An intelligent seeding operation terminal platform based on visual perception

Abstract

The present invention relates to the field of image recognition, and specifically to a terminal platform for intelligent seeding operations based on visual perception. It performs dynamic evolution analysis on the farmland geographical images according to the soil moisture conditions of the farmland, calculates the moisture characteristics of each seeding area, and labels each seeding area as a first-class area or a second-class area; it performs moisture spatial domain analysis based on the geomorphic elements of the second-class area to determine the effective seeding trajectory of the second-class area; it compares the geographical distribution images of the first-class area and the second-class area with the visual perception images generated by the seeding machine to determine whether the seeding machine enters the first-class area or the second-class area; it visually identifies the deviation between the actual seeding trajectory and the effective seeding trajectory in the second-class area, and adjusts the seed reseeding in the second-class area accordingly. The present invention can guide seeding operations through visual perception from the global scope to the local scope of the farmland, improve the seeding operation efficiency, and ensure the overall germination rate of the farmland.

Description

Technical Field

[0001] The present invention relates to the field of image recognition, and in particular to a terminal platform for intelligent seeding operations based on visual perception. Background Art

[0002] Spring plowing operations mainly involve sowing seeds on farmland. Whether the seeding time is appropriate directly affects the germination rate of seeds. Once the appropriate seeding time is missed, it will directly affect the growth trend and yield of crops, and may lead to crop failures or no harvests. To determine the appropriate seeding time, the weather during the spring plowing period can be predicted, mainly for accurate prediction of aspects such as temperature and rainfall. Through research, it has been found that the soil moisture content of farmland is an important indicator for judging whether the seeding time is appropriate. When the soil moisture content is at a relatively high level, the seeds sown in the soil can obtain the heat and moisture required for germination in a timely and sufficient manner, which is conducive to improving the seed germination rate. By monitoring the soil moisture content of farmland, a reliable reference basis can be provided for seeding operations.

[0003] Existing seeding operations all judge whether it is currently suitable for seeding based on the soil moisture content of the entire farmland. However, considering the vast area of farmland in the planting area, the seeding time of the entire farmland may be delayed due to insufficient soil moisture content in local areas. If the same seeding operation is carried out for the entire farmland area without optimizing the seeding operation in the areas with insufficient soil moisture content, the problem of low seed germination rate in the areas with insufficient soil moisture content cannot be compensated. How to achieve differential identification of soil moisture content through visual perception of the entire farmland area and adjust seed placement through visual perception of local farmland areas is of great significance for saving the workload of seeding operations, improving the efficiency of seeding operations, and ensuring the overall germination rate of farmland. Summary of the Invention

[0004] In order to avoid the low seed germination rate in the areas with insufficient local soil moisture content in the farmland caused by the same seeding operation for the entire farmland, it is necessary to achieve differential identification of soil moisture content through visual perception of the entire farmland area and adjust seed placement through visual perception of local farmland areas, so as to realize visual perception-guided seeding operations from the entire farmland area to local areas. The present invention provides a terminal platform for intelligent seeding operations based on visual perception, and the terminal platform includes the following modules:

[0005] A soil moisture evolution analysis module, which is used to perform dynamic evolution analysis on the farmland geographical image according to the soil moisture content of the farmland and calculate the soil moisture characteristics of each seeding area;

[0006] An area calibration module, which is used to calibrate each seeding area as a first type of area or a second type of area respectively according to the soil moisture characteristics, and visually identify the landform elements of the second type of area;

[0007] A trajectory determination module, configured to perform soil moisture spatial analysis on the second type of area according to the geomorphic elements, and determine an effective sowing trajectory for the second type of area;

[0008] An image comparison module, configured to generate geographical distribution images of all the first type of areas and the second type of areas, and compare them with the visual perception images generated by the seeder, to determine whether the seeder enters the first type of area or the second type of area;

[0009] A trajectory deviation identification module, configured to, when the seeder enters the second type of area, visually identify the deviation between the actual sowing trajectory of the seeder and the effective sowing trajectory;

[0010] An adjustment module, configured to adjust the seed reseeding of the seeder in the second type of area according to the deviation.

[0011] Preferably, before the soil moisture evolution analysis module performs dynamic evolution analysis on the farmland geographical image according to the soil moisture of the farmland, it includes:

[0012] Performing thermal infrared visual recognition on the entire farmland range at several time points respectively, to obtain the surface temperature and soil moisture content of the entire farmland range at each of the several time points; wherein, the several time points at least include the daytime time points and nighttime time points of several dates;

[0013] Performing inversion calculation on the surface temperature and soil moisture content at each time point to obtain the soil moisture of the farmland at each time point.

[0014] Preferably, the soil moisture evolution analysis module performs dynamic evolution analysis on the farmland geographical image according to the soil moisture of the farmland, and calculates the soil moisture characteristics of each sowing area, specifically:

[0015] Mapping the soil moisture of the farmland at each time point to the farmland geographical image divided with the boundaries of the sowing areas, and performing edge sharpening processing on the soil moisture of all the sowing areas within the farmland geographical image;

[0016] Performing dynamic time evolution analysis on the soil moisture of each sowing area within the farmland geographical image at all time points, and calculating the soil moisture characteristics of each sowing area within the corresponding time interval at all time points; wherein, the soil moisture characteristics include the moisture fluctuation change of each sowing area within the time interval.

[0017] Preferably, the area calibration module calibrates each sowing area as the first type of area or the second type of area respectively according to the soil moisture characteristics, specifically:

[0018] Estimate the cumulative time length during which each sowing area is in a stable soil moisture state according to the fluctuating changes in soil moisture in each sowing area included in the soil moisture characteristics within the corresponding time intervals at all time points; wherein, the stable soil moisture state refers to a state where the difference between the average soil moisture in the sowing area and the reference soil moisture is within a preset range.

[0019] If the cumulative time length is greater than or equal to the preset time length threshold, label the sowing area as a first type of area; if the cumulative time length is less than the preset time length threshold, label the sowing area as a second type of area.

[0020] Preferably, the area calibration module visually identifies the geomorphic elements of the second type of area, specifically:

[0021] Identify the geomorphic elements of the second type of area by recognizing the remote sensing image of the entire farmland range according to the boundary of the second type of area; wherein, the geomorphic elements include the surface contour elements of the second type of area.

[0022] Preferably, the trajectory determination module performs soil moisture spatial analysis on the second type of area according to the geomorphic elements to determine the effective sowing trajectory of the second type of area, specifically:

[0023] Perform surface space mapping on the soil moisture of the second type of area according to the surface contour elements of the second type of area included in the geomorphic elements to obtain the soil moisture distribution map of the second type of area;

[0024] Calculate the soil moisture distribution spatial vector of the grid soil moisture matrix of the soil moisture distribution map, perform cluster analysis on the soil moisture distribution spatial vector, and determine the grid areas with sufficient soil moisture within the second type of area;

[0025] Determine the effective sowing trajectory of the second type of area according to the positions of the grid areas with sufficient soil moisture.

[0026] Preferably, the image comparison module generates the geographical distribution images of all the first type of areas and the second type of areas, and compares them with the visual perception images generated by the seeder to determine whether the seeder enters the first type of area or the second type of area, specifically:

[0027] Generate the geographical distribution images of all the first type of areas and the second type of areas according to the geographical boundary positioning data of the first type of areas and the second type of areas; wherein, the geographical distribution images are marked with the respective boundaries of the first type of areas and the second type of areas.

[0028] Compare the visually perceived GIS images generated locally by the seeder with the geographical distribution images in the same spatial coordinate system to determine the relative positional relationship between the seeder and the boundaries of the first type of area and the second type of area during the operation, so as to determine whether the seeder enters the first type of area or the second type of area.

[0029] Preferably, when the seeder enters the second type of area, the trajectory deviation recognition module is used to visually recognize the deviation between the actual seeding trajectory of the seeder and the effective seeding trajectory, specifically:

[0030] When the seeder enters the second type of area during the operation, collect the image of the seed cluster that has been placed in the second type of area;

[0031] Perform background separation and seed contour recognition on the seed cluster image to determine the spatial distribution positions of the germination points of all the seeds that have been placed;

[0032] Perform continuous fitting processing on the spatial distribution positions of the germination points of the seeds to obtain the actual seeding trajectory;

[0033] Perform a spatial comparison between the actual seeding trajectory and the effective seeding trajectory to determine the deviation between the two; wherein, the deviation includes the distance deviation and the azimuth deviation between the actual seeding trajectory and the effective seeding trajectory.

[0034] Preferably, the adjustment module is used to adjust the supplementary seeding of the seeder in the second type of area according to the deviation, specifically:

[0035] Estimate the seed placement blank range between the actual seeding trajectory and the effective seeding trajectory according to the distance deviation and the azimuth deviation between the actual seeding trajectory and the effective seeding trajectory included in the deviation;

[0036] Adjust the seed supplementary placement density of the seeder in the seed placement blank range according to the seed placement blank range and the soil moisture content along the effective seeding trajectory.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] Visually perceive the entire farmland to realize the differential recognition of soil moisture content, and visually perceive the local area of the farmland to realize the adjustment of seed placement, so as to realize the visual perception-guided seeding operation from the global range to the local range of the farmland, improve the seeding operation efficiency and ensure the germination rate of the whole farmland. Brief Description of the Drawings

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings. Among them:

[0040] Figure 1 is a schematic structural diagram of a terminal platform for an intelligent seeding operation based on visual perception provided by the present invention.

[0041] Figure 2 is the soil moisture distribution of the farmland geographical image of the present invention.

[0042] Figure 3 is the distribution of the first type of area and the second type of area in the farmland of the present invention.

[0043] Figure 4 is a schematic diagram for determining the effective seeding trajectory of the present invention.

[0044] Figure 5 is a schematic diagram for determining the actual seeding trajectory of the present invention.

[0045] Figure 6 is a schematic diagram of the deviation between the effective seeding trajectory and the actual seeding trajectory of the present invention. Detailed Embodiments

[0046] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings. It can be understood that the specific embodiments described herein are only used to explain the present invention, rather than limiting the present invention. Additionally, it should be noted that for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings, rather than all the structures. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0047] The terms "including" and "having" and any variations thereof in the present invention are intended to cover non-exclusive inclusion. For example, a process, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally further include steps or units not listed, or may optionally further include other steps or units inherent to these processes, products, or devices.

[0048] Reference to "embodiment" in this document means that the specific features, structures, or characteristics described in connection with the embodiment can be included in at least one embodiment of the present invention. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0049] Please refer to Figure 1 As shown, the present invention provides a terminal platform for intelligent seeding operations based on visual perception. The terminal platform includes the following modules:

[0050] The soil moisture evolution analysis module is used to perform dynamic evolution analysis on the farmland geographical image according to the soil moisture of the farmland and calculate the soil moisture characteristics of each seeding area;

[0051] The area calibration module is used to calibrate each seeding area as a first-class area or a second-class area according to the soil moisture characteristics and visually identify the geomorphic elements of the second-class area;

[0052] The trajectory determination module is used to perform soil moisture airspace analysis on the second-class area according to the geomorphic elements and determine the effective seeding trajectory of the second-class area;

[0053] The image comparison module is used to generate the geographical distribution images of all the first-class areas and the second-class areas and compare them with the visual perception images generated by the seeder to determine whether the seeder enters the first-class area or the second-class area;

[0054] The trajectory deviation identification module is used to visually identify the deviation between the actual seeding trajectory of the seeder and the effective seeding trajectory when the seeder enters the second-class area;

[0055] The adjustment module is used to adjust the seed supplementary sowing of the seeder in the second-class area according to the deviation.

[0056] During the seeding operation, accurately predicting the soil moisture of the farmland plays a decisive role in whether the seeds can germinate normally after sowing; insufficient soil moisture will cause the seeds to be unable to obtain sufficient heat and moisture in the soil and cannot germinate at the right time. Once the seeds germinate late or cannot germinate, it will affect the crop yield; considering that the soil moisture changes both at different dates and in the early and late time periods of the same date, too large a difference in soil moisture changes is not conducive to the germination and growth of seeds. Therefore, it is necessary to perform dynamic evolution analysis on the farmland geographical image level according to the soil moisture in the global scope of the farmland to obtain the soil moisture characteristics of seeding areas such as seed placement pits in the farmland, and accurately distinguish and judge whether each seeding area has sufficient soil moisture to meet the seed germination requirements.

[0057] Considering the vast area of farmland, there may be significant differences in soil moisture conditions in different sowing areas within the farmland. For example, the soil moisture in some sowing areas is sufficient, and only by sowing according to the conventional seed placement mode can a high germination rate of seeds in these sowing areas be ensured; while the soil moisture in other sowing areas is insufficient. If sowing is carried out according to the conventional seed placement mode, some seeds will not be able to obtain sufficient heat and moisture for germination, which will affect the germination rate. At this time, it is necessary to improve the seed placement state under the original conventional seed placement mode. Therefore, according to the soil moisture characteristics, each sowing area needs to be respectively designated as the first type of area with sufficient soil moisture or the second type of area without sufficient soil moisture, so as to facilitate subsequent sowing adjustment only for the second type of area. The first type of area can maintain the original conventional seed placement mode, taking into account the sowing efficiency of the overall farmland area and the sowing optimization of the local farmland area.

[0058] For the second type of area without sufficient soil moisture, there are significant differences in the internal soil moisture distribution. For example, in some directions within the second type of area, the soil moisture is sufficient, and the range along these directions can provide relatively more heat and moisture for the seeds. If seeds are placed along the range along these directions, it can ensure a high germination rate of the seeds. However, if seeds are placed outside the range along these directions, the seed germination rate will be severely reduced. It can be seen that the range along these directions within the second type of area forms an effective sowing trajectory, which can provide a spatial position reference for subsequent optimization of the sowing range in the second type of area, facilitating the alignment of seed placement with the effective sowing trajectory as much as possible.

[0059] Considering the differences in soil moisture distribution between the first type of area and the second type of area, during the actual sowing operation process, only the sowing operation in the second type of area needs to be optimized, and for the first type of area, only the original conventional seed placement mode needs to be maintained. In order to accurately and timely implement the sowing operation change when switching from the first type of area to the second type of area, it is necessary to determine whether the seed drill is currently in the first type of area or the second type of area, so as to facilitate the seed drill to adjust the sowing mode in a timely manner. Therefore, generate the geographical distribution images of all the first type of areas and the second type of areas, and compare them with the visual perception images generated by the seed drill. Through visual perception recognition, it is possible to accurately locate whether the seed drill is in the first type of area or the second type of area, and timely and precisely achieve the optimized sowing in the second type of area.

[0060] During the process of sowing operations on agricultural land, the seeder will carry out sowing in the conventional seed placement mode for all first-class areas and all second-class areas, that is, sowing operations are carried out according to the preset conventional seed placement spatial density and placement direction. Considering the insufficient soil moisture in the second-class areas, relying solely on the above conventional seed placement mode for sowing cannot guarantee the seed germination rate in the second-class areas, and it is necessary to further implement optimized sowing on the basis of the above conventional seed placement mode for sowing. The purpose of implementing optimized sowing for the second-class areas is to enable the seeds to be placed within a range consistent with the effective sowing trajectory. For this reason, when the seeder enters the second-class areas, the visual recognition system recognizes the actual sowing trajectory formed after the seeder implements the above conventional seed placement mode for sowing, and compares the actual sowing trajectory with the effective sowing trajectory to determine the deviation between the two, providing a reference for further optimizing sowing in the second-class areas in the future.

[0061] During the process of further implementing optimized sowing for the second-class areas, it is necessary to ensure that the seeds for supplementary sowing are as consistent as possible with the effective sowing trajectory in the second-class areas. For this reason, according to the deviation between the actual sowing trajectory and the effective sowing trajectory in the second-class areas, the operable area range for the seeds of supplementary sowing is determined, so as to ensure that the seeds of supplementary sowing fall into the range consistent with the effective sowing trajectory to the greatest extent and will not overlap with the actual sowing trajectory.

[0062] Furthermore, before the soil moisture evolution analysis module conducts dynamic evolution analysis on the farmland geographical image according to the soil moisture of the farmland, it includes:

[0063] Thermal infrared visual recognition is carried out on the entire farmland range at several time points respectively to obtain the surface temperature and soil moisture content of the entire farmland range at each of the several time points; among them, the several time points at least include the daytime and nighttime time points of several dates;

[0064] Inverse calculation is carried out on the surface temperature and soil moisture content at each time point to obtain the soil moisture of the farmland at each time point.

[0065] Soil moisture is an important parameter index reflecting the soil state, and it has important reference significance for agricultural operations such as sowing. Soil moisture is related to the soil surface temperature and soil moisture content. By monitoring and analyzing the soil surface temperature data and soil moisture content data, a linear regression relationship between the soil surface temperature and the soil moisture content is established, and then inverse algorithm transformation processing is carried out on the above linear regression relationship to obtain the soil moisture.

[0066] Considering the vast area of farmland, if the distributed sampling method is used to collect soil surface temperature data and soil water content at several location points in the farmland, the obtained soil surface temperature data and soil water content data cannot comprehensively and accurately reflect the actual surface temperature and soil water content status of the entire farmland, resulting in a large error in the retrieved soil moisture content and reducing the credibility of the soil moisture content. Therefore, remote sensing thermal infrared images of the entire farmland can be collected and analyzed, and thermal infrared vision recognition can be carried out on the entire farmland to obtain the surface temperature and soil water content of the entire farmland. Considering that the soil surface temperature and soil water content are affected by the external environment. For example, during the day, the external environmental temperature and humidity are relatively high, and the soil surface temperature and soil water content are also relatively high; during the night, the external environmental temperature and humidity are relatively low, and the soil surface temperature and soil water content are also relatively low. In this way, the soil moisture content of the same farmland will fluctuate at different dates and during the morning and evening periods of the same date. As Figure 2 shown, (a) shows the soil moisture content distribution of the farmland at a certain time point (such as 13:00) during the day of a certain date, and (b) shows the soil moisture content distribution of the farmland at a certain time point (such as 21:00) during the night of a certain date. By comparing (a) and (b), it can be found that there are significant differences in the soil moisture content distribution of the same farmland at different time points within the same date. If only a single time point within a certain date is used as the basis for calculating the soil moisture content distribution of the entire farmland, the fluctuating changes in the soil moisture content distribution of the entire farmland cannot be accurately reflected, thus affecting the accuracy of subsequent distinguishing the first type of area and the second type of area within the entire farmland. Therefore, thermal infrared vision recognition is carried out on the entire farmland at several time points respectively, the surface temperature and soil water content of the entire farmland at the daytime time points and evening time points of several dates are obtained, and the soil moisture content distribution of the entire farmland at each time point is obtained through inversion calculation of the surface temperature and soil water content at each time point, providing a basis for the soil moisture content stability characteristics of all sown areas within the farmland from the time domain change level.

[0067] Furthermore, the soil moisture content evolution analysis module performs dynamic evolution analysis on the farmland geographical image according to the soil moisture content of the farmland, and calculates the soil moisture content characteristics of each sown area, specifically:

[0068] Map the soil moisture content of the farmland at each time point to the farmland geographical image divided by the boundaries of the sown areas, and perform edge sharpening processing on the soil moisture content of all sown areas within the farmland geographical image;

[0069] Perform dynamic time evolution analysis on the soil moisture content of each sown area within the farmland geographical image at all time points, and calculate the soil moisture content characteristics of each sown area within the corresponding time interval at all time points; among them, the soil moisture content characteristics include the fluctuating changes in the soil moisture content of each sown area within the time interval.

[0070] After sowing seeds in farmland, the seeds rely on the heat and moisture in the farmland soil to germinate. If the soil cannot continuously and stably supply heat and moisture to the seeds, the health and germination of the seeds cannot be guaranteed. Considering that it takes some time from sowing to seed germination, if the soil moisture changes greatly during this period, the seeds cannot continuously and stably obtain sufficient heat and moisture and will not germinate normally. Before sowing, the farmland will be plowed and a large number of seed placement pits and other sowing areas will be formed on the farmland. There are differences in the fluctuation of soil moisture in each sowing area; for example, the fluctuation range of soil moisture in some sowing areas is small and can remain relatively stable for a long time, so that heat and moisture can be continuously and stably supplied to the seeds; in other sowing areas, the fluctuation range of soil moisture is large and cannot remain relatively stable for a long time, so that heat and moisture cannot be continuously and stably supplied to the seeds.

[0071] In order to accurately judge whether each sowing area in the farmland can continuously and stably supply heat and moisture to the seeds, it is necessary to identify and analyze the change of soil moisture in the geographical image of the entire farmland; specifically, map the soil moisture of the farmland at each time point to the farmland geographical image with the boundary of the sowing area divided and perform edge sharpening processing on the soil moisture, so as to strengthen the resolution of soil moisture in different sowing areas on the farmland geographical image; then perform dynamic time evolution analysis on the soil moisture of each sowing area in the farmland geographical image at all time points, calculate the moisture fluctuation change of each sowing area in the corresponding time interval at all time points, and realize the accurate characterization of the change amplitude of soil moisture in each sowing area.

[0072] Furthermore, the region calibration module calibrates each sowing area as the first type of area or the second type of area according to the soil moisture characteristics, specifically:

[0073] According to the moisture fluctuation change of each sowing area in the corresponding time interval at all time points included in the soil moisture characteristics, estimate the cumulative time length of each sowing area in a stable soil moisture state; among them, the stable soil moisture state refers to the state where the difference between the average soil moisture of the sowing area and the reference soil moisture is within the preset range;

[0074] If the cumulative time length is greater than or equal to the preset time length threshold, the sowing area is calibrated as the first type of area; if the cumulative time length is less than the preset time length threshold, the sowing area is calibrated as the second type of area.

[0075] When the soil moisture in the sowing area within the farmland remains relatively stable and the fluctuations in soil moisture are relatively small during the corresponding time interval, it indicates that the above sowing area can continuously and stably provide heat and moisture for the seeds; when the soil moisture in the sowing area within the farmland cannot maintain a stable state and the fluctuations in soil moisture are relatively large during the corresponding time interval, it indicates that the above sowing area cannot continuously and stably provide heat and moisture for the seeds. According to the above analysis, for some sowing areas with sufficient and stable soil moisture, only sowing according to the conventional seed placement mode can ensure a high germination rate of the seeds in these sowing areas; for some sowing areas with unstable soil moisture, sowing according to the conventional seed placement mode will cause some seeds to fail to obtain sufficient heat and moisture for germination, affecting the germination rate. At this time, it is necessary to improve the seed placement state under the original conventional seed placement mode.

[0076] In order to implement zonal operation for the sowing operation in the farmland, it is necessary to distinguish and demarcate all sowing areas within the farmland. Specifically, according to the fluctuations in soil moisture of each sowing area within the corresponding time intervals at all time points, estimate the cumulative time length during which each sowing area is in a stable soil moisture state. The larger the cumulative time length during which the sowing area is in a stable soil moisture state, the more capable the sowing area is of continuously and stably providing heat and moisture for the seeds. Then, compare the cumulative time length during which the sowing area is in a stable soil moisture state with a threshold value, and demarcate each sowing area as a first-class area or a second-class area respectively. As Figure 3 shown, for all sowing areas within the farmland, the sowing area corresponding to the red frame belongs to the first-class area, and the sowing area corresponding to the green frame belongs to the second-class area. Through the above method, each sowing area is demarcated as a first-class area with sufficient soil moisture or a second-class area without sufficient soil moisture, which is convenient for subsequent sowing adjustment only for the second-class areas. The second-class areas can maintain the original conventional seed placement mode, taking into account the sowing efficiency of the entire farmland area and the sowing optimization of the local farmland area.

[0077] Furthermore, the area demarcation module visually identifies the geomorphic elements of the second-class areas, specifically:

[0078] Identify the remote sensing image of the entire farmland area according to the boundary of the second-class areas to obtain the geomorphic elements of the second-class areas; among them, the geomorphic elements include the surface contour elements of the second-class areas.

[0079] The second type of area within the farmland does not have sufficient soil moisture, and the seeds sown in the second type of area cannot continuously and stably obtain heat and moisture from the soil in the second type of area. To improve the germination rate of seeds in the second type of area, it is necessary to optimize the seeding of seeds in the second type of area so that the seeds can obtain as much heat and moisture as possible from the second type of area to ensure the survival and germination rate of the seeds. Considering the spatial differences in soil moisture within the second type of area, that is, the soil moisture in some intervals within the second type of area is large, while in other intervals it is small, based on the boundary of the second type of area, remote sensing images of the entire farmland area are identified to obtain the surface contour elements of the second type of area, which can comprehensively represent the surface contour shape of the second type of area and provide a basis for subsequent refinement of the soil moisture distribution within the second type of area.

[0080] Furthermore, the trajectory determination module conducts soil moisture spatial analysis on the second type of area according to the geomorphic elements to determine the effective seeding trajectory of the second type of area, specifically as follows:

[0081] Based on the surface contour elements of the second type of area included in the geomorphic elements, conduct surface space mapping of the soil moisture in the second type of area to obtain the soil moisture distribution map of the second type of area;

[0082] Calculate the soil moisture distribution spatial vector of the grid-based soil moisture matrix of the soil moisture distribution map, conduct cluster analysis on the soil moisture distribution spatial vector, and determine the grid areas with sufficient soil moisture within the second type of area;

[0083] Based on the positions of the grid areas with sufficient soil moisture, determine the effective seeding trajectory of the second type of area.

[0084] The soil moisture within the second type of area also shows an uneven distribution trend, with high and low differences in soil moisture in different intervals. To accurately determine the positions within the second type of area that can provide more heat and moisture, first conduct surface space mapping of the same coordinate system for the surface contour elements of the second type of area and the soil moisture in the second type of area to obtain the soil moisture distribution map of the second type of area. The above-mentioned soil moisture distribution map characterizes the size distribution of soil moisture on the global surface of the second type of area.

[0085] After equally spaced grid division of the above-mentioned soil moisture distribution map, a grid-based soil moisture matrix is obtained. Conduct soil moisture size spatial gradient simulation calculation on the above-mentioned grid-based soil moisture matrix to obtain the soil moisture distribution spatial vector; then conduct neural network model cluster analysis on all soil moisture distribution spatial vectors to determine the grid areas with sufficient soil moisture within the second type of area; among them, the grid areas with sufficient soil moisture refer to the grid cluster areas within the second type of area where the soil moisture is greater than the preset soil moisture threshold and are adjacent in spatial distribution. Finally, based on the positions of the grid areas with sufficient soil moisture, conduct continuous simulation of the spatial distribution of the grid areas to determine the effective seeding trajectory of the second type of area. For example Figure 4As shown, multiple grid spaces are equally spaced within the second type of area, and each grid space has a corresponding soil moisture level. Among them, the soil moisture in the white grid spaces is relatively low (for example, less than a preset soil moisture threshold), and the soil moisture in the gray grid spaces is relatively high (for example, greater than or equal to the preset soil moisture threshold). By performing a soil moisture distribution spatial vector sum and clustering analysis on all grid spaces within the second type of area, several effective seeding trajectories within the second type of area are obtained. It can be considered that each effective seeding trajectory and its adjacent range within the second type of area can provide more and more continuous heat and moisture supply for the seeds compared to other spatial ranges, providing a spatial position reference for subsequent optimization of the seeding range in the second type of area, and facilitating the seed placement to be as consistent as possible with the effective seeding trajectories.

[0086] Furthermore, the image comparison module generates geographical distribution images of all the first type of areas and the second type of areas, and compares them with the visual perception images generated by the seeding machine to determine whether the seeding machine enters the first type of area or the second type of area. Specifically:

[0087] Based on the geographical boundary positioning data of the first type of areas and the second type of areas, geographical distribution images of all the first type of areas and the second type of areas are generated; among them, the geographical distribution images are marked with the respective boundaries of the first type of areas and the second type of areas.

[0088] The visual perception GIS image locally generated by the seeding machine is compared with the geographical distribution image in the same spatial coordinate system to determine the relative position relationship between the seeding machine and the respective boundaries of the first type of area and the second type of area during the operation process, so as to determine whether the seeding machine enters the first type of area or the second type of area.

[0089] All the first-class areas and second-class areas within the farmland have fixed geographical boundaries. From the above analysis, it can be seen that there are differences in seeding operation requirements for the first-class areas and second-class areas respectively due to different soil moisture conditions. For the first-class areas, only by sowing according to the conventional seed placement mode can a high germination rate of the seeds inside be ensured; for the second-class areas, if sowing is carried out according to the conventional seed placement mode, some seeds will not be able to obtain sufficient heat and moisture for germination, which will affect the germination rate. At this time, on the basis of implementing the sowing operation for the second-class areas according to the conventional seed placement mode, it is also necessary to additionally implement an improved and optimized sowing operation. In order to enable the seeder to carry out differential sowing operations for the first-class areas and second-class areas, it is necessary to first determine the type of area where the seeder is currently located. Specifically, according to the geographical boundary positioning data of the first-class areas and second-class areas, a geographical distribution image marked with the boundaries of the first-class areas and second-class areas respectively is generated; then, the visual perception GIS image locally generated by the built-in GIS device during the working process of the seeder is compared with the geographical distribution image in the same spatial coordinate system to determine the relative distance and relative azimuth angle between the seeder itself and the boundaries of the first-class areas and second-class areas during the operation process, so as to determine whether the seeder enters the first-class area or the second-class area. When the seeder enters the first-class area, control the seeder to sow according to the conventional seed placement mode; when the seeder enters the second-class area, control the seeder to sow according to the conventional seed placement mode and then implement the optimized and improved sowing operation.

[0090] Furthermore, the trajectory deviation recognition module is used to, when the seeder enters the second-class area, visually recognize the deviation between the actual seeding trajectory of the seeder and the effective seeding trajectory. Specifically:

[0091] When the seeder enters the second-class area during the operation process, collect the image of the seed cluster that has been placed in the second-class area;

[0092] Perform background separation and seed contour recognition on the seed cluster image to determine the spatial distribution positions of the germination points of all the seeds that have been placed;

[0093] Perform continuous fitting processing on the spatial distribution positions of the germination points of the seeds to obtain the actual seeding trajectory;

[0094] Perform a spatial comparison between the actual seeding trajectory and the effective seeding trajectory to determine the deviation between the two; among them, the deviation includes the distance deviation and azimuth angle deviation between the actual seeding trajectory and the effective seeding trajectory.

[0095] After the seeder enters the second type of area, it first sows seeds according to the conventional seed placement mode. A certain number of seeds will be placed in the second type of area, and these seeds are distributed inside the second type of area. Seed germination forms young buds in a specific interval on the seed surface, and the distribution position of the bud points on the seed surface is relatively fixed. The distribution position of the bud points of the same type of crop seeds depends on the seed type. The soil moisture in the surrounding space of the seed bud points in the soil will affect the seed germination. The higher the soil moisture in the surrounding space of the seed bud points, the easier it is to promote the germination of the bud points. The greater the distance between the seed bud points and the soil with higher soil moisture, the less likely it is to obtain sufficient heat and moisture, affecting the germination of the seed bud points. In order to enable the seeds sown in the second type of area to obtain sufficient heat and moisture from the soil, and to prevent the seeds sown later from overlapping with the seeds sown at the beginning in space, it is necessary to determine the spatial distribution positions of the germination points of the seeds sown at the beginning, and then perform a continuous fitting process on the spatial distribution positions of the germination points of all seeds to obtain the actual sowing trajectory. As Figure 5 shown, after the initial sowing in the second type of area is completed, visual perception recognition can be carried out to obtain the image of the seed cluster that has been placed in the second type of area, and then background separation and seed contour recognition are performed on the seed cluster image to obtain the spatial distribution positions of the germination points, and finally the actual sowing trajectory is obtained by fitting. Finally, the actual sowing trajectory is compared with the effective sowing trajectory to determine the deviation between the two, providing a reference for further optimizing the sowing in the second type of area in the future.

[0096] Furthermore, the adjustment module is used to adjust the seed supplementary sowing of the seeder in the second type of area according to the deviation, specifically:

[0097] Estimate the seed placement blank range between the actual sowing trajectory and the effective sowing trajectory according to the distance deviation and azimuth deviation between the actual sowing trajectory and the effective sowing trajectory included in the deviation;

[0098] Adjust the seed supplementary placement density of the seeder in the seed placement blank range according to the seed placement blank range and the soil moisture along the effective sowing trajectory.

[0099] As Figure 6As shown, the actual sowing trajectory and the effective sowing trajectory are mapped to the same image space, and the pixel contours of the above two trajectories are recognized to obtain the distance deviation and azimuth deviation between the actual sowing trajectory and the effective sowing trajectory, so as to estimate the seed placement blank range between the actual sowing trajectory and the effective sowing trajectory; the above seed placement blank range can be, but is not limited to, the range where the number of seeds placed per unit area between the actual sowing trajectory and the effective sowing trajectory is less than the preset number threshold. Then, according to the seed placement blank range and the soil moisture along the effective sowing trajectory, the seed supplement placement density of the seeder in the seed placement blank range is adjusted, so that more seeds can be placed in the seed placement blank range and the position with higher soil moisture along the effective sowing trajectory, ensuring that the supplementary sowing seeds can obtain sufficient heat and moisture from the second type of area, and the seed germination rate in the second type of area is increased as much as possible.

[0100] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by adding a necessary general hardware platform, and of course can also be implemented by combining hardware and software. Based on such an understanding, the above technical solution can essentially or in other words be embodied in the form of a computer product, and the present invention can be implemented in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it, and other embodiments may also be used. Although the present invention has been described in detail with reference to the aforementioned embodiments, ordinary technicians in this field should understand that they can still modify the technical solutions recorded in the aforementioned embodiments, or replace some of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A terminal platform for intelligent seeding operations based on visual perception, characterized in that: The terminal platform includes the following modules: The soil moisture evolution analysis module is used to dynamically analyze the evolution of the farmland geographic image according to the soil moisture of the farmland and calculate the soil moisture characteristics of each sowing area; A region marking module, for marking each sowing region as a first type of region or a second type of region according to the soil moisture characteristics, and visually identifying geomorphic elements of the second type of region; According to the soil moisture characteristics, each sowing area is marked as a first-class area or a second-class area, specifically: According to the soil moisture fluctuation changes of each sowing area in the corresponding time interval at all time points included in the soil moisture characteristics, the cumulative time length of each sowing area in a stable soil moisture state is estimated; wherein the stable soil moisture state refers to a state in which the difference between the average soil moisture of the sowing area and the benchmark soil moisture is within a preset range; If the accumulated time length is greater than or equal to the preset time length threshold, the sowing area is marked as a first type of area; if the accumulated time length is less than the preset time length threshold, the sowing area is marked as a second type of area; The trajectory determination module is used to perform a spatial analysis of the soil moisture condition of the second type of area according to the geomorphic elements to determine an effective sowing trajectory for the second type of area, specifically: According to the surface contour elements of the second type of area included in the landform elements, the soil moisture of the second type of area is mapped on the surface space to obtain a soil moisture distribution map of the second type of area; Calculating the soil moisture distribution space vector of the gridded soil moisture matrix of the soil moisture distribution map, performing cluster analysis on the soil moisture distribution space vector, and determining the grid area with sufficient soil moisture in the second type of area, specifically: After the soil moisture distribution map is divided into equally spaced grids, a gridded moisture matrix is ​​obtained, and a spatial gradient simulation calculation of the moisture size is performed on the gridded moisture matrix to obtain a moisture distribution space vector; then a neural network model cluster analysis is performed on all the moisture distribution space vectors to determine the grid area with sufficient moisture in the second type of area; wherein the grid area with sufficient moisture refers to a grid cluster area in the second type of area where the moisture is greater than a preset moisture threshold and is adjacent in spatial distribution; Determining an effective sowing trajectory for the second type of area according to the location of the grid area with sufficient soil moisture; An image comparison module is used to generate a geographical distribution image of all first-category areas and second-category areas, and compare it with the visual perception image generated by the seed drill to determine whether the seed drill enters the first-category area or the second-category area; A trajectory deviation identification module, used for visually identifying the deviation between the actual sowing trajectory of the sowing machine and the effective sowing trajectory when the sowing machine enters the second type of area; An adjustment module is used to adjust the seeding machine's seed reseeding in the second type of area according to the deviation.

2. The terminal platform according to claim 1, characterized in that: Before the soil moisture evolution analysis module performs dynamic evolution analysis on the farmland geographic image according to the soil moisture of the farmland, it includes: Thermal infrared visual recognition is performed on the entire farmland at several time points to obtain the surface temperature and soil moisture content of the entire farmland at several time points; wherein the several time points at least include daytime time points and evening time points on several dates; The surface temperature and soil moisture content at each time point are inverted and calculated to obtain the soil moisture conditions of the farmland at each time point.

3. The terminal platform according to claim 2, characterized in that: The soil moisture evolution analysis module dynamically analyzes the farmland geographic image according to the soil moisture of the farmland, and calculates the soil moisture characteristics of each sowing area, specifically: Mapping the soil moisture condition of the farmland at each time point to a farmland geographic image divided with sowing area boundaries, and performing soil moisture condition edge sharpening processing on all sowing areas in the farmland geographic image; A dynamic time evolution analysis of the soil moisture conditions of each sowing area in the farmland geographic image at all time points is performed, and the soil moisture characteristics of each sowing area in the corresponding time interval at all time points are calculated; wherein the soil moisture characteristics include the soil moisture fluctuation changes of each sowing area in the time interval.

4. The terminal platform according to claim 1, characterized in that: The area calibration module visually identifies the geomorphic elements of the second type of area, specifically: According to the boundaries of the second type of areas, the remote sensing images of the global scope of the farmland are identified to obtain the geomorphic elements of the second type of areas; wherein the geomorphic elements include the surface contour elements of the second type of areas.

5. The terminal platform according to claim 1, characterized in that: The image comparison module generates geographical distribution images of all first-category areas and second-category areas, and compares them with the visual perception images generated by the planter to determine whether the planter enters the first-category area or the second-category area, specifically: Generate a geographical distribution image of all the first-category areas and the second-category areas according to the geographical boundary positioning data of the first-category areas and the second-category areas; wherein the geographical distribution image identifies the respective boundaries of the first-category areas and the second-category areas; The visual perception GIS image generated locally by the seeder is compared with the geographic distribution image in the same spatial coordinate system to determine the relative position relationship between the seeder and the boundaries of the first and second types of areas during operation, thereby determining whether the seeder enters the first or second type of area.

6. The terminal platform according to claim 1, characterized in that: The track deviation identification module is used to visually identify the deviation between the actual sowing track of the sowing machine and the effective sowing track when the sowing machine enters the second type of area, specifically: When the seeder enters the second type of area during operation, collecting images of seed clusters that have been placed in the second type of area; Performing background separation and seed contour recognition on the seed cluster image to determine the spatial distribution positions of the germination points of all the seeds that have been placed; Performing continuous fitting processing on the spatial distribution positions of the germination points of the seeds to obtain an actual sowing trajectory; The actual sowing trajectory is spatially compared with the effective sowing trajectory to determine the deviation between the two; wherein the deviation includes a distance deviation and an azimuth deviation between the actual sowing trajectory and the effective sowing trajectory.

7. The terminal platform according to claim 1, characterized in that: The adjustment module is used to adjust the seeding machine to re-sow seeds in the second type of area according to the deviation, specifically: According to the distance deviation and azimuth deviation between the actual sowing trajectory and the effective sowing trajectory included in the deviation, estimating the seed placement blank range between the actual sowing trajectory and the effective sowing trajectory; The seed supplementation density of the seeder in the seed placement blank range is adjusted according to the seed placement blank range and the soil moisture conditions along the effective sowing trajectory.

Images