A method and system for monitoring the spread of rice planthoppers and for control thereof

Abstract

This invention relates to a method and system for monitoring and controlling the spread of rice planthoppers, comprising: dividing the paddy field area to be monitored into monitoring sub-regions according to preset regional division rules; controlling a drone to collect images of rice leaves, rice canopy temperature data, and regional location information of each monitoring sub-region according to a preset flight route and monitoring cycle, forming multiple sets of time-series monitoring data corresponding one-to-one with each monitoring sub-region; combining rice leaf images and rice canopy temperature data to determine if rice planthopper infestation exists in the corresponding monitoring sub-region during the current monitoring cycle, and determining the linear spread rate and area infestation spread rate of rice planthoppers based on the regional location information and infestation area data determined from two adjacent monitoring sessions; determining the spraying dosage and spraying time for each infested sub-region based on the linear spread rate and area infestation spread rate, and controlling the drone to carry out spraying operations accordingly to curb the further spread of rice planthoppers.

Description

Technical Field

[0001] This invention relates to the field of agricultural pest and disease monitoring and control technology, and in particular to a method and system for monitoring and controlling the spread of rice planthoppers. Background Technology

[0002] Rice planthoppers are common and devastating pests in rice cultivation. They feed on rice sap and can cause lodging, reduced yields, or even crop failure when they break out on a large scale. They spread rapidly and over a wide area. Timely and accurate monitoring of their spread and the implementation of targeted control measures are key to effectively controlling the pest and reducing losses.

[0003] In current technologies, monitoring the spread rate of rice planthoppers mainly relies on manual inspections. Workers examine rice leaves for damage, capture adult and nymph planthoppers, and assess the infestation level in different areas to estimate the spread rate. Control measures primarily involve manual spraying of pesticides or uniform application of pesticides to fixed areas. Currently, this combination of monitoring and control methods has the following problems:

[0004] 1. Extremely low monitoring efficiency: Due to the limited scope of manual patrols, a large amount of manpower and resources are required for large areas of paddy fields, and the process is time-consuming. Real-time data on the spread of the disease cannot be obtained in a timely manner, resulting in delays in prevention and control.

[0005] 2. Poor accuracy: Manual judgment relies on experience, and different staff members have different judgment standards. It is also difficult to detect early signs of rice planthopper spread, which can easily lead to misjudgment and missed judgment, resulting in weak control measures.

[0006] 3. Insufficient safety: Paddy fields are mostly in a damp environment, and manual inspection and pesticide application are prone to getting stuck in the mud. In addition, there is a risk of snake and insect bites in some areas. At the same time, it is not possible to quickly cover remote and complex terrain paddy field areas.

[0007] 4. Poor control effect: The uniform application method does not take into account the differences in the spread speed and damage of rice planthoppers. There are problems such as environmental pollution and fertilizer waste caused by excessive application of pesticides, or the continuous spread of pests caused by insufficient application of pesticides.

[0008] 5. Disconnect between monitoring and prevention: In the current technology, monitoring data cannot directly guide prevention and control operations. It requires secondary analysis and the formulation of prevention and control plans by humans, which is cumbersome and has poor timeliness.

[0009] Furthermore, existing methods using fixed monitoring points can only monitor rice planthoppers in localized areas, failing to achieve full coverage and accurately capture their spread. This results in significant discrepancies between estimated and actual spread rates, hindering precise control. Existing drones are mostly used for single monitoring or single application of pesticides, failing to integrate monitoring, spread analysis, and precise control, thus unable to meet the high-efficiency control needs of large-scale rice paddies.

[0010] Therefore, in order to overcome the problems of low efficiency, poor accuracy, insufficient safety, poor control effect, and disconnect between monitoring and control in existing technologies for monitoring the spread rate of rice planthoppers, and to address the shortcomings of incomplete coverage of fixed monitoring points and the inability of existing drones with limited functionality to meet the needs of efficient control in large-scale rice fields, it is necessary to propose a method for monitoring the spread rate of rice planthoppers and controlling them using drones. This method should be efficient, accurate, comprehensive, and integrated with monitoring and control, so as to achieve real-time and accurate monitoring of the spread rate of rice planthoppers and integrated control, improve the timeliness, targeting, and safety of control, reduce environmental pollution and resource waste caused by excessive use of pesticides, effectively control the spread of pests, and ensure stable and increased rice yields. Summary of the Invention

[0011] The technical problem to be solved by the present invention is to provide a method and system for monitoring the spread rate and controlling rice planthoppers, addressing the shortcomings of the prior art.

[0012] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: A method for monitoring and controlling the spread of rice planthoppers, the method comprising:

[0013] S1. For the rice paddy area to be monitored, divide the monitoring sub-regions according to the preset regional division rules;

[0014] S2. According to the preset flight route and monitoring cycle, control the UAV to collect rice leaf images, rice canopy temperature data and regional location information covered by each monitoring sub-region, and form multiple sets of time-series monitoring data corresponding one-to-one with each monitoring sub-region.

[0015] S3. Based on the combined images of rice leaves and rice canopy temperature data, when it is determined that the corresponding monitoring sub-region has rice planthopper infestation in the current monitoring period, the linear spread rate and area infestation spread rate of rice planthoppers are determined based on the regional location information and infestation area data determined by two adjacent monitoring sessions.

[0016] S4. Based on the linear spread rate and the area infestation spread rate, determine the spraying dosage and spraying time for each infested sub-region, and control the drone to carry out spraying operations accordingly to curb the further radiation and spread of rice planthoppers.

[0017] Furthermore, in step S3, the infringed sub-regions are determined based on the following steps:

[0018] S31. Input the rice leaf images of each monitoring sub-region into the rice planthopper infestation identification model, and identify the corresponding rice planthopper infestation type, infestation severity level and infested leaf area percentage through the rice planthopper infestation identification model;

[0019] S32. When it is determined that there is a suspected rice planthopper infestation in the corresponding monitoring sub-area based on the type of rice planthopper infestation and the severity level of the infestation, the rice canopy temperature data of the corresponding monitoring sub-area is obtained.

[0020] S33. Calculate the overall confidence level based on the rice canopy temperature data and the corresponding percentage of affected leaf area, and determine the corresponding monitoring sub-region as an affected sub-region if the overall confidence level of the corresponding monitoring sub-region exceeds a preset confidence level threshold.

[0021] Furthermore, in step S31, the rice planthopper infestation identification model consists of a lightweight convolutional feature extraction backbone network, a rice planthopper-specific spatial-channel attention module, and a multi-task output head, wherein:

[0022] The lightweight convolutional feature extraction backbone network uses MobileNetV3 as its basic architecture.

[0023] The rice planthopper-specific spatial-channel attention module generates an attention mask in the spatial dimension based on the leaf regions that the rice planthopper prefers to infest, and dynamically adjusts the channel weights in the channel dimension based on the characteristics of chlorophyll degradation and water loss caused by rice planthopper infestation.

[0024] The multi-task output head simultaneously outputs the type of rice planthopper infestation, the severity level of the infestation, and the percentage of affected leaf area.

[0025] Furthermore, in step S33, the overall confidence level is calculated using the following formula:

[0026] ;

[0027] + =1;

[0028] ;

[0029] in, Indicates the overall confidence level. Indicates the code for the type of rice planthopper infestation. Indicates the severity level of the infringement, Represents the dynamic weighting coefficients for the canopy temperature dimension. This indicates the measured rice canopy temperature in the currently monitored sub-region. This represents the baseline canopy temperature when rice is healthy and free from disease. This indicates the maximum canopy temperature threshold under disease stress in rice. The dynamic weighting coefficient represents the proportion of damaged leaf area. This indicates the percentage of rice leaf area affected by damage in the current monitoring sub-region. This indicates the upper limit of the percentage of rice leaves affected by damage.

[0030] Furthermore, in step S3, the linear propagation rate is determined based on the following steps:

[0031] (1) By combining the coordinates of multiple vertices of the boundary of the initially damaged area, the coordinates of the geometric center point of the initially damaged area are calculated;

[0032] (2) Obtain the maximum straight-line distance from the boundary vertex of the infringed area to the geometric center point in two adjacent monitoring sessions, and determine the target radial expansion distance of the infringed boundary during the two adjacent monitoring sessions based on the difference between these two distance values;

[0033] (3) Based on the target expansion distance and the time interval between two adjacent monitoring, determine the linear spread rate of rice planthopper.

[0034] Furthermore, the linear spread rate of the rice planthopper was determined based on the following formula:

[0035] ;

[0036] in, This represents the linear spread rate of the rice planthopper, where k represents the distance from the geometric center point. Multiple sectors are formed by uniformly dividing the circumference around the origin. Indicates the first monitoring in the first time From the boundary vertex of the infringed region within each sector to the geometric center point The maximum straight-line distance, Indicates the number of the next monitoring session From the boundary vertex of the infringed region within each sector to the geometric center point The maximum straight-line distance, Indicates the time interval between two monitoring sessions. Indicates the first The weighting coefficient of each sector The first in the initial monitoring The area of ​​damage corresponding to each sector The total area of ​​damage as initially monitored, meeting the requirements. 1.

[0037] Furthermore, in step S3, the area infestation spread rate is determined based on the following steps: the area infestation spread rate of rice planthoppers is determined based on the ratio of the increase in the total area infested by rice planthoppers during two adjacent monitoring periods to the monitoring time interval.

[0038] Furthermore, in step S4, determining the spraying dosage and spraying time for each affected sub-region based on the linear spread rate and the area spread rate includes:

[0039] S41. Based on the linear spread rate and the area spread rate, the prevention and control level of each affected sub-region is classified by the dual-speed threshold comparison method.

[0040] S42. Based on the prevention and control level, linear spread rate, and area spread rate, the spraying dose for each affected sub-area is determined by the weighted superposition calculation method of basic dose and dual-velocity normalization.

[0041] S43. Based on the prevention and control level, linear spread rate, and area spread rate, the spraying time for each affected sub-area is determined by a time lead calculation method that combines dual-velocity normalized weighted summation with prevention and control level correction.

[0042] This application discloses a system for monitoring and controlling the spread of rice planthoppers. The system includes a monitoring area division module, a monitoring data acquisition module, a rice planthopper infestation identification module, and a control decision module, wherein:

[0043] The monitoring area division module is used to divide the paddy field area to be monitored into monitoring sub-regions according to preset area division rules.

[0044] The monitoring data acquisition module is used to control the UAV to collect images of rice leaves, rice canopy temperature data and regional location information covered by each monitoring sub-region according to the preset flight route and monitoring cycle, forming multiple sets of time-series monitoring data corresponding one-to-one with each monitoring sub-region.

[0045] The rice planthopper infestation identification module is used to integrate rice leaf images and rice canopy temperature data to determine whether there is a rice planthopper infestation problem in the corresponding monitoring sub-region during the current monitoring period. Based on the regional location information and infestation area data determined by two adjacent monitoring sessions, it determines the linear spread rate and area infestation spread rate of the rice planthopper.

[0046] The prevention and control decision module is used to determine the spraying dosage and spraying time for each infested sub-region based on the linear spread rate and the area infestation spread rate, and accordingly control the drone to carry out spraying operations to curb the further radiation and spread of rice planthoppers.

[0047] The beneficial effects of this invention are as follows: First, by dividing the paddy field into monitoring sub-regions according to preset rules, the drone can systematically collect rice leaf images, canopy temperature data, and regional location information of each sub-region according to a fixed flight route and monitoring cycle, providing structured multi-set time-series monitoring data for subsequent analysis; Second, by combining leaf images and canopy temperature data, the problem of rice planthopper infestation is accurately identified, and based on the regional location information and infestation area data of two adjacent monitoring sessions, the linear spread rate and area infestation spread rate are quantified from two dimensions: radial expansion rate and area growth rate, overcoming the defect that traditional single-direction extreme value determination is easily affected by local boundary abrupt changes; Finally, the monitoring and identification, speed quantification, and decision-making spraying are linked into a closed-loop process, realizing the integrated connection of real-time accurate monitoring and control of rice planthopper spread rate, improving the timeliness, pertinence, and safety of control. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating a method for monitoring and controlling the spread of rice planthoppers disclosed in this invention.

[0049] Figure 2 This is a schematic diagram showing the division of the monitoring sub-regions;

[0050] Figure 3 This is a schematic diagram of the structure of a system for monitoring and controlling the spread of rice planthoppers disclosed in this invention. Detailed Implementation

[0051] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0052] like Figure 1 As shown, this application discloses a method for monitoring and controlling the spread of rice planthoppers, the method comprising:

[0053] Step S1: For the rice paddy area to be monitored, divide the area into monitoring sub-regions according to the preset area division rules.

[0054] For details, please refer to Figure 2 When delineating the rice paddy area to be monitored, this application divides the area into several uniform monitoring sub-areas and marks the coordinate information of each monitoring sub-area.

[0055] Step S2: According to the preset flight route and monitoring cycle, control the UAV to collect images of rice leaves, rice canopy temperature data and regional location information covered by each monitoring sub-region, forming multiple sets of time-series monitoring data corresponding one-to-one with each monitoring sub-region.

[0056] Specifically, before controlling the drone to collect data, this application will debug the drone. The drone is equipped with a high-definition camera for collecting images of rice leaves, an infrared thermal imaging module for collecting rice canopy temperature data for each monitoring sub-region, a GPS positioning module for determining the geographical coordinates of each monitoring sub-region for zoned monitoring and positioning, and a pesticide spraying module for spraying the corresponding dosage of control agent on the identified infested sub-regions. In practice, this application will set a monitoring cycle, prepare control agents, and preset the spraying dosage corresponding to different infestation levels.

[0057] In one embodiment, the monitoring cycle is set to 2 days / time, and the flight altitude of the drone is controlled at 6 meters above the ground.

[0058] In one embodiment, the pesticide spraying module mounted on the drone uses an adjustable nozzle with a spraying flow rate range of 0.5-2L / acre, and the pesticide used is a highly efficient and low-toxicity insecticide specifically for rice planthoppers.

[0059] In one embodiment, the preset flight route is a grid-like flight route to perform a full-coverage scan of the target rice field. In order to ensure that there is no less than 30% overlap between the images collected by adjacent flight routes to achieve seamless connection of the monitoring sub-regions, and to avoid the image blurring caused by excessive flight speed, which would affect the recognition accuracy of the rice planthopper infestation identification model, this application controls the spacing between adjacent flight routes to 3 meters and controls the flight speed of the UAV monitoring to 2.5 meters per second.

[0060] Step S3: Combining rice leaf images and rice canopy temperature data, when rice planthopper infestation occurs in the corresponding monitoring sub-region during the current monitoring period, determine the linear spread rate and area infestation spread rate of rice planthoppers based on the regional location information and infestation area data determined from two adjacent monitoring sessions.

[0061] Specifically, this application first synthesizes the coordinates of all vertices of the boundary of the initially affected area, and then obtains the geometric center point by calculating the arithmetic mean of the horizontal and vertical coordinates. As a radial distance reference point; secondly, obtain the time of two adjacent monitoring sessions. , The corresponding total area of ​​damage in each monitoring sub-region , In terms of the increase in area ( ) and monitoring time interval ( The ratio of ) determines the rate of spread of area damage. Then, the 360° circle is evenly divided into k sectors with the geometric center point as the origin, and the maximum radial distance from the boundary vertex to the geometric center point of each sector is calculated in the two monitoring cycles. , And based on the proportion of the infested area of ​​each sector to the total infested area in the initial monitoring. The proportion is used as a weighting coefficient The linear spread rate is determined using a multi-sector weighted fusion formula (see subsequent calculation formulas for details). Therefore, the linear spread rate of the currently monitored sub-region and the speed of spread of the damage Both methods quantify the infestation and spread of rice planthoppers from two dimensions: radial expansion rate and area growth rate, respectively, providing a quantitative basis for determining the differences in spraying dosage and spraying time in each sub-region.

[0062] Step S4: Based on the linear spread rate and the area infestation spread rate, determine the spraying dosage and spraying time for each infested sub-region, and control the drone to carry out spraying operations accordingly to curb the further radiation and spread of rice planthoppers.

[0063] Specifically, this application first divides each infested sub-region into three control levels—Level 1 high spread rate zone, Level 2 unidirectional high spread rate zone, and Level 3 low spread rate zone—based on linear spread rate and area infestation spread rate using a dual-velocity threshold comparison method. Then, based on the control level and dual-dimensional spread rate, the spraying dose for each sub-region is determined using a method of weighted superposition of base dose and dual-velocity normalization, ensuring the dose increases linearly with the spread rate. Finally, a time lead calculation method combining dual-velocity normalized weighted summation and control level correction is used to determine the spraying lead for each infested sub-region relative to the current monitoring time. After control level correction, the actual spraying time is obtained, ensuring that Level 1 control zones are sprayed earliest and Level 3 control zones are sprayed latest. Ultimately, the drone is controlled to perform zoned spraying operations sequentially according to the differentiated spraying dose and time determined for each sub-region, achieving sufficient early spraying in high spread rate areas and reduced-volume delayed spraying in low spread rate areas to curb the further spread of rice planthoppers.

[0064] As can be seen from the above, the method for monitoring and controlling the spread of rice planthoppers disclosed in this application firstly divides the paddy field into monitoring sub-regions according to preset rules, enabling drones to systematically collect rice leaf images, canopy temperature data, and regional location information of each sub-region according to a fixed flight route and monitoring cycle, providing structured multi-set time-series monitoring data for subsequent analysis; secondly, it accurately identifies rice planthopper infestation problems by combining leaf images and canopy temperature data, and quantifies the linear spread rate and area infestation spread rate from two dimensions: radial expansion rate and area growth rate, based on the regional location information and infestation area data of two adjacent monitoring sessions, overcoming the defect that traditional single-direction extreme value determination is easily affected by local boundary abrupt changes; finally, it links monitoring and identification, speed quantification, and decision-making spraying into a closed-loop process, realizing the integrated connection of real-time accurate monitoring and control of rice planthopper spread rate, improving the timeliness, pertinence, and safety of control.

[0065] In one embodiment, in step S3, the infringed sub-region is determined based on the following steps:

[0066] Step S31: Input the rice leaf images of each monitoring sub-region into the rice planthopper infestation identification model. The rice planthopper infestation identification model identifies the corresponding rice planthopper infestation type, infestation severity level, and percentage of infested leaf area.

[0067] Specifically, the rice planthopper infestation identification model consists of a lightweight convolutional feature extraction backbone network, a rice planthopper-specific spatial-channel attention module, and a multi-task output head. The lightweight convolutional feature extraction backbone network uses MobileNetV3 as its basic architecture. The rice planthopper-specific spatial-channel attention module generates attention masks based on the leaf regions that rice planthoppers prefer to infest in the spatial dimension, and dynamically adjusts the channel weights based on the chlorophyll degradation and water loss characteristics caused by rice planthopper infestation in the channel dimension. The multi-task output head simultaneously outputs the rice planthopper infestation type, infestation severity level, and the percentage of affected leaf area. The rice planthopper infestation types include brown planthopper, white-backed planthopper, and gray planthopper, and the infestation severity levels include Level 1 (mild), Level 2 (moderate), and Level 3 (severe).

[0068] Step S32: When it is determined that there is a suspected rice planthopper infestation in the corresponding monitoring sub-area based on the type of rice planthopper infestation and the severity level of the infestation, the rice canopy temperature data of the corresponding monitoring sub-area is obtained.

[0069] Specifically, when determining whether there is a suspected rice planthopper infestation in a corresponding monitoring sub-area, this application will set a severity level for the infestation. At any given time, regardless of the value of the rice planthopper infestation type t, the corresponding monitoring sub-area is determined to be suspected of having rice planthopper infestation; when the infestation severity level... If the rice planthopper infestation type t = 1, i.e., white-backed planthopper, then it is determined that there is no suspected rice planthopper infestation in the corresponding monitoring sub-area. If the rice planthopper infestation type t ∈ {0,2}, i.e., brown planthopper or gray planthopper, then it is determined that there is suspected rice planthopper infestation in the corresponding monitoring sub-area.

[0070] Step S33: Calculate the overall confidence level based on the rice canopy temperature data and the corresponding percentage of damaged leaf area. If the overall confidence level of the corresponding monitoring sub-region exceeds a preset confidence level threshold, the corresponding monitoring sub-region is identified as an infested sub-region.

[0071] Specifically, when the overall confidence level of the corresponding monitoring sub-region exceeds the preset confidence level threshold, it indicates that the sub-region simultaneously meets the two major characteristics of abnormal plant physiological temperature and large-scale leaf damage, and the determination of pest and disease infestation is sufficiently credible. This application will mark the monitoring sub-region and identify it as an infested sub-region. Conversely, if the overall confidence level exceeds the preset confidence level threshold, it will be determined that there is no obvious pest and disease infestation in the rice in the region, or the abnormal signal is caused by irrelevant factors such as environmental noise or light interference. This application will not include it in the statistical scope of infested areas.

[0072] In one embodiment, in step S31, the rice planthopper infestation identification model consists of a lightweight convolutional feature extraction backbone network, a rice planthopper-specific spatial-channel attention module, and a multi-task output head, wherein:

[0073] The lightweight convolutional feature extraction backbone network uses MobileNetV3 as its architecture.

[0074] Specifically, the input end of the MobileNetV3 receives rice leaf images captured by the high-definition camera and scaled to 224×224 pixels, and the output end generates a 512-dimensional feature vector and inputs it into the rice planthopper-specific spatial-channel attention module.

[0075] The rice planthopper-specific spatial-channel attention module generates attention masks in the spatial dimension based on the leaf regions that the rice planthopper prefers to infest, and dynamically adjusts the channel weights in the channel dimension based on the chlorophyll degradation and water loss characteristics caused by the rice planthopper infestation.

[0076] Specifically, in the spatial dimension, since rice planthoppers prefer to infest the lower and middle parts of rice leaves and the leaf sheath junction, this application assigns a higher spatial weight to these areas and a lower spatial weight to non-infestation-prone areas such as leaf edges, thereby suppressing background interference. In the channel dimension, since rice planthopper infestation leads to a decrease in chlorophyll content and water loss in leaves, the corresponding image features are most prominent in the green and near-infrared bands. Therefore, this application assigns a higher weight to the feature channels corresponding to these bands and a lower weight to feature channels unrelated to infestation, such as textures and edges, thereby highlighting infestation-related features.

[0077] The multi-task output head simultaneously outputs the type of rice planthopper infestation, the severity level of the infestation, and the percentage of affected leaf area.

[0078] Specifically, the multi-task output head includes three parallel branches. In the type branch, a softmax activation function is used to output three-class classification probabilities to determine the rice planthopper infestation type as brown planthopper, white-backed planthopper, or gray planthopper. In the grade branch, a softmax activation function is used to output three-class classification probabilities to determine the infestation severity grade as level one (mild), level two (moderate), or level three (severe). In the area branch, a sigmoid activation function is used to output a continuous value of the percentage of affected leaf area.

[0079] In one embodiment, in step S33, the overall confidence level is calculated using the following formula:

[0080] ;

[0081] + =1;

[0082] = ;

[0083] in, Indicates the overall confidence level. Indicates the code for the type of rice planthopper infestation. Indicates the severity level of the infringement, Represents the dynamic weighting coefficients for the canopy temperature dimension. This indicates the measured rice canopy temperature in the currently monitored sub-region. This represents the baseline canopy temperature when rice is healthy and free from disease. This indicates the maximum canopy temperature threshold under disease stress in rice. The dynamic weighting coefficient represents the proportion of damaged leaf area. This indicates the percentage of rice leaf area affected by damage in the current monitoring sub-region. This indicates the upper limit of the percentage of rice leaves affected by damage.

[0084] It should be noted that, =0 indicates the type of brown planthopper infestation. =1 indicates the type of infestation by the white-backed planthopper. =2 indicates the type of infestation by the gray planthopper. =1 indicates a level one minor offense. =2 indicates a level two moderate injury. =3 indicates a level 3 severe attack.

[0085] Specifically, this application first performs linear normalization on the rice canopy temperature and the proportion of damaged leaf area, mapping the two indicators characterizing the stress of rice planthoppers to a unified range of 0-1. Then, based on the planthopper infestation type code and infestation severity level code, it sets dynamic complementary weights that satisfy the weight sum of 1. and The temperature weights for brown planthoppers and white-backed planthoppers increase with the level of infestation and decrease with the type of pest code. For gray planthoppers, temperature and leaf proportion are assigned equal weights. Finally, a comprehensive confidence score is obtained through linear weighted fusion, with a value between 0 and 1. The value directly reflects the reliability of the judgment that the corresponding level of rice planthopper infestation exists in the monitored sub-area.

[0086] In one embodiment, in step S3, the linear propagation rate is determined based on the following steps:

[0087] (1) By combining the coordinates of multiple vertices of the boundary of the first damaged area, the coordinates of the geometric center point of the first damaged area are calculated.

[0088] Specifically, this application will extract the coordinates of all vertices on the boundary contour of the rice planthopper infestation area identified in the initial monitoring, and record them as follows: , ... ... The form is where n is the total number of boundary vertices of the initially affected region. Let x be the x-coordinate of the i-th vertex. Let be the ordinate of the i-th vertex. Then, let be the x-coordinates of these n vertices. , ... Summing and dividing by the total number of vertices n yields the x-coordinate of the geometric center. Similarly, the ordinates of these n vertices... Summing and dividing by the total number of vertices n yields the ordinate of the geometric center point. .

[0089] (2) Obtain the maximum straight-line distance from the boundary vertex of the infringed area to the geometric center point in two adjacent monitoring sessions, and determine the target radial expansion distance of the infringed boundary during the two adjacent monitoring sessions based on the difference between these two distance values.

[0090] Specifically, this application will use the geometric center point of the initially affected area as the reference point. With a fixed reference origin, traverse all vertices of the boundary of the affected area corresponding to the first monitoring time T1. , … And calculate the distance from each vertex to the geometric center. The maximum value among the Euclidean straight-line distances is taken as the maximum straight-line distance of the first monitoring. Similarly, with the geometric center point With a fixed reference origin, traverse all vertices of the boundary of the affected area corresponding to the next monitoring time T2. , ), ( , ), ..., ( , ), and calculate the distance from each vertex to the geometric center point. The maximum value among the Euclidean straight-line distances is taken as the maximum straight-line distance for the next monitoring. Finally, based on distance With distance The difference between the two values ​​determines the radial target expansion distance of the infringement boundary during two adjacent monitoring periods. .

[0091] (3) Based on the target expansion distance and the time interval between two adjacent monitoring, determine the linear spread rate of rice planthopper.

[0092] Specifically, this application uses the geometric center point of the initially affected area. To fix the origin, the 360° circumference around it is evenly divided into k sectors; then, the data from the first monitoring of each sector is calculated separately. and the next monitoring Maximum radial distance from the vertex of the boundary of the attack to the geometric center point and And based on the proportion of the infested area of ​​each sector to the total infested area in the initial monitoring. The proportion is used as a weighting coefficient Ultimately, the multi-sector weighted fusion formula is used. The method for calculating linear spread rate, compared to the traditional approach that only takes the global maximum radial distance, eliminates the random interference of local boundary abrupt changes through spatial decomposition and area weighting, making the spread rate more reflective of the overall trend of rice planthopper infestation radiating outwards.

[0093] In one embodiment, in step S3, the linear spread rate of the rice planthopper is determined based on the following formula:

[0094] ;

[0095] in, This represents the linear spread rate of the rice planthopper, where k represents the distance from the geometric center point. Multiple sectors are formed by uniformly dividing the circumference around the origin. Indicates the first monitoring in the first time From the boundary vertex of the infringed region within each sector to the geometric center point The maximum straight-line distance, Indicates the number of the next monitoring session From the boundary vertex of the infringed region within each sector to the geometric center point The maximum straight-line distance, Indicates the time interval between two monitoring sessions. Indicates the first The weighting coefficient of each sector The first in the initial monitoring The area of ​​damage corresponding to each sector The total area of ​​damage as initially monitored, meeting the requirements. 1.

[0096] It should be noted that this application decomposes the infringed area into k sectors according to angle. After each sector independently calculates its radial expansion, the area ratio of each sector in the initial monitoring is used as the weight for weighted fusion. This makes the linear spread rate reflect the multi-directional radial expansion trend of the overall infringement, rather than the extreme value of a single direction, thus improving the robustness and representativeness of the spread rate determination.

[0097] In one embodiment, in step S3, the area infestation spread rate is determined based on the following steps: the area infestation spread rate of rice planthoppers is determined based on the ratio of the increase in the total area infested by rice planthoppers during two adjacent monitoring periods to the monitoring time interval.

[0098] Specifically, this application will first obtain the first monitoring time. The corresponding total area infested by rice planthoppers and the time of the next monitoring The corresponding total area infested by rice planthoppers Subsequently, the total area of ​​damage was measured using the data from the last monitoring session. Subtract the total area of ​​damage from the first monitoring. This allows us to obtain the increase in the total area infested by rice planthoppers between two adjacent monitoring periods. Finally, the increase in the total area of ​​damage will be... Divide by monitoring time interval The area of ​​infestation and spread of rice planthoppers was obtained. .

[0099] In one embodiment, step S4, determining the spraying dosage and spraying time for each affected sub-region based on the linear spread rate and the area spread rate, includes:

[0100] Step S41: Based on the linear spread rate and the area spread rate, the prevention and control level of each affected sub-region is classified by a dual-speed threshold comparison method.

[0101] Specifically, (1) when ≥ and ≥ At that time, set the current infringed sub-region. (1) Level I prevention and control zone (i.e., high spread rate zone); (2) When and ≥ ,or ≥ and At that time, set the current infringed sub-region. It is a secondary prevention and control zone (i.e., a zone with a one-way high spread rate); (3) when and At that time, set the current infringed sub-region. It is a level-three prevention and control zone (i.e., a low spread rate zone), among which, This represents the preset threshold for determining the linear spread rate, and its value is determined based on statistics of the historical maximum radial spread rate of rice planthoppers during the rice growing season. This represents the preset threshold for determining the rate of spread of rice planthopper infestation. Its value is determined based on the statistical analysis of the historical maximum area growth rate of rice planthoppers during the rice growing season.

[0102] Step S42: Based on the prevention and control level, linear spread rate, and area spread rate, the spraying dose for each affected sub-region is determined by using a weighted superposition calculation method that combines the base dose with dual-velocity normalization.

[0103] Specifically, each sub-region of infringement Spraying dosage Standard reference spraying dosage Together with the normalized weighted increments of the linear spread rate and the area spread rate, the calculation formula is as follows: ,in, , These are the dose weighting coefficients corresponding to the linear spread rate and the area damage spread rate, respectively, and have... > The formula is designed to reflect that the radial spread rate has a greater weight on the spray dosage than the area growth rate. The core principle of this formula is to normalize the two-dimensional spread rates by dividing them by their respective thresholds, and then weighting them together with weighted coefficients after eliminating dimensional differences. This results in the spray dosage increasing linearly with the spread rate.

[0104] Step S43: Based on the prevention and control level, linear spread rate, and area spread rate, the spraying time for each affected sub-area is determined by a time lead calculation method that combines dual-velocity normalized weighted summation with prevention and control level correction.

[0105] Specifically, each infringing sub-region i relative to the current monitoring time Spray advance Δ The value is determined by the normalized weighted sum of the linear spread rate and the area spread rate, and corrected for by the control level. The calculation formula is: Δ The spraying times for each affected sub-region i are as follows: Δ ,in, This represents the advance time factor, whose value is determined based on the estimated time from the current spread rate of rice planthoppers to the next monitoring cycle; , These are the time weighting coefficients corresponding to the linear spread rate and the area spread rate, respectively, and they have... The influence of radial spread rate on spraying timing is greater than that of area growth rate. This represents the correction coefficient for the prevention and control level, where, when the affected sub-region i is a Level 1 prevention and control zone, there is... 1.2; When the infringed sub-region i is a level-two prevention and control zone, there is 1.0; When the infringed sub-region i is a level 3 prevention and control zone, there is 0.8. The core principle of this formula is as follows: the two-dimensional spread rate is divided by the corresponding threshold for normalization to eliminate the difference in dimensions. After weighted summation by weighting coefficients, the basic advance amount is obtained by multiplying by the time advance coefficient. Then, the advance amount is adjusted by gradient through the control level correction coefficient, so that the spraying advance amount is maximized in the first-level control area and minimized in the third-level control area. This allows for priority spraying in high-spread-rate areas and delayed spraying in low-spread-rate areas, achieving dynamic matching between spraying time and the spread status of each sub-area.

[0106] In one embodiment, after completing one round of control operations, this application will continue to monitor the pests using drones and compare the rice planthopper infestation before and after control to analyze the control effect. If the infestation level in a certain area does not decrease or continues to worsen, this application will further adjust the spraying dosage, spraying time, and spraying frequency for that area until the pest is effectively controlled. If the pest is controlled, the monitoring cycle can be appropriately extended for routine monitoring to prevent a resurgence. Furthermore, the optimal spraying dosage is determined through analysis and calculation, using the following formula: Optimal Spraying Dosage Formula:

[0107] ;

[0108] in, This formula represents the previously used spraying dosage and introduces a correction factor for the control effect. Correction factor for insect population density This allows for precise calculation of drug dosage as needed.

[0109] In one embodiment, the prevention and control effect correction coefficient The calculation method is as follows: , This represents the percentage of the target area affected before prevention and control measures were implemented. This represents the percentage of affected area after pesticide application and monitoring. If the affected area has decreased compared to before treatment, then... If the dosage is <1, it can be appropriately reduced; if it does not decrease or worsens, then... If the value is greater than 1, then the dosage should be increased.

[0110] In one embodiment, the insect population density correction factor The calculation method is as follows: , The current insect population density was determined using the commonly used five-point sampling method in agriculture, and the calculation method is as follows: , , This represents the total number of sampling points. To manually count the total number of rice planthoppers at the sampling points, To manually count the number of plants at the sampling points, As the baseline insect population density for moderate infestation, in accordance with the clearly defined relationship between the degree of infestation and the weighted average number of insects per 100 plants in GB / T 15794-2009 "Specifications for Monitoring and Surveying Rice Planthoppers", and in conjunction with the three-level control decision-making system of the Ministry of Agriculture and Rural Affairs' Rice Planthopper Control Technical Regulations, the median value of the weighted average number of insects per 100 plants interval is taken as the baseline insect population density in this application. The specific values ​​are shown in Table 1 below.

[0111] Table 1

[0112]

[0113] Please refer to Figure 3 This application discloses a system for monitoring and controlling the spread of rice planthoppers. The system includes a monitoring area division module, a monitoring data acquisition module, a rice planthopper infestation identification module, and a control decision module, wherein:

[0114] The monitoring area division module is used to divide the paddy field area to be monitored into monitoring sub-regions according to preset area division rules.

[0115] The monitoring data acquisition module is used to control the UAV to collect images of rice leaves, rice canopy temperature data and regional location information covered by each monitoring sub-region according to a preset flight route and monitoring cycle, forming multiple sets of time-series monitoring data corresponding one-to-one with each monitoring sub-region.

[0116] The rice planthopper infestation identification module is used to comprehensively analyze rice leaf images and rice canopy temperature data to determine whether rice planthopper infestation exists in the corresponding monitoring sub-region during the current monitoring period. Based on the regional location information and infestation area data determined from two adjacent monitoring sessions, it determines the linear spread rate and area infestation spread rate of the rice planthopper.

[0117] The prevention and control decision module is used to determine the spraying dosage and spraying time for each infested sub-region based on the linear spread rate and the area infestation spread rate, and accordingly control the drone to carry out spraying operations to curb the further radiation and spread of rice planthoppers.

[0118] In one embodiment, the above modules are also used to implement a method for monitoring and controlling the spread of rice planthoppers as described in any of the preceding claims, and this application does not limit this method.

[0119] As can be seen from the above, the system for monitoring and controlling the spread of rice planthoppers disclosed in this application first divides the paddy field into monitoring sub-regions according to preset rules, enabling drones to systematically collect rice leaf images, canopy temperature data, and regional location information of each sub-region according to a fixed flight route and monitoring cycle, providing structured multi-set time-series monitoring data for subsequent analysis; secondly, it accurately identifies rice planthopper infestation problems by combining leaf images and canopy temperature data, and quantifies the linear spread rate and area infestation spread rate from two dimensions, namely radial expansion rate and area growth rate, based on the regional location information and infestation area data of two adjacent monitoring sessions, overcoming the defect of traditional single-direction extreme value determination being easily interfered with by local boundary abrupt changes; finally, it links monitoring and identification, speed quantification, and decision-making spraying into a closed-loop process, realizing the integrated connection of real-time accurate monitoring and control of rice planthopper spread rate, improving the timeliness, pertinence, and safety of control.

[0120] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for monitoring and controlling the spread of rice planthoppers, characterized in that, The method includes: S1. For the rice paddy area to be monitored, divide the monitoring sub-regions according to the preset regional division rules; S2. According to the preset flight route and monitoring cycle, control the UAV to collect rice leaf images, rice canopy temperature data and regional location information covered by each monitoring sub-region, and form multiple sets of time-series monitoring data corresponding one-to-one with each monitoring sub-region. S3. Based on the combined images of rice leaves and rice canopy temperature data, when it is determined that the corresponding monitoring sub-region has rice planthopper infestation in the current monitoring period, the linear spread rate and area infestation spread rate of rice planthoppers are determined based on the regional location information and infestation area data determined by two adjacent monitoring sessions. S4. Based on the linear spread rate and the area infestation spread rate, determine the spraying dosage and spraying time for each infested sub-region, and control the drone to carry out spraying operations accordingly to curb the further radiation and spread of rice planthoppers.

2. The method according to claim 1, characterized in that, In step S3, the infringed sub-regions are determined based on the following steps: S31. Input the rice leaf images of each monitoring sub-region into the rice planthopper infestation identification model, and identify the corresponding rice planthopper infestation type, infestation severity level and infested leaf area percentage through the rice planthopper infestation identification model; S32. When it is determined that there is a suspected rice planthopper infestation in the corresponding monitoring sub-area based on the type of rice planthopper infestation and the severity level of the infestation, the rice canopy temperature data of the corresponding monitoring sub-area is obtained. S33. Calculate the overall confidence level based on the rice canopy temperature data and the corresponding percentage of affected leaf area, and determine the corresponding monitoring sub-region as an affected sub-region if the overall confidence level of the corresponding monitoring sub-region exceeds a preset confidence level threshold.

3. The method according to claim 2, characterized in that, In step S31, the rice planthopper infestation identification model consists of a lightweight convolutional feature extraction backbone network, a rice planthopper-specific spatial-channel attention module, and a multi-task output head, wherein: The lightweight convolutional feature extraction backbone network uses MobileNetV3 as its basic architecture. The rice planthopper-specific spatial-channel attention module generates an attention mask in the spatial dimension based on the leaf regions that the rice planthopper prefers to infest, and dynamically adjusts the channel weights in the channel dimension based on the characteristics of chlorophyll degradation and water loss caused by rice planthopper infestation. The multi-task output head simultaneously outputs the type of rice planthopper infestation, the severity level of the infestation, and the percentage of affected leaf area.

4. The method according to claim 2, characterized in that, In step S33, the overall confidence level is calculated using the following formula: ； ； = ； in, Indicates the overall confidence level. Indicates the code for the type of rice planthopper infestation. Indicates the severity level of the infringement, Represents the dynamic weighting coefficients for the canopy temperature dimension. This indicates the measured rice canopy temperature in the currently monitored sub-region. This represents the baseline canopy temperature when rice is healthy and free from disease. This indicates the maximum canopy temperature threshold under disease stress in rice. The dynamic weighting coefficient represents the proportion of damaged leaf area. This indicates the percentage of rice leaf area affected by damage in the current monitoring sub-region. This indicates the upper limit of the percentage of rice leaves affected by damage.

5. The method according to claim 1, characterized in that, In step S3, the linear propagation rate is determined based on the following steps: (1) By combining the coordinates of multiple vertices of the boundary of the initially damaged area, the coordinates of the geometric center point of the initially damaged area are calculated; (2) Obtain the maximum straight-line distance from the boundary vertex of the infringed area to the geometric center point in two adjacent monitoring sessions, and determine the target radial expansion distance of the infringed boundary during the two adjacent monitoring sessions based on the difference between these two distance values; (3) Based on the target expansion distance and the time interval between two adjacent monitoring, determine the linear spread rate of rice planthopper.

6. The method according to claim 5, characterized in that, In step S3, the linear spread rate of the rice planthopper is determined based on the following formula: ； in, This represents the linear spread rate of the rice planthopper, where k represents the distance from the geometric center point. Multiple sectors are formed by uniformly dividing the circumference around the origin. Indicates the first monitoring in the first time From the boundary vertex of the infringed region within each sector to the geometric center point The maximum straight-line distance, Indicates the number of the next monitoring session From the boundary vertex of the infringed region within each sector to the geometric center point The maximum straight-line distance, Indicates the time interval between two monitoring sessions. Indicates the first The weighting coefficient of each sector The first in the initial monitoring The area of ​​damage corresponding to each sector The total area of ​​damage as initially monitored, meeting the requirements.

1.

7. The method according to claim 1, characterized in that, In step S3, the area infestation spread rate is determined based on the following steps: the area infestation spread rate of rice planthoppers is determined based on the ratio of the increase in the total area infested by rice planthoppers during two adjacent monitoring periods to the monitoring time interval.

8. The method according to claim 1, characterized in that, In step S4, determining the spraying dosage and spraying time for each affected sub-region based on the linear spread rate and the area spread rate includes: S41. Based on the linear spread rate and the area spread rate, the prevention and control level of each affected sub-region is classified by the dual-speed threshold comparison method. S42. Based on the prevention and control level, linear spread rate, and area spread rate, the spraying dose for each affected sub-area is determined by the weighted superposition calculation method of basic dose and dual-velocity normalization. S43. Based on the prevention and control level, linear spread rate, and area spread rate, the spraying time for each affected sub-area is determined by a time lead calculation method that combines dual-velocity normalized weighted summation with prevention and control level correction.

9. A system for monitoring and controlling the spread of rice planthoppers, characterized in that, The system includes a monitoring area division module, a monitoring data acquisition module, a rice planthopper infestation identification module, and a prevention and control decision-making module, wherein: The monitoring area division module is used to divide the paddy field area to be monitored into monitoring sub-regions according to preset area division rules. The monitoring data acquisition module is used to control the UAV to collect images of rice leaves, rice canopy temperature data and regional location information covered by each monitoring sub-region according to the preset flight route and monitoring cycle, forming multiple sets of time-series monitoring data corresponding one-to-one with each monitoring sub-region. The rice planthopper infestation identification module is used to integrate rice leaf images and rice canopy temperature data to determine whether there is a rice planthopper infestation problem in the corresponding monitoring sub-region during the current monitoring period. Based on the regional location information and infestation area data determined by two adjacent monitoring sessions, it determines the linear spread rate and area infestation spread rate of the rice planthopper. The prevention and control decision module is used to determine the spraying dosage and spraying time for each infested sub-region based on the linear spread rate and the area infestation spread rate, and accordingly control the drone to carry out spraying operations to curb the further radiation and spread of rice planthoppers.

Images