A pesticide application control method, system and medium for a plant protection drone

Abstract

The present invention discloses a pesticide application control method, system and medium for a plant protection UAV, aiming to provide an efficient and accurate pesticide application scheme for crops. The method comprises the following steps: first, basic information of the target farmland is obtained. Secondly, the enrichment of pests is determined by using image recognition technology, and a pest enrichment report is generated. Then, a pest control scheme is generated based on historical pesticide application dosages and pest enrichment reports. According to the pest control scheme, the optimal pesticide application time period for the plant protection UAV is determined. Then, within the optimal pesticide application time period, pesticide application is performed on crops. Finally, the effect of the crops after the pesticide application operation is judged to evaluate the effect of the pesticide application. The present invention realizes the integration of farmland information acquisition, pest monitoring, pesticide application time determination and pesticide application effect evaluation, thereby improving the efficiency and accuracy of pesticide application by plant protection UAVs.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural pesticide application, and in particular to a pesticide application control method, system and medium of a plant protection unmanned aerial vehicle. Background Art

[0002] At present, there is a widespread problem of insect pests harming crops in agricultural production, so plant protection pesticide application is one of the important means to ensure the healthy growth of crops. Traditional pesticide application methods mainly rely on manual operation or mechanical equipment, but these methods have problems such as low efficiency, uneven application, excessive use of pesticides and environmental pollution, and lack the ability to accurately apply pesticides. As an emerging agricultural pesticide application tool, plant protection drones have the advantages of flexibility, efficiency and speed, and can achieve accurate pesticide application, reduce the use of pesticides, reduce environmental pollution, and improve crop yield and quality. However, most plant protection drones currently still use traditional timed and quantitative pesticide application, and fail to carry out personalized pesticide application according to the actual situation of specific farmland. The lack of monitoring and analysis of pest information leads to poor pesticide application effect. Therefore, there is an urgent need for a plant protection drone pesticide application control method that can integrate farmland information, monitor pests, intelligently judge the best time for pesticide application, and evaluate the effect of pesticide application. This method should be based on advanced image recognition technology and data analysis methods, and can realize the personalization of crop pesticide application, improve the efficiency and accuracy of pesticide application, and achieve scientific, environmentally friendly and sustainable agricultural production goals. Summary of the invention

[0003] In order to solve at least one of the above technical problems, the present invention proposes a pesticide application control method, system and medium for a crop protection UAV.

[0004] A first aspect of the present invention provides a pesticide application control method for a plant protection UAV, comprising:

[0005] Obtaining basic information of the target farmland, wherein the basic information of the farmland includes location information of the target farmland, crop type information, and pest information;

[0006] Determine the pest enrichment situation based on image recognition technology and generate pest enrichment report;

[0007] Obtain the historical pesticide dosage of each crop planting area in the target area, and generate a pest control plan based on the historical pesticide dosage and pest enrichment report;

[0008] Determine the best time for pesticide application by plant protection drone according to the pest control plan;

[0009] Apply pesticides to crops based on pest control plans and optimal application time periods;

[0010] Determine the effect of pesticide application on crops after the application.

[0011] In this solution, the basic information of the target farmland is obtained, and the basic information of the farmland includes the location information of the target farmland, the crop type information, and the pest information, specifically:

[0012] Based on the GPS positioning device, the target farmland location information is obtained, the location information includes longitude and latitude data, a map model is constructed, the target farmland location information is imported into the map model, a target farmland map model is generated, and the map model is displayed on a preset display;

[0013] Acquire crop type information according to farmland planting information, wherein the crop type information includes crop name and growth height;

[0014] According to the farmland survey data, the pest information in the farmland is obtained, and the pest information includes the pest name and pest characteristics.

[0015] In this solution, the pest enrichment situation is judged based on image recognition technology, and a pest enrichment report is generated, specifically:

[0016] Acquire video data of a process of performing an enrichment operation on pests in a preset area, extract video frame images of the video data, and obtain a video frame image set;

[0017] Extracting pixel information of each frame image of the video frame image set;

[0018] Calculate a pixel covariance matrix of the video frame image set according to the pixel information, and calculate an eigenvector and an eigenvalue of the video frame image set according to the pixel covariance matrix;

[0019] The eigenvalues ​​are sorted from large to small, and the first eigenvalue is selected as the principal component of the eigenvector, and the original pixel information is projected onto the principal component to obtain a video frame image set data of a first dimensionality reduction;

[0020] Calculate the similarity matrix of the video frame image set data with a reduced dimension, use the similarity matrix to calculate the conditional probability between data points, and obtain a probability distribution graph;

[0021] Initialize the data point positions of the second dimension reduction, and calculate the conditional probability between the data points after the second dimension reduction according to the probability distribution diagram;

[0022] The KL divergence value is calculated according to the conditional probability between the probability distribution graph and the data points of the secondary dimensionality reduction, and the position of the data points of the secondary dimensionality reduction is adjusted cyclically until the KL divergence value reaches a preset divergence value, thereby obtaining a set of secondary dimensionality reduction video frame images;

[0023] Build image processing models based on image recognition technology;

[0024] Importing each video frame image of the secondary dimensionality reduction video frame image set into the image processing model, marking the position of each pest with a dot, and obtaining a marked image;

[0025] The enrichment degree of the pests is determined according to the marked images, and a pest enrichment report is generated, wherein the pest enrichment report includes the number of the pests in the target farmland.

[0026] In this scheme, the historical pesticide dosage of each crop planting area in the target area is obtained, and a pest control scheme is generated according to the historical pesticide dosage and pest enrichment report, specifically:

[0027] Obtain the historical pesticide dosage of each crop planting area in the target area, and determine whether the historical pesticide dosage meets the pollution standard. If it does not meet the pollution standard, generate a pesticide control plan for plant protection drones;

[0028] If the pollution standard is reached, the phototropism information of the natural enemy insects of the pests in the target farmland is obtained, and the number threshold of the natural enemy insects is set;

[0029] Initialize a preset number of natural enemy insect light induction devices within a preset time period to induce the natural enemy insects to the target farmland;

[0030] Observe the quantity change information of the natural enemy insects of the pests in real time through the infrared detection device, and draw the quantity change information into a quantity-time change graph in real time;

[0031] If the number of natural enemy insects in the target farmland is not greater than the number threshold after the natural enemy insects are induced, a natural enemy insect induction network is set according to the preset number of natural enemy insect light induction devices;

[0032] If the number of natural enemy insects in the target farmland is greater than the quantity threshold after the natural enemy insects are induced, the natural enemy insect light induction equipment will be evenly reduced, and the natural enemy insects will be drained until the number of natural enemy insects in the target farmland after induction is no greater than the quantity threshold.

[0033] In this scheme, the optimal time period for applying pesticides by plant protection drones is determined according to the pest control scheme, specifically:

[0034] Obtaining timestamp information of the video frame image;

[0035] According to the timestamp information of pest enrichment report and video frame image, draw the number-time curve of pest quantity change based on time change;

[0036] According to the change information of the quantity-time curve, determine the time period with the largest number of pests enriched;

[0037] According to the time period, the optimal time period for spraying pesticides by the plant protection drone is determined.

[0038] In this scheme, the pesticide application operation on crops based on the pest enrichment report and the optimal pesticide application time period is specifically as follows:

[0039] Dividing the target farmland into N small areas in a grid manner, and displaying the divided small areas in the target farmland map model;

[0040] During the optimal application period, the camera device of the plant protection drone acquires image information of each small area;

[0041] Extracting features from the image according to the image information, comparing the extracted features with the features of the pests to obtain the pests in the image, and counting the pests to obtain information on the number of pests in each small area;

[0042] If the number of pests is greater than the preset number, the small area is marked as the area to be sprayed, and the location information of the area to be sprayed is obtained and marked in the target farmland map model;

[0043] Based on the location information of the area to be sprayed, set the starting and ending points of the spraying route of the plant protection drone;

[0044] Based on the Dijkstra algorithm, the location information of the area to be sprayed is searched to obtain the shortest path for the plant protection drone to spray pesticides and form a spraying route;

[0045] According to the number of pests, the location of the area to be sprayed, and the type of crops, the amount of pesticides applied by the plant protection drone at different locations can be adjusted in real time.

[0046] In this scheme, the effect of pesticide application on crops after the pesticide application operation is judged as follows:

[0047] Selecting a preset percentage of post-pesticide application video data within a preset time period of the area after the pesticide application operation, and extracting post-pesticide application video frame data;

[0048] Calculate the optical flow vector of the video frame data after drug administration to obtain the optical flow field;

[0049] Perform motion filtering on the optical flow field to obtain the optical flow information of pest movement;

[0050] Based on the optical flow information, the optical flow field is thresholded and the connected pixels are combined into the pest movement trajectory;

[0051] Count the movement trajectories of pests to obtain the number of pests after application of pesticides;

[0052] Compare the number of pests after application with the number of pests enriched to determine the effect of the application.

[0053] The second aspect of the present invention further provides a pesticide application control system for a plant protection UAV, the system comprising: a memory and a processor, the memory comprising a pesticide application control method program for the plant protection UAV, and when the pesticide application control method program for the plant protection UAV is executed by the processor, the following steps are implemented:

[0054] Obtaining basic information of the target farmland, wherein the basic information of the farmland includes location information of the target farmland, crop type information, and pest information;

[0055] Determine the pest enrichment situation based on image recognition technology and generate pest enrichment report;

[0056] Obtain the historical pesticide dosage of each crop planting area in the target area, and generate a pest control plan based on the historical pesticide dosage and pest enrichment report;

[0057] Determine the best time for pesticide application by plant protection drone according to the pest control plan;

[0058] Apply pesticides to crops based on pest control plans and optimal application time periods;

[0059] Determine the effect of pesticide application on crops after the application.

[0060] In this scheme, the historical pesticide dosage of each crop planting area in the target area is obtained, and a pest control scheme is generated according to the historical pesticide dosage and pest enrichment report, specifically:

[0061] Obtain the historical pesticide dosage of each crop planting area in the target area, and determine whether the historical pesticide dosage meets the pollution standard. If it does not meet the pollution standard, generate a pesticide control plan for plant protection drones;

[0062] If the pollution standard is reached, the phototropism information of the natural enemy insects of the pests in the target farmland is obtained, and the number threshold of the natural enemy insects is set;

[0063] Initialize a preset number of natural enemy insect light induction devices within a preset time period to induce the natural enemy insects to the target farmland;

[0064] Observe the quantity change information of the natural enemy insects of the pests in real time through the infrared detection device, and draw the quantity change information into a quantity-time change graph in real time;

[0065] If the number of natural enemy insects in the target farmland is not greater than the number threshold after the natural enemy insects are induced, a natural enemy insect induction network is set according to the preset number of natural enemy insect light induction devices;

[0066] If the number of natural enemy insects in the target farmland is greater than the quantity threshold after the natural enemy insects are induced, the natural enemy insect light induction equipment will be evenly reduced, and the natural enemy insects will be drained until the number of natural enemy insects in the target farmland after induction is no greater than the quantity threshold.

[0067] The third aspect of the present invention also provides a computer-readable storage medium, which includes a pesticide application control method program for a plant protection drone. When the pesticide application control method program for the plant protection drone is executed by a processor, the steps of the pesticide application control method for the plant protection drone as described in any one of the above items are implemented.

[0068] The present invention discloses a pesticide application control method, system and medium for a plant protection UAV, aiming to provide an efficient and accurate pesticide application scheme for crops. The method comprises the following steps: first, basic information of the target farmland is obtained. Secondly, the enrichment of pests is determined by using image recognition technology, and a pest enrichment report is generated. Then, a pest control scheme is generated based on historical pesticide application dosages and pest enrichment reports. According to the pest control scheme, the optimal pesticide application time period for the plant protection UAV is determined. Then, within the optimal pesticide application time period, pesticide application is performed on crops. Finally, the effect of the crops after the pesticide application operation is judged to evaluate the effect of the pesticide application. The present invention realizes the integration of farmland information acquisition, pest monitoring, pesticide application time determination and pesticide application effect evaluation, thereby improving the efficiency and accuracy of pesticide application by plant protection UAVs. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 A flow chart showing a method for controlling the application of pesticides by a plant protection UAV according to the present invention is shown;

[0070] Figure 2 The flow chart of generating a pest control scheme according to the present invention is shown;

[0071] Figure 3 The present invention shows a flow chart of applying pesticides to crops;

[0072] Figure 4 A block diagram of a pesticide application control system of a crop protection UAV of the present invention is shown. DETAILED DESCRIPTION

[0073] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0074] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0075] Figure 1 A flow chart of a pesticide application control method of a plant protection UAV of the present invention is shown.

[0076] like Figure 1 As shown, the first aspect of the present invention provides a pesticide application control method of a plant protection UAV, comprising:

[0077] S102, obtaining basic information of the target farmland, wherein the basic information of the farmland includes location information of the target farmland, crop type information, and pest information;

[0078] S104, judging the pest enrichment situation based on image recognition technology and generating a pest enrichment report;

[0079] S106, obtaining the historical pesticide dosage of each crop planting area in the target area, and generating a pest control plan according to the historical pesticide dosage and the pest enrichment report;

[0080] S108, determining the best time period for applying pesticides by the plant protection drone according to the pest control plan;

[0081] S110, applying pesticides to crops based on the pest control plan and the optimal application time period;

[0082] S112, judging the effect of the pesticide application on the crops after the pesticide application operation.

[0083] According to an embodiment of the present invention, the basic information of the target farmland is obtained, and the basic information of the farmland includes the location information of the target farmland, the crop type information, and the pest information, specifically:

[0084] Based on the GPS positioning device, the target farmland location information is obtained, the location information includes longitude and latitude data, a map model is constructed, the target farmland location information is imported into the map model, a target farmland map model is generated, and the map model is displayed on a preset display;

[0085] Acquire crop type information according to farmland planting information, wherein the crop type information includes crop name and growth height;

[0086] According to the farmland survey data, the pest information in the farmland is obtained, and the pest information includes the pest name and pest characteristics.

[0087] According to an embodiment of the present invention, the pest enrichment situation is judged based on image recognition technology, and a pest enrichment report is generated, specifically:

[0088] Acquire video data of a process of performing an enrichment operation on pests in a preset area, extract video frame images of the video data, and obtain a video frame image set;

[0089] Extracting pixel information of each frame image of the video frame image set;

[0090] Calculate a pixel covariance matrix of the video frame image set according to the pixel information, and calculate an eigenvector and an eigenvalue of the video frame image set according to the pixel covariance matrix;

[0091] The eigenvalues ​​are sorted from large to small, and the first eigenvalue is selected as the principal component of the eigenvector, and the original pixel information is projected onto the principal component to obtain a video frame image set data of a first dimensionality reduction;

[0092] Calculate the similarity matrix of the video frame image set data with a reduced dimension, use the similarity matrix to calculate the conditional probability between data points, and obtain a probability distribution graph;

[0093] Initialize the data point positions of the second dimension reduction, and calculate the conditional probability between the data points after the second dimension reduction according to the probability distribution diagram;

[0094] The KL divergence value is calculated according to the conditional probability between the probability distribution graph and the data points of the secondary dimensionality reduction, and the position of the data points of the secondary dimensionality reduction is adjusted cyclically until the KL divergence value reaches a preset divergence value, thereby obtaining a set of secondary dimensionality reduction video frame images;

[0095] Build image processing models based on image recognition technology;

[0096] Importing each video frame image of the secondary dimensionality reduction video frame image set into the image processing model, marking the position of each pest with a dot, and obtaining a marked image;

[0097] The enrichment degree of the pests is determined according to the marked images, and a pest enrichment report is generated, wherein the pest enrichment report includes the number of the pests in the target farmland.

[0098] It should be noted that in the embodiment of the present invention, due to the huge amount of data in the video frame image set, the video frame image set is first subjected to a dimensionality reduction process through PCA, and then the video frame image set is subjected to a secondary dimensionality reduction process using t-sne, and finally useful video frame data is obtained, and redundant data in the video frame data is eliminated, thereby improving the data processing efficiency of the system, further improving the data reception delay phenomenon caused by data processing of the plant protection UAV, and improving the pesticide application accuracy of the plant protection UAV.

[0099] Figure 2 The flowchart of generating the pest control scheme of the present invention is shown.

[0100] According to an embodiment of the present invention, the acquisition of historical pesticide dosages for each crop planting area in the target area and the generation of a pest control plan based on the historical pesticide dosages and pest enrichment reports are specifically as follows:

[0101] S202, obtaining the historical pesticide dosage of each crop planting area in the target area, determining whether the historical pesticide dosage reaches the pollution standard, and if it does not reach the pollution standard, generating a pesticide control plan for plant protection drones;

[0102] S204, if the pollution standard is reached, obtaining the phototropism information of the natural enemy insects of the pests in the target farmland, and setting a number threshold of the natural enemy insects;

[0103] S206, initializing a preset number of natural enemy insect light induction devices within a preset time period to induce the natural enemy insects to the target farmland;

[0104] S208, observing the quantity change information of the natural enemy insects of the pests in real time through an infrared detection device, and drawing the quantity change information into a quantity-time change graph in real time;

[0105] S210, if the number of natural enemy insects in the target farmland is not greater than the number threshold after the natural enemy insects are induced, a natural enemy insect induction network is set according to the preset number of natural enemy insect light induction devices;

[0106] S212, if the number of natural enemy insects in the target farmland is greater than the quantity threshold after the natural enemy insects are induced, the natural enemy insect light induction equipment is evenly reduced, and the natural enemy insects are drained until the number of natural enemy insects in the target farmland after induction is no greater than the quantity threshold.

[0107] It should be noted that in the embodiment of the present invention, it is first determined whether the historical pesticide dosage has reached the pollution standard. If the pollution standard is reached, the natural enemy insects of the pests are induced to prey on the pests in the target farmland, thereby avoiding further pollution to the environment by re-application of pesticides, achieving the effect of environmental protection and harmless killing of pests, and monitoring the number of natural enemy insects in real time to avoid excessive number of natural enemy insects causing harm to crops; the number threshold of the natural enemy insects is the maximum number of natural enemy insects that will not cause harm to the crops in the target farmland; the preset time period refers to the night, which avoids the influence of sunlight on the lighting equipment, resulting in poor natural enemy insect induction effect.

[0108] According to an embodiment of the present invention, the optimal time period for applying pesticides by the plant protection drone is determined according to the pest enrichment report, specifically:

[0109] Obtaining timestamp information of the video frame image;

[0110] According to the timestamp information of pest enrichment report and video frame image, draw the number-time curve of pest quantity change based on time change;

[0111] According to the change information of the quantity-time curve, determine the time period with the largest number of pests enriched;

[0112] According to the time period, the optimal time period for spraying pesticides by the plant protection drone is determined.

[0113] It should be noted that the number-time curve is drawn by analyzing the timestamp information of the video frame images and combining the number information of pests in the target farmland in the pest enrichment report. This curve can clearly show the change of pest number over time, so that farm managers can intuitively understand the dynamic changes of pest enrichment; according to the change information of the number-time curve, the system can accurately determine the time period with the largest number of pest enrichment, which is the best time period for plant protection and pesticide application. During this time period, the use of plant protection drones for pesticide application can effectively target pest groups, reduce the use of pesticides, and maximize the effect of plant protection, achieving the goals of high efficiency, energy saving, and environmental protection.

[0114] Figure 3 The flowchart of the operation of applying pesticides to crops according to the present invention is shown.

[0115] According to an embodiment of the present invention, the operation of applying pesticides to crops based on the pest enrichment report and the optimal time period for applying pesticides is specifically as follows:

[0116] S302, dividing the target farmland into N small areas in a grid manner, and displaying the divided small areas in the target farmland map model;

[0117] S304, acquiring image information of each small area based on the camera device of the plant protection drone during the optimal pesticide application time period;

[0118] S306, extracting features from the image according to the image information, comparing the extracted features with pest features, obtaining pests in the image, and counting the pests to obtain pest quantity information for each small area;

[0119] S308, if the number of pests is greater than the preset number, the small area is marked as an area to be sprayed, and the location information of the area to be sprayed is obtained and marked in the target farmland map model;

[0120] S310, setting the starting point and end point of the pesticide application route of the plant protection drone based on the location information of the area to be applied;

[0121] S312, searching the location information of the area to be sprayed based on the Dijkstra algorithm, obtaining the shortest path for the plant protection UAV to spray, and forming a spraying route;

[0122] S314, according to the number information of pests, the location information of the area to be sprayed, and the crop type information, the amount of pesticides applied by the plant protection drone at different locations is controlled in real time.

[0123] It should be noted that the target farmland is divided into N small areas in a grid manner so that the farmland can be managed and sprayed more finely; the preset number is the maximum number of pests in crops that does not cause harm to the crops set by the farmland manager; the Dijkstra algorithm is an algorithm for solving the shortest path in a graph, which can help plant protection drones plan the optimal spraying route, thereby saving time and resources; in an embodiment of the present invention, the amount of pesticide applied by the plant protection drone at different locations is controlled in real time, so that the plant protection drone can be flexibly adjusted according to actual conditions, and pesticides can be used in areas with high pest density, reducing the waste of pesticides while ensuring that crops are fully protected.

[0124] According to an embodiment of the present invention, judging the effect of pesticide application on the crops after the pesticide application operation is specifically as follows:

[0125] Selecting a preset percentage of post-pesticide application video data within a preset time period of the area after the pesticide application operation, and extracting post-pesticide application video frame data;

[0126] Calculate the optical flow vector of the video frame data after drug administration to obtain the optical flow field;

[0127] Perform motion filtering on the optical flow field to obtain the optical flow information of pest movement;

[0128] Based on the optical flow information, the optical flow field is thresholded and the connected pixels are combined into the pest movement trajectory;

[0129] Count the movement trajectories of pests to obtain the number of pests after application of pesticides;

[0130] Compare the number of pests after application with the number of pests enriched to determine the effect of the application.

[0131] It should be noted that the selection of a preset percentage of the area after the pesticide application operation to obtain video data to judge the effect of the pesticide application can reduce the amount of data processing; the optical flow vector refers to the displacement of pixels in time in a continuous image, and the optical flow field is an image composed of the optical flow vectors of all pixels. By calculating the optical flow field, the movement of each pixel in the image can be understood; by comparing the number of pests after pesticide application with the number of pest enrichment, the effect of the pesticide application operation can be objectively evaluated. If the number of pests after pesticide application is significantly reduced and is smaller than the number of pest enrichment, it means that the pesticide application operation has achieved good results and successfully controlled the number of pests. On the contrary, if the number of pests is still large after pesticide application, it may be necessary to further optimize the pesticide application plan to improve the effect of pesticide application.

[0132] According to an embodiment of the present invention, it also includes:

[0133] Acquire multispectral training data of crops in different health states, wherein the health states include healthy, sub-healthy, and unhealthy;

[0134] Establish a plant health assessment model and import multispectral training data into the plant health assessment model for training;

[0135] Acquire multispectral data of crops in target farmland through multispectral sensors;

[0136] Preprocessing the multispectral data, wherein the preprocessing includes radiation correction and noise removal;

[0137] Import the pre-processed multispectral data into the plant health assessment model to conduct plant health assessment and determine the health status of the plant;

[0138] Control the start time and spraying time of the plant protection drone according to the health status of the plants.

[0139] It should be noted that, in the embodiment of the present invention, by establishing a plant health assessment model to assess the health status of plants, human resources for monitoring the health status of crops are reduced, and the management efficiency of crops is improved.

[0140] In addition, the acquisition of the historical pesticide dosage of each crop planting area in the target area and the determination of whether the historical pesticide dosage reaches the pollution standard specifically include the following steps:

[0141] Through the farmland pesticide application data, the historical pesticide dosage of each crop planting area in the target area, the degradation cycle of the applied pesticide in the soil, and the pesticide spraying soil residue were obtained;

[0142] Determine the pollution standards of pesticide components in soil according to environmental protection standards;

[0143] The soil pesticide deposition degree is predicted based on the historical pesticide dosage, the degradation cycle of the applied pesticide in the soil, and the residual amount of pesticide sprayed in the soil;

[0144] The soil pesticide deposition degree is judged by the pollution standard. If the pesticide deposition degree exceeds the pollution standard, it is determined that the historical application dosage has reached the pollution standard.

[0145] It should be noted that by obtaining historical pesticide application dosages to predict soil pesticide deposition, judging whether the farmland meets the pollution standards based on the soil pesticide deposition, it saves the testing of farmland soil, effectively saves time, and improves the efficiency of pollution assessment.

[0146] According to an embodiment of the present invention, it also includes:

[0147] Obtain the growth cycle information, planting time information, and historical environmental data of crops in the target farmland;

[0148] Predict the environmental data of crops during a preset growth cycle based on the crop growth cycle information, planting time information, and historical environmental data. The environmental data includes temperature, humidity, wind speed, and rainfall.

[0149] Obtain information on pest species that appear at different growth stages during the historical growth cycle of crops;

[0150] Acquire information on species of natural enemy insects of the pests according to information on species of the pests, and retrieve data on suitable environments for the survival of the natural enemy insects;

[0151] Comparing the environmental data of the crops during the preset growth cycle with the environmental data suitable for the survival of natural enemy insects, the natural enemy insect species suitable for the preset growth cycle are obtained;

[0152] The phototaxis information of the suitable natural enemy insect species is obtained, and the suitable natural enemy insect species are induced to the target farmland during the preset growth period of the crop.

[0153] It should be noted that the types of pests are different in different growth stages of crops, so the types of natural enemy insects that need to be attracted to the farmland to prey on pests will also be different. Since the suitable living environments of different natural enemy insects are different, in order to avoid poor induction efficiency of natural enemy insects, in an embodiment of the present invention, based on the predicted environmental data of crops in different growth cycles, natural enemy insects that are suitable for survival in the environmental data can be induced to improve the induction efficiency of natural enemy insects.

[0154] Figure 4 A block diagram of a pesticide application control system of a crop protection UAV of the present invention is shown.

[0155] The second aspect of the present invention further provides a pesticide application control system 4 for a plant protection UAV, the system comprising: a memory 41 and a processor 42, wherein the memory comprises a pesticide application control method program for the plant protection UAV, and when the pesticide application control method program for the plant protection UAV is executed by the processor, the following steps are implemented:

[0156] Obtaining basic information of the target farmland, wherein the basic information of the farmland includes location information of the target farmland, crop type information, and pest information;

[0157] Determine the pest enrichment situation based on image recognition technology and generate pest enrichment report;

[0158] Obtain the historical pesticide dosage of each crop planting area in the target area, and generate a pest control plan based on the historical pesticide dosage and pest enrichment report;

[0159] Determine the best time for pesticide application by plant protection drone according to the pest control plan;

[0160] Apply pesticides to crops based on pest control plans and optimal application time periods;

[0161] Determine the effect of pesticide application on crops after the application.

[0162] According to an embodiment of the present invention, the basic information of the target farmland is obtained, and the basic information of the farmland includes the location information of the target farmland, the crop type information, and the pest information, specifically:

[0163] Based on the GPS positioning device, the target farmland location information is obtained, the location information includes longitude and latitude data, a map model is constructed, the target farmland location information is imported into the map model, a target farmland map model is generated, and the map model is displayed on a preset display;

[0164] Acquire crop type information according to farmland planting information, wherein the crop type information includes crop name and growth height;

[0165] According to the farmland survey data, the pest information in the farmland is obtained, and the pest information includes the pest name and pest characteristics.

[0166] According to an embodiment of the present invention, the pest enrichment situation is judged based on image recognition technology, and a pest enrichment report is generated, specifically:

[0167] Acquire video data of a process of performing an enrichment operation on pests in a preset area, extract video frame images of the video data, and obtain a video frame image set;

[0168] Extracting pixel information of each frame image of the video frame image set;

[0169] Calculate a pixel covariance matrix of the video frame image set according to the pixel information, and calculate an eigenvector and an eigenvalue of the video frame image set according to the pixel covariance matrix;

[0170] The eigenvalues ​​are sorted from large to small, and the first eigenvalue is selected as the principal component of the eigenvector, and the original pixel information is projected onto the principal component to obtain a video frame image set data of a first dimensionality reduction;

[0171] Calculate the similarity matrix of the video frame image set data with a reduced dimension, use the similarity matrix to calculate the conditional probability between data points, and obtain a probability distribution graph;

[0172] Initialize the data point positions of the second dimension reduction, and calculate the conditional probability between the data points after the second dimension reduction according to the probability distribution diagram;

[0173] The KL divergence value is calculated according to the conditional probability between the probability distribution graph and the data points of the secondary dimensionality reduction, and the position of the data points of the secondary dimensionality reduction is adjusted cyclically until the KL divergence value reaches a preset divergence value, thereby obtaining a set of secondary dimensionality reduction video frame images;

[0174] Build image processing models based on image recognition technology;

[0175] Importing each video frame image of the secondary dimensionality reduction video frame image set into the image processing model, marking the position of each pest with a dot, and obtaining a marked image;

[0176] The enrichment degree of the pests is determined according to the marked images, and a pest enrichment report is generated, wherein the pest enrichment report includes the number of the pests in the target farmland.

[0177] It should be noted that in the embodiment of the present invention, due to the huge amount of data in the video frame image set, the video frame image set is first subjected to a dimensionality reduction process through PCA, and then the video frame image set is subjected to a secondary dimensionality reduction process using t-sne, and finally useful video frame data is obtained, and redundant data in the video frame data is eliminated, thereby improving the data processing efficiency of the system, further improving the data reception delay phenomenon caused by data processing of the plant protection UAV, and improving the pesticide application accuracy of the plant protection UAV.

[0178] According to an embodiment of the present invention, the acquisition of historical pesticide dosages for each crop planting area in the target area and the generation of a pest control plan based on the historical pesticide dosages and pest enrichment reports are specifically as follows:

[0179] Obtain the historical pesticide dosage of each crop planting area in the target area, and determine whether the historical pesticide dosage meets the pollution standard. If it does not meet the pollution standard, generate a pesticide control plan for plant protection drones;

[0180] If the pollution standard is reached, the phototropism information of the natural enemy insects of the pests in the target farmland is obtained, and the number threshold of the natural enemy insects is set;

[0181] Initialize a preset number of natural enemy insect light induction devices within a preset time period to induce the natural enemy insects to the target farmland;

[0182] Observe the quantity change information of the natural enemy insects of the pests in real time through the infrared detection device, and draw the quantity change information into a quantity-time change graph in real time;

[0183] If the number of natural enemy insects in the target farmland is not greater than the number threshold after the natural enemy insects are induced, a natural enemy insect induction network is set according to the preset number of natural enemy insect light induction devices;

[0184] If the number of natural enemy insects in the target farmland is greater than the quantity threshold after the natural enemy insects are induced, the natural enemy insect light induction equipment will be evenly reduced, and the natural enemy insects will be drained until the number of natural enemy insects in the target farmland after induction is no greater than the quantity threshold.

[0185] It should be noted that in the embodiment of the present invention, it is first determined whether the historical pesticide dosage has reached the pollution standard. If the pollution standard is reached, the natural enemy insects of the pests are induced to prey on the pests in the target farmland, thereby avoiding further pollution to the environment by re-application of pesticides, achieving the effect of environmental protection and harmless killing of pests, and monitoring the number of natural enemy insects in real time to avoid excessive number of natural enemy insects causing harm to crops; the number threshold of the natural enemy insects is the maximum number of natural enemy insects that will not cause harm to the crops in the target farmland; the preset time period refers to the night, which avoids the influence of sunlight on the lighting equipment, resulting in poor natural enemy insect induction effect.

[0186] According to an embodiment of the present invention, the optimal time period for applying pesticides by the plant protection drone is determined according to the pest enrichment report, specifically:

[0187] Obtaining timestamp information of the video frame image;

[0188] According to the timestamp information of pest enrichment report and video frame image, draw the number-time curve of pest quantity change based on time change;

[0189] According to the change information of the quantity-time curve, determine the time period with the largest number of pests enriched;

[0190] According to the time period, the optimal time period for spraying pesticides by the plant protection drone is determined.

[0191] It should be noted that the number-time curve is drawn by analyzing the timestamp information of the video frame images and combining the number information of pests in the target farmland in the pest enrichment report. This curve can clearly show the change of pest number over time, so that farm managers can intuitively understand the dynamic changes of pest enrichment; according to the change information of the number-time curve, the system can accurately determine the time period with the largest number of pest enrichment, which is the best time period for plant protection and pesticide application. During this time period, the use of plant protection drones for pesticide application can effectively target pest groups, reduce the use of pesticides, and maximize the effect of plant protection, achieving the goals of high efficiency, energy saving, and environmental protection.

[0192] According to an embodiment of the present invention, the operation of applying pesticides to crops based on the pest enrichment report and the optimal time period for applying pesticides is specifically as follows:

[0193] Dividing the target farmland into N small areas in a grid manner, and displaying the divided small areas in the target farmland map model;

[0194] During the optimal application period, the camera device of the plant protection drone acquires image information of each small area;

[0195] Extracting features from the image according to the image information, comparing the extracted features with the features of the pests to obtain the pests in the image, and counting the pests to obtain information on the number of pests in each small area;

[0196] If the number of pests is greater than the preset number, the small area is marked as the area to be sprayed, and the location information of the area to be sprayed is obtained and marked in the target farmland map model;

[0197] Based on the location information of the area to be sprayed, set the starting and ending points of the spraying route of the plant protection drone;

[0198] Based on the Dijkstra algorithm, the location information of the area to be sprayed is searched to obtain the shortest path for the plant protection drone to spray pesticides and form a spraying route;

[0199] According to the number of pests, the location of the area to be sprayed, and the type of crops, the amount of pesticides applied by the plant protection drone at different locations can be adjusted in real time.

[0200] It should be noted that the target farmland is divided into N small areas in a grid manner so that the farmland can be managed and sprayed more finely; the preset number is the maximum number of pests in crops that does not cause harm to the crops set by the farmland manager; the Dijkstra algorithm is an algorithm for solving the shortest path in a graph, which can help plant protection drones plan the optimal spraying route, thereby saving time and resources; in an embodiment of the present invention, the amount of pesticide applied by the plant protection drone at different locations is controlled in real time, so that the plant protection drone can be flexibly adjusted according to actual conditions, and pesticides can be used in areas with high pest density, reducing the waste of pesticides while ensuring that crops are fully protected.

[0201] According to an embodiment of the present invention, judging the effect of pesticide application on the crops after the pesticide application operation is specifically as follows:

[0202] Selecting a preset percentage of post-pesticide application video data within a preset time period of the area after the pesticide application operation, and extracting post-pesticide application video frame data;

[0203] Calculate the optical flow vector of the video frame data after drug administration to obtain the optical flow field;

[0204] Perform motion filtering on the optical flow field to obtain the optical flow information of pest movement;

[0205] Based on the optical flow information, the optical flow field is thresholded and the connected pixels are combined into the pest movement trajectory;

[0206] Count the movement trajectories of pests to obtain the number of pests after application of pesticides;

[0207] Compare the number of pests after application with the number of pests enriched to determine the effect of the application.

[0208] It should be noted that the selection of a preset percentage of the area after the pesticide application operation to obtain video data to judge the effect of the pesticide application can reduce the amount of data processing; the optical flow vector refers to the displacement of pixels in time in a continuous image, and the optical flow field is an image composed of the optical flow vectors of all pixels. By calculating the optical flow field, the movement of each pixel in the image can be understood; by comparing the number of pests after pesticide application with the number of pest enrichment, the effect of the pesticide application operation can be objectively evaluated. If the number of pests after pesticide application is significantly reduced and is smaller than the number of pest enrichment, it means that the pesticide application operation has achieved good results and successfully controlled the number of pests. On the contrary, if the number of pests is still large after pesticide application, it may be necessary to further optimize the pesticide application plan to improve the effect of pesticide application.

[0209] According to an embodiment of the present invention, it also includes:

[0210] Acquire multispectral training data of crops in different health states, wherein the health states include healthy, sub-healthy, and unhealthy;

[0211] Establish a plant health assessment model and import multispectral training data into the plant health assessment model for training;

[0212] Acquire multispectral data of crops in target farmland through multispectral sensors;

[0213] Preprocessing the multispectral data, wherein the preprocessing includes radiation correction and noise removal;

[0214] Import the pre-processed multispectral data into the plant health assessment model to conduct plant health assessment and determine the health status of the plant;

[0215] Control the start time and spraying time of the plant protection drone according to the health status of the plants.

[0216] It should be noted that, in the embodiment of the present invention, by establishing a plant health assessment model to assess the health status of plants, human resources for monitoring the health status of crops are reduced, and the management efficiency of crops is improved.

[0217] In addition, the acquisition of the historical pesticide dosage of each crop planting area in the target area and the determination of whether the historical pesticide dosage reaches the pollution standard specifically include the following steps:

[0218] Through the farmland pesticide application data, the historical pesticide dosage of each crop planting area in the target area, the degradation cycle of the applied pesticide in the soil, and the pesticide spraying soil residue were obtained;

[0219] Determine the pollution standards of pesticide components in soil according to environmental protection standards;

[0220] The soil pesticide deposition degree is predicted based on the historical pesticide dosage, the degradation cycle of the applied pesticide in the soil, and the residual amount of pesticide sprayed in the soil;

[0221] The soil pesticide deposition degree is judged by the pollution standard. If the pesticide deposition degree exceeds the pollution standard, it is determined that the historical application dosage has reached the pollution standard.

[0222] It should be noted that by obtaining historical pesticide application dosages to predict soil pesticide deposition, judging whether the farmland meets the pollution standards based on the soil pesticide deposition, it saves the testing of farmland soil, effectively saves time, and improves the efficiency of pollution assessment.

[0223] According to an embodiment of the present invention, it also includes:

[0224] Obtain the growth cycle information, planting time information, and historical environmental data of crops in the target farmland;

[0225] Predict the environmental data of crops during a preset growth cycle based on the crop growth cycle information, planting time information, and historical environmental data. The environmental data includes temperature, humidity, wind speed, and rainfall.

[0226] Obtain information on pest species that appear at different growth stages during the historical growth cycle of crops;

[0227] Acquire information on species of natural enemy insects of the pests according to information on species of the pests, and retrieve data on suitable environments for the survival of the natural enemy insects;

[0228] Comparing the environmental data of the crops during the preset growth cycle with the environmental data suitable for the survival of natural enemy insects, the natural enemy insect species suitable for the preset growth cycle are obtained;

[0229] The phototaxis information of the suitable natural enemy insect species is obtained, and the suitable natural enemy insect species are induced to the target farmland during the preset growth period of the crop.

[0230] It should be noted that the types of pests are different in different growth stages of crops, so the types of natural enemy insects that need to be attracted to the farmland to prey on pests will also be different. Since the suitable living environments of different natural enemy insects are different, in order to avoid poor induction efficiency of natural enemy insects, in an embodiment of the present invention, based on the predicted environmental data of crops in different growth cycles, natural enemy insects that are suitable for survival in the environmental data can be induced to improve the induction efficiency of natural enemy insects.

[0231] The third aspect of the present invention also provides a computer-readable storage medium, which includes a pesticide application control method program for a plant protection drone. When the pesticide application control method program for a plant protection drone is executed by a processor, the steps of the pesticide application control method for a plant protection drone as described in any one of the above items are implemented.

[0232] The present invention discloses a pesticide application control method, system and medium for a plant protection UAV, aiming to provide an efficient and accurate pesticide application scheme for crops. The method comprises the following steps: first, basic information of the target farmland is obtained. Secondly, the enrichment of pests is determined by using image recognition technology, and a pest enrichment report is generated. Then, a pest control scheme is generated based on historical pesticide application dosages and pest enrichment reports. According to the pest control scheme, the optimal pesticide application time period for the plant protection UAV is determined. Then, within the optimal pesticide application time period, pesticide application is performed on crops. Finally, the effect of the crops after the pesticide application operation is judged to evaluate the effect of the pesticide application. The present invention realizes the integration of farmland information acquisition, pest monitoring, pesticide application time determination and pesticide application effect evaluation, thereby improving the efficiency and accuracy of pesticide application by plant protection UAVs.

[0233] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0234] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0235] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0236] Those skilled in the art can understand that: all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, and other media that can store program codes.

[0237] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0238] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for controlling the application of pesticides by a plant protection drone, characterized in that: The following steps are involved: Obtaining basic information of the target farmland, wherein the basic information of the farmland includes location information of the target farmland, crop type information, and pest information; Determine the pest enrichment situation based on image recognition technology and generate pest enrichment report; Obtain the historical pesticide dosage of each crop planting area in the target area, and generate a pest control plan based on the historical pesticide dosage and pest enrichment report; Determine the best time for pesticide application by plant protection drone according to the pest control plan; Apply pesticides to crops based on pest control plans and optimal application time periods; Determine the effect of pesticide application on crops after the pesticide application; The image recognition technology is used to judge the pest enrichment situation and generate a pest enrichment report, specifically: Acquire video data of a process of performing an enrichment operation on pests in a preset area, extract video frame images of the video data, and obtain a video frame image set; Extracting pixel information of each frame image of the video frame image set; Calculate a pixel covariance matrix of the video frame image set according to the pixel information, and calculate an eigenvector and an eigenvalue of the video frame image set according to the pixel covariance matrix; The eigenvalues ​​are sorted from large to small, and the first eigenvalue is selected as the principal component of the eigenvector, and the original pixel information is projected onto the principal component to obtain a video frame image set data of a first dimensionality reduction; Calculate the similarity matrix of the video frame image set data with a reduced dimension, use the similarity matrix to calculate the conditional probability between data points, and obtain a probability distribution graph; Initialize the data point positions of the second dimension reduction, and calculate the conditional probability between the data points after the second dimension reduction according to the probability distribution diagram; The KL divergence value is calculated according to the conditional probability between the probability distribution graph and the data points of the secondary dimensionality reduction, and the position of the data points of the secondary dimensionality reduction is adjusted cyclically until the KL divergence value reaches a preset divergence value, thereby obtaining a set of secondary dimensionality reduction video frame images; Build image processing models based on image recognition technology; Importing each video frame image of the secondary dimensionality reduction video frame image set into the image processing model, marking the position of each pest with a dot, and obtaining a marked image; The enrichment degree of the pests is determined according to the marked images, and a pest enrichment report is generated, wherein the pest enrichment report includes the number of the pests in the target farmland.

2. The pesticide application control method of a crop protection drone according to claim 1, characterized in that: The basic information of the target farmland is obtained, and the basic information of the farmland includes the location information of the target farmland, the crop type information, and the pest information, specifically: Based on the GPS positioning device, the target farmland location information is obtained, the location information includes longitude and latitude data, a map model is constructed, the target farmland location information is imported into the map model, a target farmland map model is generated, and the map model is displayed on a preset display; Acquire crop type information according to farmland planting information, wherein the crop type information includes crop name and growth height; According to the farmland survey data, the pest information in the farmland is obtained, and the pest information includes the pest name and pest characteristics.

3. The pesticide application control method of a crop protection drone according to claim 1, characterized in that: The method of obtaining the historical pesticide dosage of each crop planting area in the target area and generating a pest control plan according to the historical pesticide dosage and the pest enrichment report is as follows: Obtain the historical pesticide dosage of each crop planting area in the target area, and determine whether the historical pesticide dosage meets the pollution standard. If it does not meet the pollution standard, generate a pesticide control plan for plant protection drones; If the pollution standard is reached, the phototropism information of the natural enemy insects of the pests in the target farmland is obtained, and the number threshold of the natural enemy insects is set; Initialize a preset number of natural enemy insect light induction devices within a preset time period to induce the natural enemy insects to the target farmland; Observe the quantity change information of the natural enemy insects of the pests in real time through the infrared detection device, and draw the quantity change information into a quantity-time change graph in real time; If the number of natural enemy insects in the target farmland is not greater than the number threshold after the natural enemy insects are induced, a natural enemy insect induction network is set according to the preset number of natural enemy insect light induction devices; If the number of natural enemy insects in the target farmland is greater than the quantity threshold after the natural enemy insects are induced, the natural enemy insect light induction equipment will be evenly reduced, and the natural enemy insects will be drained until the number of natural enemy insects in the target farmland after induction is no greater than the quantity threshold.

4. The pesticide application control method of a crop protection drone according to claim 1, characterized in that: According to the pest control plan, the best time period for applying pesticides by plant protection drones is determined as follows: Obtaining timestamp information of the video frame image; According to the timestamp information of pest enrichment report and video frame image, draw the number-time curve of pest quantity change based on time change; According to the change information of the quantity-time curve, determine the time period with the largest number of pests enriched; According to the time period, the optimal time period for spraying pesticides by the plant protection drone is determined.

5. The pesticide application control method of a crop protection drone according to claim 2, characterized in that: The operation of applying pesticides to crops based on the pest control plan and the optimal application time period is specifically as follows: Dividing the target farmland into N small areas in a grid manner, and displaying the divided small areas in the target farmland map model; During the optimal application period, the camera device of the plant protection drone acquires image information of each small area; Extracting features from the image according to the image information, comparing the extracted features with the features of the pests to obtain the pests in the image, and counting the pests to obtain information on the number of pests in each small area; If the number of pests is greater than the preset number, the small area is marked as the area to be sprayed, and the location information of the area to be sprayed is obtained and marked in the target farmland map model; Based on the location information of the area to be sprayed, set the starting and ending points of the spraying route of the plant protection drone; Based on the Dijkstra algorithm, the location information of the area to be sprayed is searched to obtain the shortest path for the plant protection drone to spray pesticides and form a spraying route; According to the number of pests, the location of the area to be sprayed, and the type of crops, the amount of pesticides applied by the plant protection drone at different locations can be adjusted in real time.

6. The pesticide application control method of a crop protection drone according to claim 1, characterized in that: The method of judging the effect of pesticide application on the crops after the pesticide application operation is as follows: Selecting a preset percentage of post-pesticide application video data within a preset time period of the area after the pesticide application operation, and extracting post-pesticide application video frame data; Calculate the optical flow vector of the video frame data after drug administration to obtain the optical flow field; Perform motion filtering on the optical flow field to obtain the optical flow information of pest movement; Based on the optical flow information, the optical flow field is thresholded and the connected pixels are combined into the pest movement trajectory; Count the movement trajectories of pests to obtain the number of pests after application of pesticides; Compare the number of pests after application with the number of pests enriched to determine the effect of the application.

7. A pesticide application control system for a crop protection drone, characterized in that: The pesticide application control system of the plant protection UAV includes a storage device and a processor. The storage device includes a pesticide application control method program of the plant protection UAV. When the pesticide application control method program of the plant protection UAV is executed by the processor, the following steps are implemented: Obtaining basic information of the target farmland, wherein the basic information of the farmland includes location information of the target farmland, crop type information, and pest information; Determine the pest enrichment situation based on image recognition technology and generate pest enrichment report; Obtain the historical pesticide dosage of each crop planting area in the target area, and generate a pest control plan based on the historical pesticide dosage and pest enrichment report; Determine the best time for pesticide application by plant protection drone according to the pest control plan; Apply pesticides to crops based on pest control plans and optimal application time periods; Determine the effect of pesticide application on crops after the pesticide application; The image recognition technology is used to judge the pest enrichment situation and generate a pest enrichment report, specifically: Acquire video data of a process of performing an enrichment operation on pests in a preset area, extract video frame images of the video data, and obtain a video frame image set; Extracting pixel information of each frame image of the video frame image set; Calculate a pixel covariance matrix of the video frame image set according to the pixel information, and calculate an eigenvector and an eigenvalue of the video frame image set according to the pixel covariance matrix; The eigenvalues ​​are sorted from large to small, and the first eigenvalue is selected as the principal component of the eigenvector, and the original pixel information is projected onto the principal component to obtain a video frame image set data of a first dimensionality reduction; Calculate the similarity matrix of the video frame image set data with a reduced dimension, use the similarity matrix to calculate the conditional probability between data points, and obtain a probability distribution graph; Initialize the data point positions of the second dimension reduction, and calculate the conditional probability between the data points after the second dimension reduction according to the probability distribution diagram; The KL divergence value is calculated according to the conditional probability between the probability distribution graph and the data points of the secondary dimensionality reduction, and the position of the data points of the secondary dimensionality reduction is adjusted cyclically until the KL divergence value reaches a preset divergence value, thereby obtaining a set of secondary dimensionality reduction video frame images; Build image processing models based on image recognition technology; Importing each video frame image of the secondary dimensionality reduction video frame image set into the image processing model, marking the position of each pest with a dot, and obtaining a marked image; The enrichment degree of the pests is determined according to the marked images, and a pest enrichment report is generated, wherein the pest enrichment report includes the number of the pests in the target farmland.

8. The pesticide application control system of a crop protection drone according to claim 7, characterized in that: The method of obtaining the historical pesticide dosage of each crop planting area in the target area and generating a pest control plan according to the historical pesticide dosage and the pest enrichment report is as follows: Obtain the historical pesticide dosage of each crop planting area in the target area, and determine whether the historical pesticide dosage meets the pollution standard. If it does not meet the pollution standard, generate a pesticide control plan for plant protection drones; If the pollution standard is reached, the phototropism information of the natural enemy insects of the pests in the target farmland is obtained, and the number threshold of the natural enemy insects is set; Initialize a preset number of natural enemy insect light induction devices within a preset time period to induce the natural enemy insects to the target farmland; Observe the quantity change information of the natural enemy insects of the pests in real time through the infrared detection device, and draw the quantity change information into a quantity-time change graph in real time; If the number of natural enemy insects in the target farmland is not greater than the number threshold after the natural enemy insects are induced, a natural enemy insect induction network is set according to the preset number of natural enemy insect light induction devices; If the number of natural enemy insects in the target farmland is greater than the quantity threshold after the natural enemy insects are induced, the natural enemy insect light induction equipment will be evenly reduced, and the natural enemy insects will be drained until the number of natural enemy insects in the target farmland after induction is no greater than the quantity threshold.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a pesticide application control method program for a plant protection UAV. When the pesticide application control method program for a plant protection UAV is executed by a processor, the pesticide application control method for a plant protection UAV as described in any one of claims 1 to 6 is implemented.

Images