Rice planting environment monitoring method and system based on UAV technology

Abstract

The present invention discloses a rice planting environment monitoring method and system based on drone technology. The method includes: field survey, autonomous planning of monitoring routes, monitoring data collection and processing, construction of a three-dimensional model of rice fields, and optimization of the rice field growth environment. The present invention belongs to the field of smart agricultural technology, and specifically refers to a rice planting environment monitoring method and system based on drone technology. This solution uses calculated target values to plan actions to form a basic route, uses a blocking mechanism to optimize the route, can randomly unblock when the drone is trapped to ensure normal flight, and monitors and assesses risks in blind spots at the edge of the rice field; uses rice field images taken by a multi-camera to find plane points, selects seed points through Harris response values, increases the area of seed points and extracts the initial plane based on different feature information in the rice field image, and reconstructs the rice field based on grammatical rule segmentation, providing a richer and more accurate data foundation for three-dimensional reconstruction of rice fields.

Description

Technical Field

[0001] The present invention belongs to the field of smart agriculture technology, and specifically is a rice planting environment monitoring method and system based on drone technology. Background Art

[0002] A rice cultivation environment monitoring method based on drone technology utilizes drones equipped with sensors to conduct aerial photography and data collection of rice-growing areas. The collected data is analyzed and processed to provide timely information on rice growth conditions and pest and disease conditions. The goal is to utilize this monitoring information to enable farmers to more scientifically and accurately manage their fields, thereby improving rice yield and quality, reducing costs, and minimizing resource waste. However, existing drone-based rice cultivation environment monitoring methods have technical issues such as difficulty autonomously planning efficient, comprehensive, and emergency-responsive flight routes for rice field monitoring. Drones are unable to break free from obstacles when encountering special circumstances, and entering blind spots can easily lead to collision risks, resulting in unnecessary economic losses. Furthermore, the spatial structure and morphology of rice fields are unclear, making comprehensive data collection difficult. This inability to accurately capture the rice field environment makes it difficult to detect potential environmental risks and anomalies, hindering the protection of rice field ecosystems. Summary of the Invention

[0003] In view of the above situation, in order to overcome the defects of the existing technology, the present invention provides a rice planting environment monitoring method and system based on drone technology. In order to solve the technical problems that drones are difficult to independently plan efficient, comprehensive and emergency-response flight routes in rice field monitoring, and cannot break free from obstacles when trapped in special circumstances, enter visual blind spots, cause collision risks, and cause unnecessary economic losses, the method adopts the method of calculating target values to reasonably plan actions to form basic routes, utilizes the blocking mechanism to encourage drones to explore unexplored areas to optimize routes, can randomly unblock drones when trapped to ensure normal flight, and monitor blind spots at the edges of rice fields. Measurement and risk assessment; In order to address the technical problems that the spatial structure and morphology of rice fields are unclear, it is difficult to conduct comprehensive data collection, and it is impossible to obtain accurate rice field environmental conditions, which makes it difficult to detect potential environmental risks and anomalies, and is not conducive to protecting rice field ecosystems, rice field images taken by multi-cameras are used to find 2D plane points, and seed points are selected according to Harris response values. The area of seed points is increased and the initial plane is extracted according to different feature information in the rice field images. The rice fields are reconstructed based on grammatical rule segmentation. Highly compatible mobile nodes are designed to adapt to the actual environment and integrated with sensor networks to provide a richer and more accurate data foundation for three-dimensional reconstruction of rice fields.

[0004] The technical solution adopted by the present invention is as follows: The rice planting environment monitoring method based on drone technology provided by the present invention comprises the following steps:

[0005] Step S1: Field survey, specifically determining the rice field monitoring area and entering it into the UAV monitoring system;

[0006] Step S2: Autonomously plan the monitoring route. Specifically, the target value is calculated to determine the drone's action, forming the basis of the rice field monitoring route. A blocking mechanism is set to encourage the drone to explore unexplored areas to form a better route. By judging the drone's navigation status and randomly removing nearby blocking areas when trapped, the blind spots at the edge of the rice field are monitored and risk assessment indicators are calculated.

[0007] Step S3: Monitoring data collection and processing, specifically the drone monitors according to the planned route, obtains and stores rice field image data;

[0008] Step S4: Constructing a 3D model of the rice field. Specifically, the method involves finding 2D plane points in the rice field images captured by a multi-camera system, selecting seed points based on Harris response values, increasing the area of the seed points based on different feature information in the rice field images, extracting the initial plane, and reconstructing the rice field based on grammatical rule segmentation. A highly compatible mobile node is designed to adapt to the actual environment and integrated with the sensor network, providing a richer and more accurate data foundation for the 3D reconstruction of the rice field.

[0009] Step S5: Optimizing the rice field growth environment, specifically adjusting the rice planting layout according to the three-dimensional model of the rice field.

[0010] Furthermore, in step S1, the field survey includes the following steps:

[0011] Step S11: Determine the monitoring target and scope, clarify the size of the rice planting area to be monitored, conduct on-site field investigations to understand the topography and surrounding environment, and assess the safety and feasibility of UAV flights;

[0012] Step S12: Based on the survey results, the boundaries and approximate scope of the monitoring area are preliminarily planned on the map, and the finalized rice field monitoring area information is entered into the UAV monitoring system to prepare for subsequent flight monitoring.

[0013] Furthermore, in step S2, the autonomous planning of the monitoring route includes the following steps:

[0014] Step S21: Calculate the target value, which is used to select the action of the drone. The drone determines the next state and action based on the target value, and selects the action that is expected to bring the greatest return. This is the basis for forming the rice field monitoring route. The formula used is as follows:

[0015] ;

[0016] In the formula, y represents the target value, represents the reward obtained at time t+1, γ represents the discount factor, Indicates the state of the drone at the next moment, represents the parameters of the Q network at time t, represents the parameters of the target network, represents the corresponding drone action that maximizes the Q function;

[0017] Step S22: Setting a blocking mechanism. The blocking mechanism is used to mark the explored rice field monitoring area. When the drone selects an action according to the target value, the blocking mechanism will restrict the drone from entering the explored area, prompting the drone to explore the unexplored area, thereby forming a more efficient and better route.

[0018] The judgment function is used to determine whether the drone can navigate normally. If the drone is stuck in the current state, the adjacent blocked areas will be randomly unblocked to prevent the drone from being trapped during the monitoring process. The judgment function is set up with the following formula:

[0019] ;

[0020] Where, Represents the judgment function. When the judgment function value is 1, it means that the drone is in the normal navigation state; when the judgment function value is -1, it means that the drone is stuck in the current state. u represents the index of the rice field monitoring area, and N represents the set of blocked areas.

[0021] Step S23: Monitoring the blind spot at the edge of the paddy field. When the drone reaches the edge area and enters the visual blind spot, the edge blind spot risk assessment index is calculated. The formula used is as follows:

[0022] ;

[0023] In the formula, B represents the edge blind zone risk assessment index, δ represents the adjustment coefficient, represents the visual visibility probability in the edge area, D represents the distance from the UAV to the edge of the rice field, and M represents the density of obstacles in the blind spot;

[0024] Step S24: Monitoring route formation, based on the selection of actions and the risk assessment of blind spots at the edge of the rice field, continuous exploration and adjustment are carried out to form an autonomously planned monitoring route, while avoiding wandering in place, improving learning efficiency and the rationality of route planning.

[0025] Furthermore, in step S3, the monitoring data collection and processing includes the following steps:

[0026] Step S31: The UAV monitors the rice field according to the autonomously planned monitoring route, and uses a multi-camera system to photograph the rice field to obtain image data of the rice field;

[0027] Step S32: De-noising, contrast enhancement, and brightness adjustment are performed on the collected image data to improve image quality;

[0028] Step S33: extracting features from the rice field image, including the contour, color features, and texture features of the rice field;

[0029] Step S34: Target recognition: identifying the rice field area and rice plants based on the extracted features and marking them;

[0030] Step S35: storing the processed image data.

[0031] Furthermore, in step S4, constructing the three-dimensional model of the rice field includes the following steps:

[0032] Step S41: Find plane points based on the point cloud data. Using the rice field image captured by the multi-camera, the depth information of each pixel in the image is calculated through a stereo matching algorithm to generate the corresponding 3D point cloud data. The dimensional features are then classified and the corresponding 2D plane points are found.

[0033] The dimensional features represent the shape features and height features of different rice field monitoring areas;

[0034] Step S42: extracting the initial plane of the rice field, selecting seed points from the 2D plane points using Harris response values, and using information from the rice field image to determine dimensional features, point surface distance, and growth angle information to increase the area of the seed points. When the curvature of the seed point exceeds a preset threshold, the growth stops and the initial plane is extracted;

[0035] The dimensional features represent features related to the size, scale and proportion of the rice field, helping to determine the appropriate range and scale for extracting the initial plane, ensuring that the extracted plane matches the size of the actual rice field;

[0036] The point surface distance represents the distance information between the seed point and the surface of the surrounding object, which is used to understand the spatial relationship between the rice field and the surrounding environment;

[0037] The growth angle information determines the angle and method of seed point expansion, which directly affects the formation and range of the initial plane;

[0038] The seed point serves as a starting point to guide the direction and range of the entire monitoring expansion;

[0039] The area of the seed point is continuously expanded to determine the size of the plane region to be segmented, thereby shaping the scale of the initial plane;

[0040] The seed point curvature is used to measure the curvature direction of the seed point, and the monitoring area is controlled by comparing it with a preset threshold to ensure that the extracted plane meets the curvature requirement;

[0041] The preset threshold is a limit value used to determine when the curvature growth of the seed point stops;

[0042] Step S421: Selection of seed points. First, determine the adjacent nodes of each plane point, calculate the covariance matrix and Harris response value to judge the importance of the point, and use the Delaunay triangulation method and α-shape algorithm to determine whether the point is on the boundary. Finally, determine whether to select the point as a seed point and perform the first judgment step to calculate the Harris response value. The formula used is as follows:

[0043] ;

[0044] ;

[0045] Where Num represents the number of adjacent nodes, M represents the covariance matrix, and i represents the index of the adjacent node. Represents the coordinate value of the i-th adjacent node on the x-axis, Represents the coordinate value of the i-th adjacent node on the y-axis, Represents the coordinate value of the i-th adjacent node on the z-axis, R represents the Harris response value, det(M) represents the determinant of the covariance matrix, tr() represents the trace of the covariance matrix, when the Harris response value is greater than or equal to the preset response threshold, it is judged as a seed point for preliminary prediction; when the Harris response value is less than the preset response threshold, it is judged as a non-seed point;

[0046] The next judgment step is to use the Delaunay triangulation method and the α-shape algorithm to determine whether the initially predicted seed point is on the boundary. If the point is not on the boundary, the initially predicted seed point is classified as a seed point; if the point is on the boundary, the initially predicted seed point is classified as a non-seed point.

[0047] Step S422: Calculate the area growth process of the seed point. The formula used is as follows:

[0048] ;

[0049] ;

[0050] Where, Indicates the value after the area of the seed point increases at time t+1. represents the value of the seed point area at time t, k represents the growth coefficient, represents the rice field image information at time t, LGT represents the dimensional feature, which is used to determine the basic unit and proportion of the seed point area growth, k1 represents the weight of the dimensional feature in the rice field image information, d represents the point surface distance, which is used to control the range and boundary of the seed point area growth according to the distance from other points, k2 represents the weight of the point surface distance in the rice field image information, ξ represents the growth angle information, which is used to guide the extension angle of the seed point area growth, and k3 represents the weight of the growth angle information in the rice field image information;

[0051] Step S43: Reconstructing the paddy field based on grammatical rule segmentation. The paddy field area is segmented using actual point cloud data, and the initial paddy field plane is divided into different paddy field blocks. These paddy field blocks are then described and segmented by defining grammatical rules, thereby extracting and reconstructing each area of the paddy field, limiting the number of polygonal patches, and improving the reconstruction effect of sharp feature areas.

[0052] The grammatical rules represent a rule system for describing the segmentation and reconstruction of rice paddy areas;

[0053] The sharp feature areas refer to the boundaries of rice paddies, the junctions between ridges and rice paddies, the junctions between irrigation channels and rice paddies, and the edges of larger stones in rice paddies;

[0054] Step S44: Design highly compatible mobile nodes. Considering that the parameters in the actual environment change relatively stably, highly compatible mobile nodes are designed to adapt well to the actual rice field environment and dynamically integrate the sensor network well to solve the data loss problem caused by failed nodes, improve network reliability, and provide a richer and more accurate data basis for the three-dimensional reconstruction of rice fields.

[0055] Furthermore, in step S5, the optimization of the rice field growth environment is specifically to analyze the undulations and slopes in the terrain based on the three-dimensional reconstruction model of the rice field, improve irrigation and drainage conditions, adjust the rice planting layout, carry out scientific planting planning, and judge areas prone to breeding of pests and diseases based on the model, and take preventive measures in advance.

[0056] The rice planting environment monitoring system based on drone technology provided by the present invention includes a field survey module, an autonomous monitoring route planning module, a monitoring data collection and processing module, a rice field three-dimensional model construction module, and a rice field growth environment optimization module;

[0057] The field survey module specifically determines the rice field monitoring area and enters it into the drone monitoring system;

[0058] The autonomous monitoring route planning module specifically calculates target values to determine drone actions, forming the basis for rice field monitoring routes. It also sets a blocking mechanism to encourage drones to explore unexplored areas to form a more optimal route. By judging the drone's navigation status and randomly removing nearby blocking areas when trapped, it monitors blind spots at the edge of rice fields and calculates risk assessment indicators.

[0059] The monitoring data acquisition and processing module is specifically a drone that monitors according to the planned route, obtains rice field image data and stores it;

[0060] The rice field 3D model construction module specifically uses rice field images captured by a multi-camera to find 2D plane points, selects seed points based on Harris response values, increases the area of seed points and extracts initial planes based on different feature information in the rice field images, reconstructs the rice field based on grammatical rule segmentation, designs highly compatible mobile nodes to adapt to the actual environment and integrates sensor networks, providing a richer and more accurate data foundation for rice field 3D reconstruction;

[0061] The module for optimizing the rice field growth environment specifically adjusts the rice planting layout according to the three-dimensional model of the rice field.

[0062] The beneficial results achieved by the present invention using the above scheme are as follows:

[0063] (1) In order to solve the technical problems that UAVs are difficult to autonomously plan efficient, comprehensive and emergency-response flight routes in rice field monitoring, and are unable to break free from obstacles when trapped in special situations, entering visual blind spots and causing collision risks, resulting in unnecessary economic losses, we use the calculated target value to rationally plan actions to form a basic route, use the blocking mechanism to encourage UAVs to explore unexplored areas to optimize the route, and can randomly unblock the UAV when it is trapped to ensure normal flight. We also monitor and conduct risk assessment on the blind spots at the edge of rice fields.

[0064] (2) In order to solve the technical problems that the spatial structure and morphology of rice fields are unclear, it is difficult to collect comprehensive data and obtain accurate environmental conditions of rice fields, which makes it difficult to detect potential environmental risks and anomalies, and is not conducive to protecting the rice field ecosystem, the rice field images taken by multi-camera are used to find 2D plane points, and the seed points are selected by Harris response value. The area of seed points is increased and the initial plane is extracted according to the different feature information in the rice field images. The rice fields are reconstructed based on grammatical rule segmentation. Highly compatible mobile nodes are designed to adapt to the actual environment and integrated with sensor networks to provide a richer and more accurate data basis for the three-dimensional reconstruction of rice fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 A schematic diagram of the process of the rice planting environment monitoring method based on drone technology provided by the present invention;

[0066] Figure 2 A schematic diagram of a rice planting environment monitoring system based on drone technology provided by the present invention;

[0067] Figure 3 Schematic diagram of the process of step S2;

[0068] Figure 4 Schematic diagram of the process of step S4.

[0069] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0070] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0071] In the description of the present invention, it should be understood that terms such as "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they should not be understood as limiting the present invention.

[0072] Example 1, see Figure 1 The present invention provides a method for monitoring a rice planting environment based on drone technology, which comprises the following steps:

[0073] Step S1: Field survey, specifically determining the rice field monitoring area and entering it into the UAV monitoring system;

[0074] Step S2: Autonomously plan the monitoring route. Specifically, the target value is calculated to determine the drone's action, forming the basis of the rice field monitoring route. A blocking mechanism is set to encourage the drone to explore unexplored areas to form a better route. By judging the drone's navigation status and randomly removing nearby blocking areas when trapped, the blind spots at the edge of the rice field are monitored and risk assessment indicators are calculated.

[0075] Step S3: Monitoring data collection and processing, specifically the drone monitors according to the planned route, obtains and stores rice field image data;

[0076] Step S4: Constructing a 3D model of the rice field. Specifically, the method involves finding 2D plane points in the rice field images captured by a multi-camera system, selecting seed points based on Harris response values, increasing the area of the seed points based on different feature information in the rice field images, extracting the initial plane, and reconstructing the rice field based on grammatical rule segmentation. A highly compatible mobile node is designed to adapt to the actual environment and integrated with the sensor network, providing a richer and more accurate data foundation for the 3D reconstruction of the rice field.

[0077] Step S5: Optimizing the rice field growth environment, specifically adjusting the rice planting layout according to the three-dimensional model of the rice field.

[0078] Example 2, see Figure 1 This embodiment is based on the above embodiment. In step S1, the field survey includes the following steps:

[0079] Step S11: Determine the monitoring target and scope, clarify the size of the rice planting area to be monitored, conduct on-site field investigations to understand the topography and surrounding environment, and assess the safety and feasibility of UAV flights;

[0080] Step S12: Based on the survey results, the boundaries and approximate scope of the monitoring area are preliminarily planned on the map, and the finalized rice field monitoring area information is entered into the UAV monitoring system to prepare for subsequent flight monitoring.

[0081] Example 3, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. In step S2, the autonomous planning of the monitoring route includes the following steps:

[0082] Step S21: Calculate the target value, which is used to select the action of the drone. The drone determines the next state and action based on the target value, and selects the action that is expected to bring the greatest return. This is the basis for forming the rice field monitoring route. The formula used is as follows:

[0083] ;

[0084] In the formula, y represents the target value, represents the reward obtained at time t+1, γ represents the discount factor, Indicates the state of the drone at the next moment, represents the parameters of the Q network at time t, represents the parameters of the target network, represents the corresponding drone action that maximizes the Q function;

[0085] Step S22: Setting a blocking mechanism. The blocking mechanism is used to mark the explored rice field monitoring area. When the drone selects an action according to the target value, the blocking mechanism will restrict the drone from entering the explored area, prompting the drone to explore the unexplored area, thereby forming a more efficient and better route.

[0086] The judgment function is used to determine whether the drone can navigate normally. If the drone is stuck in the current state, the adjacent blocked areas will be randomly unblocked to prevent the drone from being trapped during the monitoring process. The judgment function is set up with the following formula:

[0087] ;

[0088] Where, Represents the judgment function. When the judgment function value is 1, it means that the drone is in the normal navigation state; when the judgment function value is -1, it means that the drone is stuck in the current state. u represents the index of the rice field monitoring area, and N represents the set of blocked areas.

[0089] Step S23: Monitoring the blind spot at the edge of the paddy field. When the drone reaches the edge area and enters the visual blind spot, the edge blind spot risk assessment index is calculated. The formula used is as follows:

[0090] ;

[0091] In the formula, B represents the edge blind zone risk assessment index, δ represents the adjustment coefficient, represents the visual visibility probability in the edge area, D represents the distance from the UAV to the edge of the rice field, and M represents the density of obstacles in the blind spot;

[0092] Step S24: Monitoring route formation, based on the selection of actions and the risk assessment of blind spots at the edge of the rice field, continuous exploration and adjustment are carried out to form an autonomously planned monitoring route, while avoiding wandering in place, improving learning efficiency and the rationality of route planning.

[0093] By performing the above operations, the calculated target value is used to rationally plan the action to form a basic route, the blocking mechanism is used to encourage the UAV to explore unexplored areas to optimize the route, and the blockage can be randomly unblocked when the UAV is trapped to ensure normal flight, as well as monitoring and risk assessment of the blind spots at the edge of the rice field. This solves the technical problems that UAVs are difficult to autonomously plan efficient, comprehensive and emergency-responsive flight routes in rice field monitoring, and are unable to break free from obstacles when trapped in special circumstances, entering visual blind spots and causing collision risks, resulting in unnecessary economic losses.

[0094] Example 4, see Figure 1 This embodiment is based on the above embodiment. In step S3, the monitoring data collection and processing includes the following steps:

[0095] Step S31: The UAV monitors the rice field according to the autonomously planned monitoring route, and uses a multi-camera system to photograph the rice field to obtain image data of the rice field;

[0096] Step S32: De-noising, contrast enhancement, and brightness adjustment are performed on the collected image data to improve image quality;

[0097] Step S33: extracting features from the rice field image, including the contour, color features, and texture features of the rice field;

[0098] Step S34: Target recognition: identifying the rice field area and rice plants based on the extracted features and marking them;

[0099] Step S35: storing the processed image data.

[0100] Example 5, see Figure 1 and Figure 4 This embodiment is based on the above embodiment. In step S4, the construction of the three-dimensional model of the rice field includes the following steps:

[0101] Step S41: Find plane points based on the point cloud data. Using the rice field image captured by the multi-camera, the depth information of each pixel in the image is calculated through a stereo matching algorithm to generate the corresponding 3D point cloud data. The dimensional features are then classified and the corresponding 2D plane points are found.

[0102] The dimensional features represent the shape features and height features of different rice field monitoring areas;

[0103] Step S42: extracting the initial plane of the rice field, selecting seed points from the 2D plane points using Harris response values, and using information from the rice field image to determine dimensional features, point surface distance, and growth angle information to increase the area of the seed points. When the curvature of the seed point exceeds a preset threshold, the growth stops and the initial plane is extracted;

[0104] The dimensional features represent features related to the size, scale and proportion of the rice field, helping to determine the appropriate range and scale for extracting the initial plane, ensuring that the extracted plane matches the size of the actual rice field;

[0105] The point surface distance represents the distance information between the seed point and the surface of the surrounding object, which is used to understand the spatial relationship between the rice field and the surrounding environment;

[0106] The growth angle information determines the angle and method of seed point expansion, which directly affects the formation and range of the initial plane;

[0107] The seed point serves as a starting point to guide the direction and range of the entire monitoring expansion;

[0108] The area of the seed point is continuously expanded to determine the size of the plane region to be segmented, thereby shaping the scale of the initial plane;

[0109] The seed point curvature is used to measure the curvature direction of the seed point, and the monitoring area is controlled by comparing it with a preset threshold to ensure that the extracted plane meets the curvature requirement;

[0110] The preset threshold is a limit value used to determine when the curvature growth of the seed point stops;

[0111] Step S421: Selection of seed points. First, determine the adjacent nodes of each plane point, calculate the covariance matrix and Harris response value to judge the importance of the point, and use the Delaunay triangulation method and α-shape algorithm to determine whether the point is on the boundary. Finally, determine whether to select the point as a seed point and perform the first judgment step to calculate the Harris response value. The formula used is as follows:

[0112] ;

[0113] ;

[0114] Where Num represents the number of adjacent nodes, M represents the covariance matrix, and i represents the index of the adjacent node. Represents the coordinate value of the i-th adjacent node on the x-axis, Represents the coordinate value of the i-th adjacent node on the y-axis, Represents the coordinate value of the i-th adjacent node on the z-axis, R represents the Harris response value, det(M) represents the determinant of the covariance matrix, tr() represents the trace of the covariance matrix, when the Harris response value is greater than or equal to the preset response threshold, it is judged as a seed point for preliminary prediction; when the Harris response value is less than the preset response threshold, it is judged as a non-seed point;

[0115] The next judgment step is to use the Delaunay triangulation method and the α-shape algorithm to determine whether the initially predicted seed point is on the boundary. If the point is not on the boundary, the initially predicted seed point is classified as a seed point; if the point is on the boundary, the initially predicted seed point is classified as a non-seed point.

[0116] Step S422: Calculate the area growth process of the seed point. The formula used is as follows:

[0117] ;

[0118] ;

[0119] Where, Indicates the value after the area of the seed point increases at time t+1. represents the value of the seed point area at time t, k represents the growth coefficient, represents the rice field image information at time t, LGT represents the dimensional feature, which is used to determine the basic unit and proportion of the seed point area growth, k1 represents the weight of the dimensional feature in the rice field image information, d represents the point surface distance, which is used to control the range and boundary of the seed point area growth according to the distance from other points, k2 represents the weight of the point surface distance in the rice field image information, ξ represents the growth angle information, which is used to guide the extension angle of the seed point area growth, and k3 represents the weight of the growth angle information in the rice field image information;

[0120] Step S43: Reconstructing the paddy field based on grammatical rule segmentation. The paddy field area is segmented using actual point cloud data, and the initial paddy field plane is divided into different paddy field blocks. These paddy field blocks are then described and segmented by defining grammatical rules, thereby extracting and reconstructing each area of the paddy field, limiting the number of polygonal patches, and improving the reconstruction effect of sharp feature areas.

[0121] The grammatical rules represent a rule system for describing the segmentation and reconstruction of rice paddy areas;

[0122] The sharp feature areas refer to the boundaries of rice paddies, the junctions between ridges and rice paddies, the junctions between irrigation channels and rice paddies, and the edges of larger stones in rice paddies;

[0123] Step S44: Design highly compatible mobile nodes. Considering that the parameters in the actual environment change relatively stably, highly compatible mobile nodes are designed to adapt well to the actual rice field environment and dynamically integrate the sensor network well to solve the data loss problem caused by failed nodes, improve network reliability, and provide a richer and more accurate data basis for the three-dimensional reconstruction of rice fields.

[0124] By performing the above operations, 2D plane points are found in rice field images taken by a multi-camera, seed points are selected through Harris response values, the area of seed points is increased and the initial plane is extracted according to different feature information in the rice field image, and rice fields are reconstructed based on grammatical rule segmentation. Highly compatible mobile nodes are designed to adapt to the actual environment and integrate sensor networks, providing a richer and more accurate data foundation for 3D reconstruction of rice fields. This solves the technical problems of unclear spatial structure and morphology of rice fields, difficulty in comprehensive data collection, inability to obtain accurate rice field environmental conditions, difficulty in detecting potential environmental risks and anomalies, and disadvantageous for protecting rice field ecosystems.

[0125] Example 6, see Figure 1This embodiment is based on the above embodiment. In step S5, the optimization of the rice field growth environment is specifically to analyze the undulations and slopes in the terrain according to the three-dimensional reconstruction model of the rice field, improve the irrigation and drainage conditions, adjust the rice planting layout, carry out scientific planting planning, and judge the areas prone to breeding of diseases and pests according to the model, and take preventive measures in advance.

[0126] Example 7, see Figure 2 This embodiment is based on the above embodiment. The rice planting environment monitoring system based on drone technology provided by the present invention includes a field survey module, an autonomous monitoring route planning module, a monitoring data acquisition and processing module, a rice field three-dimensional model construction module, and a rice field growth environment optimization module;

[0127] The field survey module specifically determines the rice field monitoring area and enters it into the drone monitoring system;

[0128] The autonomous monitoring route planning module specifically calculates target values to determine drone actions, forming the basis for rice field monitoring routes. It also sets a blocking mechanism to encourage drones to explore unexplored areas to form a more optimal route. By judging the drone's navigation status and randomly removing nearby blocking areas when trapped, it monitors blind spots at the edge of rice fields and calculates risk assessment indicators.

[0129] The monitoring data acquisition and processing module is specifically a drone that monitors according to the planned route, obtains rice field image data and stores it;

[0130] The rice field 3D model construction module specifically uses rice field images captured by a multi-camera to find 2D plane points, selects seed points based on Harris response values, increases the area of seed points and extracts initial planes based on different feature information in the rice field images, reconstructs the rice field based on grammatical rule segmentation, designs highly compatible mobile nodes to adapt to the actual environment and integrates sensor networks, providing a richer and more accurate data foundation for rice field 3D reconstruction;

[0131] The module for optimizing the rice field growth environment specifically adjusts the rice planting layout according to the three-dimensional model of the rice field.

[0132] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0133] While the embodiments of the present invention have been shown and described, it will be apparent to those skilled in the art that various changes, modifications, substitutions, and alterations can be made to the embodiments without departing from the principles and spirit of the invention.

[0134] The present invention and its embodiments are described above. This description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs structures and embodiments similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. A rice planting environment monitoring method based on drone technology is characterized by: The method comprises the following steps: Step S1: Field survey, specifically determining the rice field monitoring area and entering it into the UAV monitoring system; Step S2: Autonomously plan the monitoring route. Specifically, the target value is calculated to determine the drone's action, forming the basis of the rice field monitoring route. A blocking mechanism is set to encourage the drone to explore unexplored areas to form a better route. By judging the drone's navigation status and randomly removing nearby blocking areas when trapped, the blind spots at the edge of the rice field are monitored and risk assessment indicators are calculated. Step S3: Monitoring data collection and processing, specifically the drone monitors according to the planned route, obtains and stores rice field image data; Step S4: Constructing a 3D model of the rice field. Specifically, the method involves finding 2D plane points in the rice field images captured by a multi-camera system, selecting seed points based on Harris response values, increasing the area of the seed points based on different feature information in the rice field images, extracting the initial plane, and reconstructing the rice field based on grammatical rule segmentation. A highly compatible mobile node is designed to adapt to the actual environment and integrated with the sensor network to provide a data foundation for the 3D reconstruction of the rice field. Step S5: Optimizing the rice field growth environment, specifically adjusting the rice planting layout based on the three-dimensional model of the rice field; In step S2, the autonomous planning of the monitoring route includes the following steps: Step S21: Calculate the target value for selecting the action of the drone. The formula used is as follows: ； In the formula, y represents the target value, represents the reward obtained at time t+1, γ represents the discount factor, Indicates the state of the drone at the next moment, represents the parameters of the Q network at time t, represents the parameters of the target network, represents the corresponding drone action that maximizes the Q function; Step S22: Setting a blocking mechanism. The blocking mechanism is used to mark the explored rice field monitoring area. When the drone selects an action according to the target value, the blocking mechanism will restrict the drone from entering the explored area, prompting the drone to explore the unexplored area and form a more optimal route. The judgment function is used to determine whether the drone can navigate normally. The judgment function is set up with the following formula: ； Where, Represents the judgment function. When the judgment function value is 1, it means that the drone is in the normal navigation state; when the judgment function value is -1, it means that the drone is stuck in the current state. u represents the index of the rice field monitoring area, and N represents the set of blocked areas. Step S23: Monitoring the blind spot at the edge of the paddy field. When the drone reaches the edge area and enters the visual blind spot, the edge blind spot risk assessment index is calculated. The formula used is as follows: ； In the formula, B represents the edge blind zone risk assessment index, δ represents the adjustment coefficient, represents the visual visibility probability in the edge area, D represents the distance from the UAV to the edge of the rice field, and M represents the density of obstacles in the blind spot; Step S24: The monitoring route is formed. Based on the action selection and the risk assessment of the blind spot at the edge of the rice field, continuous exploration and adjustment are carried out to form an autonomously planned monitoring route to avoid wandering in place.

2. The rice planting environment monitoring method based on drone technology according to claim 1 is characterized in that: In step S4, the construction of the three-dimensional model of the paddy field includes the following steps: Step S41: Find plane points based on the point cloud data. Using the rice field image captured by the multi-camera, the depth information of each pixel in the image is calculated through a stereo matching algorithm to generate the corresponding 3D point cloud data. The dimensional features are then classified and the corresponding 2D plane points are found. Step S42: extracting the initial plane of the rice field, selecting seed points from the 2D plane points using Harris response values, and increasing the area of the seed points using information in the rice field image to determine dimensional features, point surface distance, and growth angle information. When the curvature of the seed point exceeds a preset threshold, the growth stops, and the initial plane of the rice field is extracted. The process includes the following steps: Step S421: Select the seed point, perform the first judgment step, and calculate the Harris response value. The formula used is as follows: ； ； Where Num represents the number of adjacent nodes, M represents the covariance matrix, and i represents the index of the adjacent node. Represents the coordinate value of the i-th adjacent node on the x-axis, Represents the coordinate value of the i-th adjacent node on the y-axis, Represents the coordinate value of the i-th adjacent node on the z-axis, R represents the Harris response value, det(M) represents the determinant of the covariance matrix, tr() represents the trace of the covariance matrix, when the Harris response value is greater than or equal to the preset response threshold, it is judged as a seed point for preliminary prediction; when the Harris response value is less than the preset response threshold, it is judged as a non-seed point; The next judgment step is to use the Delaunay triangulation method and the α-shape algorithm to determine whether the initially predicted seed point is on the boundary. If the point is not on the boundary, the initially predicted seed point is classified as a seed point; if the point is on the boundary, the initially predicted seed point is classified as a non-seed point. Step S422: Calculate the area growth process of the seed point. The formula used is as follows: ； ； Where, Indicates the value after the area of the seed point increases at time t+1. represents the value of the seed point area at time t, k represents the growth coefficient, represents the rice field image information at time t, LGT represents the dimensional feature, which is used to determine the basic unit and proportion of the seed point area growth, k1 represents the weight of the dimensional feature in the rice field image information, d represents the point surface distance, which is used to control the range and boundary of the seed point area growth according to the distance from other points, k2 represents the weight of the point surface distance in the rice field image information, ξ represents the growth angle information, which is used to guide the extension angle of the seed point area growth, and k3 represents the weight of the growth angle information in the rice field image information; Step S43: Reconstructing the paddy field based on grammatical rule segmentation, using actual point cloud data to segment the paddy field area, and dividing the initial paddy field plane into different paddy field blocks. Then, these paddy field blocks are described and segmented by defining grammatical rules, thereby realizing extraction and reconstruction of each area of the paddy field; Step S44: Design highly compatible mobile nodes. Considering that parameters in the actual environment change relatively stably, design highly compatible mobile nodes to dynamically integrate the sensor network in the actual rice field environment to solve the data loss problem caused by failed nodes and provide a richer and more accurate data basis for the three-dimensional reconstruction of rice fields.

3. The rice planting environment monitoring method based on drone technology according to claim 1 is characterized in that: In step S1, the field survey includes the following steps: Step S11: Determine the monitoring target and scope, clarify the size of the rice planting area to be monitored, conduct on-site field investigations to understand the topography and surrounding environment, and assess the safety and feasibility of UAV flights; Step S12: Based on the survey results, the boundaries and scope of the monitoring area are preliminarily planned on the map, and the finalized rice field monitoring area information is entered into the UAV monitoring system to prepare for subsequent flight monitoring.

4. The rice planting environment monitoring method based on drone technology according to claim 1 is characterized in that: In step S3, the monitoring data collection and processing includes the following steps: Step S31: The UAV monitors the rice field according to the autonomously planned monitoring route, and uses a multi-camera system to photograph the rice field to obtain image data of the rice field; Step S32: De-noising, contrast enhancement, and brightness adjustment are performed on the collected image data to improve image quality; Step S33: extracting features from the rice field image; Step S34: Target recognition: identifying the rice field area and rice plants based on the extracted features and marking them; Step S35: storing the processed image data.

5. The rice planting environment monitoring method based on drone technology according to claim 1 is characterized in that: In step S5, the optimization of the rice field growth environment is specifically to analyze the undulations and slopes in the terrain based on the three-dimensional reconstruction model of the rice field, improve irrigation and drainage conditions, adjust the rice planting layout, carry out scientific planting planning, and determine areas prone to breeding of pests and diseases based on the model, and take preventive measures in advance.

6. A rice planting environment monitoring system based on drone technology, for implementing the rice planting environment monitoring method based on drone technology as described in any one of claims 1 to 5, characterized in that: It includes a field survey module, an autonomous monitoring route planning module, a monitoring data collection and processing module, a rice field three-dimensional model construction module, and a rice field growth environment optimization module.

7. The rice planting environment monitoring system based on drone technology according to claim 6 is characterized by: The field survey module specifically determines the rice field monitoring area and enters it into the drone monitoring system; The autonomous monitoring route planning module specifically calculates target values to determine drone actions, forming the basis for rice field monitoring routes. It also sets a blocking mechanism to encourage drones to explore unexplored areas to form a more optimal route. By judging the drone's navigation status and randomly removing nearby blocking areas when trapped, it monitors blind spots at the edge of rice fields and calculates risk assessment indicators. The monitoring data acquisition and processing module is specifically a drone that monitors according to the planned route, obtains rice field image data and stores it; The rice field 3D model construction module specifically uses rice field images captured by a multi-camera to find 2D plane points, selects seed points based on Harris response values, increases the area of seed points and extracts initial planes based on different feature information in the rice field images, reconstructs the rice field based on grammatical rule segmentation, designs highly compatible mobile nodes to adapt to the actual environment and integrates sensor networks to provide a data foundation for rice field 3D reconstruction; The module for optimizing the rice field growth environment specifically adjusts the rice planting layout according to the three-dimensional model of the rice field.

Images