A method for precise topdressing of rice using agricultural drones

Abstract

This invention discloses a method for precise topdressing of rice using agricultural drones, comprising: monitoring seedling growth information during the rice growth process and constructing a spatial distribution map of seedling condition index; generating a topdressing application rate prescription map based on the spatial distribution map of seedling condition index and a preset dynamic curve of suitable rice growth indicators; determining the navigation path map of the agricultural drone; and fusing the topdressing application rate prescription map with the navigation path map, and performing topdressing operations using the agricultural drone based on the fusion result. This method eliminates the need for cumbersome steps such as hyperspectral unmixing and dimensionality reduction modeling, improving decision-making efficiency. It accurately matches the required fertilizer amount, avoiding over-fertilization; the fusion of the navigation path and prescription map enables finer-grained variable spraying, eliminating the need to rely on gridded constraints on the smallest operational unit, thus improving spray uniformity.

Description

Technical Field

[0001] This invention relates to the field of intelligent agricultural power machinery technology, and in particular to a method for precise topdressing of rice using unmanned aerial vehicles (UAVs). Background Technology

[0002] In recent years, the use of agricultural drones for topdressing in rice production has developed rapidly. However, existing topdressing operations using agricultural drones largely rely on the experience of field managers, lacking effective decision-making basis.

[0003] Application publication number CN111670668A discloses a method for precise topdressing of rice using agricultural drones based on hyperspectral remote sensing prescription maps. This method combines remote sensing diagnosis of topdressing during the rice tillering stage with precise operation of agricultural drones. It uses drone hyperspectral technology to establish a prescription map for topdressing during the rice tillering stage. Based on this, and combined with the operation parameters of agricultural drones, the plots to be topdressed are divided into grids to form a spraying amount suitable for precise topdressing operations within the plots. Finally, precise topdressing is carried out using agricultural drones. The following technical problems exist: Hyperspectral data processing is complex and costly, requiring cumbersome steps such as hyperspectral unmixing and dimensionality reduction modeling, involving specialized equipment (such as the GaiaSky-mini imager) and algorithms (MNF, PPI, PSO-ELM); it cannot respond to the different fertilizer requirements during key growth stages such as tillering and jointing; it is difficult to adapt to different soil / climate conditions based on fixed parameters, resulting in poor adaptability of static fertilization models; drone operations are disconnected from agronomic needs, and navigation paths are separated from fertilization decisions, leading to uneven spraying, and reliance on gridded constraints for the smallest operational unit. Summary of the Invention

[0004] This invention aims to at least partially solve one of the technical problems in the aforementioned technologies. Therefore, the purpose of this invention is to propose a method for precise topdressing of rice using unmanned aerial vehicles (UAVs), which eliminates the need for cumbersome steps such as hyperspectral unmixing and dimensionality reduction modeling, thus improving decision-making efficiency. It accurately matches the required fertilizer amount, avoiding over-fertilization; the navigation path and prescription map are integrated to achieve finer-grained variable spraying, without relying on gridded constraints on the smallest working unit, improving spray uniformity.

[0005] To achieve the above objectives, this invention proposes a method for precise topdressing of rice using unmanned aerial vehicles (UAVs), comprising:

[0006] Monitor rice seedling growth information during the rice growth process and construct a spatial distribution map of seedling condition index;

[0007] Based on the spatial distribution map of seedling condition index and the preset dynamic curve of suitable rice growth index, a topdressing application prescription map is generated.

[0008] Determine the navigation path map for rice farming drones;

[0009] The topdressing application rate prescription map is integrated with the navigation path map, and the topdressing operation is carried out by rice agricultural drones based on the integration results.

[0010] According to some embodiments of the present invention, monitoring rice seedling growth information during the rice growth process and constructing a spatial distribution map of seedling condition index includes:

[0011] Based on the drone's flight over the rice growing area, the first monitoring image is acquired using the onboard RGB camera, and the second monitoring image is acquired using the onboard multispectral camera;

[0012] The rice cover information was determined based on the first monitoring image;

[0013] The leaf area index information of rice was determined based on the second monitoring image;

[0014] Based on the coverage information and leaf area index information, seedling growth information is determined, and a spatial distribution map of seedling index is constructed.

[0015] According to some embodiments of the present invention, determining rice coverage information based on a first monitoring image includes:

[0016] The first monitoring image is stitched together to obtain a panoramic image;

[0017] Histogram equalization and contrast stretching are performed on the panoramic image to obtain an enhanced image;

[0018] The enhanced image is then denoised to obtain a denoised image.

[0019] The denoised image is segmented based on color features to obtain a first segmentation result; the denoised image is segmented based on texture features to obtain a second segmentation result; the crop region is determined by mutual correction based on the first and second segmentation results.

[0020] The ratio of the number of pixels in the crop area to the number of pixels in the monitoring area corresponding to the denoised image is calculated to determine the rice coverage information.

[0021] According to some embodiments of the present invention, determining the leaf area index information of rice based on a second monitoring image includes:

[0022] Generate orthophoto maps of reflectance for each band based on the second monitoring image;

[0023] Based on grayboard data, the DN values ​​of the reflectance orthophoto map of each band are converted into surface reflectance values. Band registration is performed on the reflectance orthophoto maps of all bands to determine the multispectral band reflectance data.

[0024] The normalized differential vegetation index was calculated based on multispectral reflectance data. The leaf area index of rice was determined by fitting the normalized differential vegetation index to a regression analysis model.

[0025] According to some embodiments of the present invention, seedling growth information is determined based on coverage information and leaf area index information, and a spatial distribution map of seedling index is constructed, including:

[0026] Based on the coverage information, growth stage assessment and growth uniformity assessment are performed to obtain the first assessment result;

[0027] Based on leaf area index information, photosynthetic capacity and growth potential were assessed to obtain the second assessment result;

[0028] Based on the results of the first and second assessments and a pre-set data table, the seedling growth level is determined as seedling growth information; a spatial distribution map of the seedling index is constructed based on the seedling growth information.

[0029] According to some embodiments of the present invention, a topdressing application rate prescription map is generated based on the spatial distribution map of seedling condition index and the preset dynamic curve of suitable rice growth index, including:

[0030] The dynamic curve of the preset suitable growth index for rice is determined based on historical experimental data and agricultural models; the historical experimental data are the dynamic changes of the coverage information and leaf area index information of the same crop at different yield levels; the agricultural model is the growth curve of rice growth model simulating ideal conditions.

[0031] Determine the local coverage information and leaf area index information corresponding to the spatial distribution map of seedling condition index;

[0032] The deviation information between the coverage information and leaf area index information and the corresponding data points of the dynamic curve of suitable growth indicators for rice is calculated. Based on the deviation information, a preset deviation-topdressing amount data table is queried. Based on the query results and the spatial distribution map of seedling condition index, a topdressing application amount prescription map is generated.

[0033] According to some embodiments of the present invention, a topdressing application rate prescription map is generated based on the query results and the spatial distribution map of the seedling condition index, including:

[0034] Based on the query results, determine the amount of topdressing to be applied in the corresponding local area of ​​the spatial distribution map of the seedling condition index.

[0035] At the boundaries of local areas with different topdressing application rates, the topdressing application rate corresponding to the boundary of the spatial distribution map of seedling condition index is determined based on the S-curve interpolation method.

[0036] Based on the topdressing application amount corresponding to the local area and the topdressing application amount at the corresponding boundary of the seedling condition index spatial distribution map, a topdressing application amount prescription map is generated.

[0037] According to some embodiments of the present invention, determining the topdressing application amount corresponding to the boundary of the spatial distribution map of seedling condition index based on the S-curve interpolation method includes:

[0038] N(d)=N1+(N2-N1)×[1 / (1+e -k×(d-d0) )]

[0039] Where N(d) is the amount of topdressing applied at a distance d from the boundary; N1 and N2 are the local topdressing applications on both sides; e is the natural constant; k is the smoothing coefficient; d is the distance from the boundary; and d0 is the boundary position.

[0040] According to some embodiments of the present invention, determining the navigation path map for a rice farming drone includes:

[0041] The non-topdressing and topdressing areas of the rice growing region are determined according to the topdressing application rate prescription map;

[0042] Obtain 3D map information and obstacle information of the topdressing area, and set the starting point and key points for navigation path planning;

[0043] Set direct flight path and obstacle avoidance path based on 3D map information and obstacle information;

[0044] The navigation path map for rice farming drones is determined based on the starting point and key points, direct flight path and obstacle avoidance path of the navigation path planning.

[0045] According to some embodiments of the present invention, a topdressing application rate prescription map is fused with a navigation path map, and topdressing operations are performed using a rice agricultural drone based on the fusion result, including:

[0046] The fertilizer application rate prescription map and navigation path map are converted to the same coordinate system; the same coordinate system is the UTM projection coordinate system.

[0047] The topdressing application rate prescription map is rasterized in the same coordinate system, and a point vector layer is created. Each point corresponds to a grid center, and a field is added to the grid center to mark the topdressing application rate.

[0048] In the same coordinate system, perform path statistics on the navigation path map, determine the path length, and perform equal interval interpolation. For each interpolated path point, search for the nearest neighbor point in the point vector layer. Based on the search results, determine several fusion points between the topdressing application rate prescription map and the navigation path map.

[0049] Preliminary fusion is performed based on several fusion points to obtain preliminary fusion results;

[0050] Based on the preliminary fusion results and the UAV's flight parameters, the final fusion result is obtained; the flight parameters include flight speed, flight altitude, and flight direction.

[0051] Topdressing operations were carried out using agricultural drones for rice cultivation based on the fusion results.

[0052] This invention proposes a method for precise topdressing of rice using agricultural drones. Based on a spatial distribution map of seedling condition index, it facilitates rapid quantitative diagnosis of indicators such as nitrogen nutrition in rice. A topdressing application prescription map is generated based on the seedling condition index spatial distribution map and a preset dynamic curve of suitable rice growth indicators. This prescription map is then fused with a navigation path map, and topdressing is performed using an agricultural drone based on the fusion result. This method eliminates the need for cumbersome steps such as hyperspectral unmixing and dimensionality reduction modeling, improving decision-making efficiency. It accurately matches the required fertilizer amount, avoiding over-fertilization. The fusion of the navigation path and prescription map enables finer-grained variable spraying, eliminating the need for gridded constraints on the smallest operational unit and improving spray uniformity.

[0053] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0054] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0055] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0056] Figure 1 This is a flowchart of a method for precise topdressing of rice using an unmanned aerial vehicle according to an embodiment of the present invention;

[0057] Figure 2 This is a schematic diagram of a rice agricultural drone monitoring rice according to an embodiment of the present invention;

[0058] Figure 3 This is a schematic diagram of leaf area index information according to an embodiment of the present invention;

[0059] Figure 4 This is a schematic diagram of a topdressing application rate prescription according to an embodiment of the present invention;

[0060] Figure 5 This is a schematic diagram of a navigation path map according to an embodiment of the present invention. Detailed Implementation

[0061] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0062] like Figures 1-5 As shown in the figure, this invention proposes a method for precise topdressing of rice using a drone, including steps S1-S4:

[0063] S1. Monitor the seedling growth information during the rice growth process and construct a spatial distribution map of the seedling index;

[0064] S2. Generate a topdressing application rate prescription map based on the spatial distribution map of seedling condition index and the preset dynamic curve of suitable rice growth index.

[0065] S3. Determine the navigation path map for rice farming drones;

[0066] S4. Integrate the topdressing application rate prescription map with the navigation path map, and carry out topdressing operations using rice agricultural drones based on the integration results.

[0067] The working principle of the above technical solution is as follows: Seedling growth information represents the growth status of rice, including coverage and leaf area index. The spatial distribution map of the seedling index is a data map obtained by organizing the growth information and location distribution of rice. A preset dynamic curve of suitable rice growth indicators is compared with the spatial distribution map of the seedling index. Based on the matching results, it is easy to determine and accurately recommend the appropriate amount of topdressing, and provide technical guidance in the form of a spatial application prescription map, generating a topdressing application prescription map. The navigation path map is determined based on the fertilization area, starting point, and key points (boundary points of the area) of the topdressing application prescription map. The topdressing application prescription map and the navigation path map are integrated, and based on the fusion result, precise variable-rate topdressing operations for rice are carried out using rice agricultural drones based on real-time seedling conditions.

[0068] The beneficial effects of the above technical solution are as follows: Based on the spatial distribution map of the seedling condition index, it facilitates rapid quantitative diagnosis of indicators such as nitrogen nutrition in rice; based on the spatial distribution map of the seedling condition index and the preset dynamic curve of suitable rice growth indicators, a topdressing application prescription map is generated; the topdressing application prescription map is integrated with the navigation path map, and topdressing operations are carried out using rice agricultural drones based on the fusion result. This eliminates the need for cumbersome steps such as hyperspectral unmixing and dimensionality reduction modeling, improving decision-making efficiency. It accurately matches the required fertilizer amount, avoiding over-fertilization; the integration of the navigation path and prescription map enables finer-grained variable spraying, eliminating the need to rely on gridded constraints on the smallest operational unit, thus improving spray uniformity.

[0069] According to some embodiments of the present invention, monitoring rice seedling growth information during the rice growth process and constructing a spatial distribution map of seedling condition index includes:

[0070] Based on the drone's flight over the rice growing area, the first monitoring image is acquired using the onboard RGB camera, and the second monitoring image is acquired using the onboard multispectral camera;

[0071] The rice cover information was determined based on the first monitoring image;

[0072] The leaf area index information of rice was determined based on the second monitoring image;

[0073] Based on the coverage information and leaf area index information, seedling growth information is determined, and a spatial distribution map of seedling index is constructed.

[0074] The working principle of the above technical solution is as follows: Visible light images are acquired using an RGB camera, i.e., the first monitoring image. Multispectral images are acquired using a multispectral camera, i.e., the second monitoring image. Rice canopy coverage is extracted from the first monitoring image. Five-band images (blue, green, red, red-edge, and near-infrared) are acquired from the second monitoring image, and NDVI (Normalized Difference Vegetation Index) is calculated to retrieve the leaf area index (LAI). Based on different growth stages of rice, corresponding weighting coefficients are preset for coverage and LAI information. Weighted calculations are performed based on these weighting coefficients to determine seedling growth information. For example, during the tillering stage, the preset weighting coefficients for coverage and LAI information are 0.6 and 0.4, respectively; during the jointing stage, the preset weighting coefficients are 0.4 and 0.6, respectively. The seedling growth information is mapped to geographic coordinates to form a meter-resolution grid map, determining the spatial distribution map of the seedling index.

[0075] The beneficial effects of the above technical solution are: by using different monitoring methods, coverage information and leaf area index information can be determined, and based on these two pieces of information, it is convenient to comprehensively assess seedling growth information and improve the accuracy of constructing a spatial distribution map of seedling index.

[0076] According to some embodiments of the present invention, determining rice coverage information based on a first monitoring image includes:

[0077] The first monitoring image is stitched together to obtain a panoramic image;

[0078] Histogram equalization and contrast stretching are performed on the panoramic image to obtain an enhanced image;

[0079] The enhanced image is then denoised to obtain a denoised image.

[0080] The denoised image is segmented based on color features to obtain a first segmentation result; the denoised image is segmented based on texture features to obtain a second segmentation result; the crop region is determined by mutual correction based on the first and second segmentation results.

[0081] The ratio of the number of pixels in the crop area to the number of pixels in the monitoring area corresponding to the denoised image is calculated to determine the rice coverage information.

[0082] The working principle of the above technical solution is as follows: The first monitoring image is composed of multiple small-area images. Based on image stitching processing, a panoramic image is easily obtained, which facilitates expanding the monitoring range and obtaining more comprehensive information on the rice growth area. Histogram equalization and contrast stretching processing are performed on the panoramic image to obtain an enhanced image, which improves the visual effect of the image and makes the grayscale difference between the rice area and the background area more obvious. Noise reduction processing is performed on the enhanced image to obtain a denoised image, eliminating the influence of image noise and improving image quality. The denoised image is segmented based on color features to obtain the first segmentation result. The values ​​of each channel in the denoised image are determined in the HSV color space, i.e., the denoised image is converted from the RGB color space to the HSV color space, and the values ​​of the H, S, and V channels are determined. Rice characteristic color thresholds are set for the H, S, and V channels respectively. The H channel facilitates the exclusion of non-green areas, the S channel facilitates the exclusion of low-saturation soil and highly illuminated water surfaces, and the V channel facilitates the exclusion of excessively dark / bright areas. For example: the threshold for the H channel (hue) is 80°-160°; the threshold for the S channel (saturation) is 20%-70%; and the threshold for the V channel (brightness) is 30%-60%. The pixels in the image are divided into crop regions and background regions. The denoised image is segmented based on texture features to obtain a second segmentation result. The texture features of the pixels in the image are analyzed, including contrast, energy, and entropy. Thresholds for contrast, energy, and entropy corresponding to rice are set (the thresholds are determined based on the crop's growth stage and corresponding historical data). Pixels whose texture features meet the corresponding thresholds are considered crop regions, thus distinguishing between crop regions and background regions. The segmentation results based on color and texture features are fused to remove incorrectly segmented regions and correct inaccurate boundaries, resulting in more accurate crop regions. The ratio of the number of pixels in the crop region to the number of pixels in the corresponding monitoring region of the denoised image is calculated to determine the rice coverage information.

[0083] The beneficial effects of the above technical solution are as follows: Image stitching, histogram equalization, contrast stretching, and noise reduction are performed on the first monitoring image to obtain an accurate denoised image. The denoised image is then segmented using color features and texture features, and the two segmentation results are mutually corrected to obtain an accurate crop region. The ratio of the number of pixels in the crop region to the number of pixels in the corresponding monitoring region of the denoised image is calculated, facilitating the accurate determination of rice coverage information.

[0084] According to some embodiments of the present invention, determining the leaf area index information of rice based on a second monitoring image includes:

[0085] Generate orthophoto maps of reflectance for each band based on the second monitoring image;

[0086] Based on grayboard data, the DN values ​​of the reflectance orthophoto map of each band are converted into surface reflectance values. Band registration is performed on the reflectance orthophoto maps of all bands to determine the multispectral band reflectance data.

[0087] The normalized differential vegetation index was calculated based on multispectral reflectance data. The leaf area index of rice was determined by fitting the normalized differential vegetation index to a regression analysis model.

[0088] The working principle of the above technical solution is as follows: Reflectance orthophoto maps for each band are generated based on the second monitoring image. The grayboard data is determined based on grayboards placed on the ground, providing a reference target with known and stable reflectance. Based on the grayboard data, the DN values ​​of the reflectance orthophoto maps for each band are converted into surface reflectance values, facilitating the elimination of the influence of sensor and lighting conditions on the image data and accurately reflecting the true reflectance characteristics of ground features. Band registration is performed on all bands of reflectance orthophoto maps to determine multispectral reflectance data; images of different bands are aligned to the same spatial coordinate system to ensure that corresponding pixels in different band images represent the same ground feature location, facilitating subsequent calculation of leaf area index (LAI) information using multispectral reflectance data. The leaf area index (LAI) is the ratio of the total vertical projection area of ​​plant leaves per unit land area to the land area. Based on the normalized difference vegetation index of the sample area, a regression analysis model of leaf area index and normalized difference vegetation index was established and fitted to obtain an accurate model. Based on the model, the leaf area index of rice was determined by the obtained normalized difference vegetation index.

[0089] The beneficial effects of the above technical solution are as follows: Based on grayboard data, the DN value of the reflectance orthophoto map of each band is converted into the surface reflectance value, the reflectance orthophoto maps of all bands are registered, and the multispectral reflectance data is determined; the normalized differential vegetation index is calculated based on the multispectral reflectance data, and the accurate leaf area index information is obtained by processing it through a regression analysis model.

[0090] According to some embodiments of the present invention, seedling growth information is determined based on coverage information and leaf area index information, and a spatial distribution map of seedling index is constructed, including:

[0091] Based on the coverage information, growth stage assessment and growth uniformity assessment are performed to obtain the first assessment result;

[0092] Based on leaf area index information, photosynthetic capacity and growth potential were assessed to obtain the second assessment result;

[0093] Based on the results of the first and second assessments and a pre-set data table, the seedling growth level is determined as seedling growth information; a spatial distribution map of the seedling index is constructed based on the seedling growth information.

[0094] The working principle of the above technical solution is as follows: Growth stage and growth uniformity are assessed based on coverage information to obtain a first assessment result; photosynthetic capacity and growth potential are assessed based on leaf area index information to obtain a second assessment result; based on the first and second assessment results and a preset data table (which serves as a comparison table of the first assessment result, second assessment result, and seedling growth level), the seedling growth level is determined as seedling growth information; a spatial distribution map of the seedling index is constructed based on the seedling growth information.

[0095] The beneficial effects of the above technical solution are: growth stage assessment and growth uniformity assessment are performed through coverage information, and photosynthetic capacity assessment and growth potential prediction assessment are performed through leaf area index information, which facilitates the comprehensive assessment results of seedling condition and thus facilitates the accurate construction of a spatial distribution map of seedling condition index.

[0096] According to some embodiments of the present invention, a topdressing application rate prescription map is generated based on the spatial distribution map of seedling condition index and the preset dynamic curve of suitable rice growth index, including:

[0097] The dynamic curve of the preset suitable growth index for rice is determined based on historical experimental data and agricultural models; the historical experimental data are the dynamic changes of the coverage information and leaf area index information of the same crop at different yield levels; the agricultural model is the growth curve of rice growth model simulating ideal conditions.

[0098] Determine the local coverage information and leaf area index information corresponding to the spatial distribution map of seedling condition index;

[0099] The deviation information between the coverage information and leaf area index information and the corresponding data points of the dynamic curve of suitable growth indicators for rice is calculated. Based on the deviation information, a preset deviation-topdressing amount data table is queried. Based on the query results and the spatial distribution map of seedling condition index, a topdressing application amount prescription map is generated.

[0100] The working principle of the above technical solution is as follows: Based on historical experimental data and agricultural models, a preset dynamic curve of suitable growth indicators for rice is determined; the historical experimental data refers to the dynamic changes in the coverage information and leaf area index information of the same crop at different yield levels; the agricultural model is a rice growth model simulating the growth curve under ideal conditions; the coverage information and leaf area index information corresponding to the local area of ​​the seedling index spatial distribution map are determined; the deviation information between the coverage information and leaf area index information and the corresponding data points of the suitable growth indicator dynamic curve for rice is calculated; based on the deviation information, a preset deviation-topdressing application data table is queried; and based on the query results combined with the seedling index spatial distribution map, a topdressing application prescription map is generated.

[0101] The beneficial effects of the above technical solution are: it facilitates the accurate determination of the dynamic curve of suitable growth indicators for rice, realizes dynamic monitoring of rice, and facilitates the accurate determination of the topdressing application rate prescription map based on the deviation information of the coverage information and leaf area index information from the corresponding data points (including the coverage information and leaf area index information) of the dynamic curve of suitable growth indicators for rice.

[0102] According to some embodiments of the present invention, a topdressing application rate prescription map is generated based on the query results and the spatial distribution map of the seedling condition index, including:

[0103] Based on the query results, determine the amount of topdressing to be applied in the corresponding local area of ​​the spatial distribution map of the seedling condition index.

[0104] At the boundaries of local areas with different topdressing application rates, the topdressing application rate corresponding to the boundary of the spatial distribution map of seedling condition index is determined based on the S-curve interpolation method.

[0105] Based on the topdressing application amount corresponding to the local area and the topdressing application amount at the corresponding boundary of the seedling condition index spatial distribution map, a topdressing application amount prescription map is generated.

[0106] The working principle of the above technical solution is as follows: Based on the query results, the topdressing application amount corresponding to the local area of ​​the seedling condition index spatial distribution map is determined, that is, the topdressing application amount of each local area. However, there are boundary areas between each local area. For the boundary areas, the topdressing application amount corresponding to the boundary of the seedling condition index spatial distribution map is determined based on the S-curve interpolation method. Based on the topdressing application amount of the local area corresponding to the seedling condition index spatial distribution map and the topdressing application amount of the corresponding boundary, a topdressing application amount prescription map is generated.

[0107] The beneficial effects of the above technical solution are as follows: considering the application amount of topdressing based on the spatial distribution map of seedling condition index corresponding to the local area and the application amount of topdressing based on the corresponding boundary, it avoids unreasonable topdressing in the boundary area and improves the accuracy of the topdressing application amount prescription map.

[0108] According to some embodiments of the present invention, determining the topdressing application amount corresponding to the boundary of the spatial distribution map of seedling condition index based on the S-curve interpolation method includes:

[0109] N(d)=N1+(N2-N1)×[1 / (1+e -k×(d-d0) )]

[0110] Where N(d) is the amount of topdressing applied at a distance d from the boundary; N1 and N2 are the local topdressing applications on both sides; e is the natural constant; k is the smoothing coefficient; d is the distance from the boundary; and d0 is the boundary position.

[0111] The working principle and beneficial effects of the above technical solution are as follows: k is a smoothing coefficient, which controls the transition speed of the curve from the initial state to the stable state. The larger the k value, the steeper the curve, meaning that the change in topdressing amount from N1 to N2 is more rapid; the smaller the k value, the smoother the curve, and the more gradual the change. d is the distance from the boundary, which determines the amount of topdressing applied at different spatial locations. d0 is the boundary position, which determines the spatial position of the curve. The two sides of the boundary corresponding to the seedling condition index spatial distribution map may have different seedling conditions, thus requiring different amounts of topdressing. Based on the distance d from the boundary, a smooth transition is made between the topdressing amounts N1 and N2 on both sides, making the distribution of topdressing amount more consistent with the actual changes in seedling condition, avoiding abrupt changes in topdressing amount, and promoting balanced crop growth.

[0112] According to some embodiments of the present invention, determining the navigation path map for a rice farming drone includes:

[0113] The non-topdressing and topdressing areas of the rice growing region are determined according to the topdressing application rate prescription map;

[0114] Obtain 3D map information and obstacle information of the topdressing area, and set the starting point and key points for navigation path planning;

[0115] Set direct flight path and obstacle avoidance path based on 3D map information and obstacle information;

[0116] The navigation path map for rice farming drones is determined based on the starting point and key points, direct flight path and obstacle avoidance path of the navigation path planning.

[0117] The working principle and beneficial effects of the above technical solution are as follows: Based on the topdressing application rate prescription map, the non-topdressing and topdressing areas of the rice growing region are determined; 3D map information and obstacle information of the topdressing area are obtained, and the starting point and key points of the navigation path planning are set; the key points are the boundary points of the area. Direct flight paths and obstacle avoidance paths are set based on the 3D map information and obstacle information; the navigation path map of the rice agricultural drone is determined based on the set starting point and key points, direct flight path, and obstacle avoidance path. This facilitates accurate determination of the navigation path map of the rice agricultural drone.

[0118] According to some embodiments of the present invention, a topdressing application rate prescription map is fused with a navigation path map, and topdressing operations are performed using a rice agricultural drone based on the fusion result, including:

[0119] The fertilizer application rate prescription map and navigation path map are converted to the same coordinate system; the same coordinate system is the UTM projection coordinate system.

[0120] The topdressing application rate prescription map is rasterized in the same coordinate system, and a point vector layer is created. Each point corresponds to a grid center, and a field is added to the grid center to mark the topdressing application rate.

[0121] In the same coordinate system, perform path statistics on the navigation path map, determine the path length, and perform equal interval interpolation. For each interpolated path point, search for the nearest neighbor point in the point vector layer. Based on the search results, determine several fusion points between the topdressing application rate prescription map and the navigation path map.

[0122] Preliminary fusion is performed based on several fusion points to obtain preliminary fusion results;

[0123] Based on the preliminary fusion results and the UAV's flight parameters, the final fusion result is obtained; the flight parameters include flight speed, flight altitude, and flight direction.

[0124] Topdressing operations were carried out using agricultural drones for rice cultivation based on the fusion results.

[0125] The working principle of the above technical solution is as follows: The topdressing application rate prescription map and the navigation path map are converted to the same coordinate system; the topdressing application rate prescription map is rasterized. An appropriate raster size is set, such as 1m × 1m. A point vector layer is created, with the center of each raster as a point. A field is added to the attribute table of the point vector layer to mark the topdressing application rate corresponding to each point. Path statistics are performed on the navigation path map in the same coordinate system to determine the path length, and equidistant interpolation is performed. For each interpolated path point, the nearest neighbor point is searched in the point vector layer. Based on the search results, several fusion points are determined between the topdressing application rate prescription map and the navigation path map. The information of the determined fusion points (including the path point location and the corresponding topdressing application rate) is integrated to generate a new layer as the preliminary fusion result. This layer contains the navigation path and the fertilizer application rate information corresponding to each path point. Based on the preliminary fusion results and the flight parameters of the UAV, the final fusion result is obtained; the flight parameters include flight speed, flight altitude and flight direction; that is, the flight parameters of the corresponding UAV are matched according to the amount of topdressing applied, and topdressing operations are carried out by rice agricultural UAVs based on the fusion results.

[0126] The beneficial effects of the above technical solution are as follows: Converting the topdressing application rate prescription map and navigation path map to the same coordinate system facilitates image fusion processing. The topdressing application rate prescription map and navigation path map are processed separately in the same coordinate system to determine several fusion points. Preliminary fusion is performed based on these fusion points to obtain a preliminary fusion result. Combined with the UAV's flight parameters, the final fusion result is obtained. Based on the fusion result, topdressing operations are performed using rice agricultural UAVs, improving the accuracy of topdressing operations using rice agricultural UAVs.

[0127] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for precise topdressing of rice using unmanned aerial vehicles (UAVs), characterized in that, include: Monitor rice seedling growth information during the rice growth process and construct a spatial distribution map of seedling condition index; Based on the spatial distribution map of seedling condition index and the preset dynamic curve of suitable rice growth index, a topdressing application prescription map is generated. Determine the navigation path map for rice farming drones; The topdressing application rate prescription map and navigation path map are integrated, and the topdressing operation is carried out by rice agricultural drones based on the final integration result; Monitoring rice seedling growth information during the growth process and constructing a spatial distribution map of seedling condition index, including: Based on the drone's flight over the rice growing area, the first monitoring image is acquired using the onboard RGB camera, and the second monitoring image is acquired using the onboard multispectral camera; The rice cover information was determined based on the first monitoring image; The leaf area index information of rice was determined based on the second monitoring image; Based on the coverage information and leaf area index information, seedling growth information is determined, and a spatial distribution map of seedling index is constructed. Based on the spatial distribution map of seedling condition index and the preset dynamic curve of suitable rice growth indicators, a topdressing application prescription map is generated, including: The dynamic curve of the preset suitable growth index for rice is determined based on historical experimental data and agricultural models; the historical experimental data are the dynamic changes of the coverage information and leaf area index information of the same crop at different yield levels; the agricultural model is the growth curve of rice growth model simulating ideal conditions. Determine the local coverage information and leaf area index information corresponding to the spatial distribution map of seedling condition index; The deviation information between the coverage information and leaf area index information and the corresponding data points of the dynamic curve of suitable growth indicators for rice is calculated. Based on the deviation information, the preset deviation-topdressing amount data table is queried. Based on the query results and the spatial distribution map of seedling condition index, a topdressing application amount operation prescription map is generated. Based on the query results and the spatial distribution map of the seedling condition index, a topdressing application prescription map is generated, including: Based on the query results, determine the amount of topdressing to be applied in the corresponding local area of ​​the spatial distribution map of the seedling condition index. At the boundaries of local areas with different topdressing application rates, the topdressing application rate corresponding to the boundary of the spatial distribution map of seedling condition index is determined based on the S-curve interpolation method. Based on the topdressing application amount corresponding to the local area and the topdressing application amount corresponding to the boundary of the seedling index spatial distribution map, a topdressing application amount operation prescription map is generated. The topdressing application rate was determined based on the S-curve interpolation method, corresponding to the boundary of the spatial distribution map of the seedling condition index, including: ； in, The corresponding distance from the boundary is The amount of topdressing applied; , This refers to the amount of topdressing applied to the local areas on both sides; It is a natural constant; For smoothing coefficients; This represents the distance from the boundary. This represents the boundary position.

2. The method for precise topdressing of rice using unmanned aerial vehicles as described in claim 1, characterized in that, The rice cover information was determined based on the first monitoring image, including: The first monitoring image is stitched together to obtain a panoramic image; Histogram equalization and contrast stretching are performed on the panoramic image to obtain an enhanced image; The enhanced image is then denoised to obtain a denoised image. The denoised image is segmented based on color features to obtain a first segmentation result; the denoised image is segmented based on texture features to obtain a second segmentation result; the crop region is determined by mutual correction based on the first and second segmentation results. The ratio of the number of pixels in the crop area to the number of pixels in the monitoring area corresponding to the denoised image is calculated to determine the rice coverage information.

3. The method for precise topdressing of rice using unmanned aerial vehicles as described in claim 1, characterized in that, The leaf area index information of rice was determined based on the second monitoring image, including: Generate orthophoto maps of reflectance for each band based on the second monitoring image; Based on grayboard data, the DN values ​​of the reflectance orthophoto map of each band are converted into surface reflectance values. Band registration is performed on the reflectance orthophoto maps of all bands to determine the multispectral band reflectance data. The normalized differential vegetation index was calculated based on multispectral reflectance data. The leaf area index of rice was determined by fitting the normalized differential vegetation index to a regression analysis model.

4. The method for precise topdressing of rice using unmanned aerial vehicles as described in claim 1, characterized in that, Based on the coverage information and leaf area index information, seedling growth information is determined, and a spatial distribution map of seedling index is constructed, including: Based on the coverage information, growth stage assessment and growth uniformity assessment are performed to obtain the first assessment result; Based on leaf area index information, photosynthetic capacity and growth potential were assessed to obtain the second assessment result; Based on the results of the first and second assessments and a pre-set data table, the seedling growth level is determined as seedling growth information; a spatial distribution map of the seedling index is constructed based on the seedling growth information.

5. The method for precise topdressing of rice using unmanned aerial vehicles as described in claim 1, characterized in that, Determine the navigation path map for rice farming drones, including: The non-topdressing and topdressing areas of the rice growing region are determined based on the topdressing application rate prescription map; Obtain 3D map information and obstacle information of the topdressing area, and set the starting point and key points for navigation path planning; Set direct flight path and obstacle avoidance path based on 3D map information and obstacle information; The navigation path map for rice farming drones is determined based on the starting point and key points, direct flight path and obstacle avoidance path of the navigation path planning.

6. The method for precise topdressing of rice using unmanned aerial vehicles as described in claim 1, characterized in that, The topdressing application prescription map and navigation path map are integrated, and topdressing operations are carried out using rice agricultural drones based on the final fusion result, including: The fertilizer application rate prescription map and navigation path map are converted to the same coordinate system; the same coordinate system is the UTM projection coordinate system. The topdressing application rate prescription map is rasterized in the same coordinate system, and a point vector layer is created. Each point corresponds to a grid center, and a field is added to the grid center to mark the topdressing application rate. In the same coordinate system, the navigation path map is statistically analyzed to determine the path length and perform equidistant interpolation. For each interpolated path point, the nearest neighbor point is searched in the point vector layer. Based on the search results, several fusion points of the topdressing application rate prescription map and the navigation path map are determined. Preliminary fusion is performed based on several fusion points to obtain preliminary fusion results; Based on the preliminary fusion results and the UAV's flight parameters, the final fusion result is obtained; the flight parameters include flight speed, flight altitude, and flight direction. Based on the final fusion results, topdressing operations were carried out using agricultural drones for rice cultivation.

Images