each crop is fitted to obtain multiple closed regions.

6. The intelligent control method for pesticide spraying on crops according to claim 1, characterized in that, Determining the relative position of each of the closed regions within the pesticide spraying area includes: For each closed region, determine the projected position of the centroid of the closed region in the flight direction of the UAV, and the projected distance between the centroid of the closed region and the UAV; When the projected position is located in front of the direction of flight of the UAV, the relative position is determined to be the sum of the radius of the pesticide spraying range and the projected distance; When the projected position is located behind the direction of flight of the UAV, the relative position is determined as the difference between the radius of the pesticide spraying range and the projected distance.

7. The intelligent control method for pesticide spraying on crops according to claim 3, characterized in that, The step of determining the flight control parameters corresponding to each sampling time based on the dwell time weights of multiple sampling times and a preset dwell time includes: A weight reference value is determined based on the dwell time weights at multiple sampling times. The weight reference value is the average of the dwell time weights at multiple sampling times, or the average of the maximum and minimum values ​​among the dwell time weights at multiple sampling times. The flight control parameters corresponding to each sampling time are determined based on the ratio of the dwell time weight to the weight reference value at each sampling time, and the preset dwell time.

8. The intelligent control method for pesticide spraying on crops according to claim 1, characterized in that, The drone flies at a fixed altitude, and the spraying rate of the drone remains constant.

9. The intelligent control method for pesticide spraying on crops according to claim 1, characterized in that, The interval between any two adjacent sampling times is the same across multiple sampling times.

Images