Crop population canopy three-dimensional point cloud completion reconstruction method and system

Abstract

The invention provides a crop population canopy three-dimensional point cloud completion reconstruction method and system. The method comprises the following steps: obtaining to-be-reconstructed canopy population three-dimensional point cloud data and seedling stage crop population data corresponding to a target crop population; determining center coordinate information and azimuth angle information corresponding to each seedling-stage crop plant in the seedling-stage crop population data; according to the center coordinate information and the azimuth angle information, clipping the to-be-reconstructed canopy group three-dimensional point cloud data, and obtaining target point cloud regions corresponding to crop plants in each growth period in the to-be-reconstructed canopy group three-dimensional point cloud data; and performing interpolation processing on the point cloud plant data in the current target point cloud area based on the alternative plant point cloud data corresponding to the current target point cloud area to obtain a canopy three-dimensional point cloud completion reconstruction result. According to the invention, a more accurate three-dimensional point cloud completion reconstruction result can be obtained.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and particularly to a method and system for three-dimensional point cloud completion and reconstruction of a crop population canopy. Background Art

[0002] The crop population structure is an organizational system for fulfilling crop production functions, and its morphological structure characteristics have always been the most basic way for crop science researchers to understand, analyze, and evaluate crops.

[0003] In the field environment, there are already relatively mature methods and platforms such as using unmanned aerial vehicles (UAVs), orbital phenotyping platforms, backpack radars, and ground-based radar scans to obtain phenotypic data related to crop canopy structures. However, there are many missing data in these collected data. Especially when using UAVs and orbital phenotyping platforms, only the data of the upper part of the canopy can be obtained, and there are a large number of missing data in the middle and lower parts of the canopy. Currently, there is no method for point cloud data completion and reconstruction for crop canopies, and there are only a few point cloud completion methods for crop leaves, which cannot solve the problem of crop canopy point cloud completion.

[0004] Therefore, there is an urgent need for a method and system for three-dimensional point cloud completion and reconstruction of a crop population canopy to solve the above problems. Summary of the Invention

[0005] Aiming at the problems existing in the prior art, the present invention provides a method and system for three-dimensional point cloud completion and reconstruction of a crop population canopy.

[0006] The present invention provides a method for three-dimensional point cloud completion and reconstruction of a crop population canopy, including: Obtaining three-dimensional point cloud data of the canopy population to be reconstructed corresponding to the target crop population and seedling-stage crop population data, where the three-dimensional point cloud data of the canopy to be reconstructed is the three-dimensional point cloud data of the canopy population of the target crop population at other growth stages except the seedling stage; the seedling-stage crop population data is the image of the seedling-stage crop population or the three-dimensional point cloud data of the seedling-stage crop population; Determining the central coordinate information and azimuth angle information corresponding to each seedling-stage crop plant in the seedling-stage crop population data; According to the central coordinate information and the azimuth angle information, cropping the three-dimensional point cloud data of the canopy population to be reconstructed to obtain the target point cloud regions corresponding to the crop plants at each growth stage in the three-dimensional point cloud data of the canopy population to be reconstructed; Interpolate the point cloud plant data within the current target point cloud region based on the alternative plant point cloud data corresponding to the current target point cloud region. After determining that the interpolation process of the point cloud plant data within all the target point cloud regions in the to-be-reconstructed canopy population three-dimensional point cloud data is completed, obtain the canopy three-dimensional point cloud completion and reconstruction result corresponding to the target crop population; wherein, the alternative plant point cloud data is obtained based on the plant point cloud data corresponding to the crop plants in the growth period within the preset region in the to-be-reconstructed canopy population three-dimensional point cloud data.

[0007] According to a method for completing and reconstructing the three-dimensional point cloud of a crop population canopy provided by the present invention, the alternative plant point cloud data is constructed through the following steps: Based on the crop variety type corresponding to the target crop population, determine the crop type plots corresponding to each crop variety type in the to-be-reconstructed canopy population three-dimensional point cloud data; Take any non-edge region within each crop type plot as the preset region, and obtain the preset number of single-plant point cloud data within the preset region; Perform a first preprocessing on the single-plant point cloud data to obtain the preprocessed single-plant point cloud data, wherein the first preprocessing at least includes size reduction processing, point cloud direction correction processing, and noise point removal processing; According to the preprocessed single-plant point cloud data and the keyword information corresponding to the preprocessed single-plant point cloud data, construct the alternative plant point cloud data, wherein the keyword information at least includes the crop variety, planting density, growth period information, number of leaves, and plant height of the crop plants.

[0008] According to a method for completing and reconstructing the three-dimensional point cloud of a crop population canopy provided by the present invention, the determination of the central coordinate information and azimuth information corresponding to each seedling-stage crop plant in the seedling-stage crop population data includes: Based on the ground projection density fast segmentation algorithm, obtain the seedling-stage crop plant point cloud data corresponding to each seedling-stage crop plant in the seedling-stage crop population point cloud data; or, based on an image detection model, obtain the seedling-stage crop plant point cloud data corresponding to each seedling-stage crop plant in the seedling-stage crop population image According to the central points of the seedling-stage crop plants in the seedling-stage crop plant point cloud data, obtain the central coordinate information corresponding to each seedling-stage crop plant in the seedling-stage crop population data; Based on the principal component analysis algorithm, obtain the main direction of each seedling-stage crop plant in the seedling-stage crop plant point cloud data, and determine the azimuth information according to the angle between the main direction of the plant and the horizontal axis in the two-dimensional plane.

[0009] According to a method for three-dimensional point cloud completion and reconstruction of a crop population canopy provided by the present invention, before cropping the three-dimensional point cloud data of the to-be-reconstructed canopy population according to the central coordinate information and the azimuth information, the method further includes: When it is determined that the three-dimensional point cloud data of the to-be-reconstructed canopy population is obtained by an orbital phenotyping platform, based on the rotation transformation matrix, register and align the three-dimensional point cloud data of the to-be-reconstructed canopy population with the point cloud data of each seedling crop plant in the seedling crop population data; When it is determined that the three-dimensional point cloud data of the to-be-reconstructed canopy population is obtained by a drone platform, based on a preset reference object, register and align the three-dimensional point cloud data of the to-be-reconstructed canopy population with the point cloud data of each seedling crop plant in the seedling crop population data; According to the registration and alignment result, use the central coordinate information and the azimuth information corresponding to each seedling crop plant in the seedling crop population data as the central coordinate information and the azimuth information corresponding to each crop plant at each growth stage in the three-dimensional point cloud data of the to-be-reconstructed canopy population; The cropping of the three-dimensional point cloud data of the to-be-reconstructed canopy population according to the central coordinate information and the azimuth information includes: Based on the central coordinate information and the azimuth information corresponding to each crop plant at each growth stage in the three-dimensional point cloud data of the to-be-reconstructed canopy population, construct a cropping rectangular area for each crop plant at each growth stage; According to the cropping rectangular area, crop the three-dimensional point cloud data of the to-be-reconstructed canopy population to obtain the target point cloud area corresponding to each crop plant at each growth stage in the three-dimensional point cloud data of the to-be-reconstructed canopy population.

[0010] According to a method for three-dimensional point cloud completion and reconstruction of a crop population canopy provided by the present invention, the interpolation processing of the point cloud plant data in the current target point cloud area based on the alternative plant point cloud data corresponding to the current target point cloud area includes: Based on the crop variety, the planting density, the growth stage information, the number of leaves, and the plant height of the crop plants at the growth stage in the current target point cloud area, determine a preset number of to-be-determined alternative plant point cloud data from multiple alternative plant point cloud data; Based on the plant height of the crop plants at the growth stage in the current target point cloud area, perform scaling processing on the to-be-determined alternative plant point cloud data to obtain the scaled to-be-determined alternative plant point cloud data; Based on the azimuth information of the seedling-stage crop plants corresponding to the crop plants in the current target point cloud region, rotate the to-be-determined alternative plant point cloud data after the scaling process to obtain the rotated to-be-determined alternative plant point cloud data; Calculate the Chamfer distance between each of the rotated to-be-determined alternative plant point cloud data and the point cloud plant data of the crop plants in the current target point cloud region, and use the rotated to-be-determined alternative plant point cloud data corresponding to the minimum Chamfer distance as the target alternative plant point cloud data; Based on the target alternative plant point cloud data, perform interpolation processing on the point cloud plant data of the crop plants in the current target point cloud region to obtain the interpolated point cloud plant data, so as to obtain the canopy three-dimensional point cloud completion and reconstruction result corresponding to the target crop population according to the interpolated point cloud plant data.

[0011] According to a method for completing and reconstructing the three-dimensional point cloud of a crop population canopy provided by the present invention, the method further includes: Perform a second preprocessing on the three-dimensional point cloud data of the canopy population to be reconstructed to obtain the preprocessed three-dimensional point cloud data of the canopy population to be reconstructed, wherein the second preprocessing includes coordinate system calibration processing, point cloud denoising processing, ground point removal processing, and crop type plot segmentation processing; The step of cropping the three-dimensional point cloud data of the canopy population to be reconstructed according to the central coordinate information and the azimuth information to obtain the target point cloud regions corresponding to the crop plants in each growth stage in the three-dimensional point cloud data of the canopy population to be reconstructed further includes: Crop the preprocessed three-dimensional point cloud data of the canopy population to be reconstructed according to the central coordinate information and the azimuth information to obtain the target point cloud regions corresponding to the crop plants in each growth stage in the preprocessed three-dimensional point cloud data of the canopy population to be reconstructed.

[0012] The present invention also provides a system for completing and reconstructing the three-dimensional point cloud of a crop population canopy, including: A point cloud data acquisition module, configured to acquire the three-dimensional point cloud data of the canopy population to be reconstructed and the seedling-stage crop population data corresponding to the target crop population, wherein the three-dimensional point cloud data of the canopy to be reconstructed is the three-dimensional point cloud data of the canopy population of the target crop population at other growth stages except the seedling stage; the seedling-stage crop population data is the seedling-stage crop population image or the seedling-stage crop population point cloud data; A seedling-stage plant positioning module, configured to determine the central coordinate information and the azimuth information corresponding to each seedling-stage crop plant in the seedling-stage crop population data; The point cloud data registration and segmentation module is used to crop the three-dimensional point cloud data of the to-be-reconstructed canopy population according to the central coordinate information and the azimuth information, so as to obtain the target point cloud regions corresponding to the crop plants at each growth stage in the three-dimensional point cloud data of the to-be-reconstructed canopy population; The point cloud completion and reconstruction module is used to perform interpolation processing on the point cloud plant data in the current target point cloud region based on the alternative plant point cloud data corresponding to the current target point cloud region. After determining that the interpolation processing of the point cloud plant data in all the target point cloud regions in the three-dimensional point cloud data of the to-be-reconstructed canopy population is completed, the three-dimensional point cloud completion and reconstruction result corresponding to the target crop population is obtained; wherein, the alternative plant point cloud data is obtained based on the plant point cloud data corresponding to the crop plants at the growth stage in the preset region in the three-dimensional point cloud data of the to-be-reconstructed canopy population.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the method for three-dimensional point cloud completion and reconstruction of the crop population canopy as described in any one of the above is implemented.

[0014] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for three-dimensional point cloud completion and reconstruction of the crop population canopy as described in any one of the above is implemented.

[0015] The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, the method for three-dimensional point cloud completion and reconstruction of the crop population canopy as described in any one of the above is implemented.

[0016] The method and system for three-dimensional point cloud completion and reconstruction of the crop population canopy provided by the present invention extract the central coordinates and azimuths of each plant from the seedling stage crop population data, and then use this information to crop the non-seedling stage canopy point cloud data to obtain the point cloud regions of the crop plants at each growth stage. Then, based on the plant point cloud data in the preset region as an alternative, interpolation processing is performed on the point cloud data in the current target point cloud region to obtain a more accurate three-dimensional point cloud completion and reconstruction result of the crop population canopy. Description of the Drawings

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1Schematic flow chart of the method for completing and reconstructing the three-dimensional point cloud of the crop population canopy provided by the present invention; Figure 2 Schematic structural diagram of the system for completing and reconstructing the three-dimensional point cloud of the crop population canopy provided by the present invention; Figure 3 Schematic structural diagram of the electronic device provided by the present invention. Specific embodiments

[0019] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0020] In the field environment, the technology of using an unmanned aerial vehicle (UAV) platform to obtain crop canopy structure phenotype data has been relatively mature. However, the UAV platform has problems such as low data resolution, operation relying on manual labor, being easily affected by weather, and limited ability to continuously obtain time-series data.

[0021] In the aspect of three-dimensional data acquisition of crop canopies, Light Laser Detection and Ranging (LiDAR), depth cameras, structured light, Time of Flight (ToF), etc. are the main sensors and technologies. Among them, LiDAR is outstanding due to its stability and high precision in field three-dimensional data acquisition, especially having more advantages in high-resolution data acquisition. In addition, the data quality is affected by the distance between the sensor and the crop. Due to the relatively high flight altitude of the UAV and being affected by high-altitude airflows, there are limitations in data acquisition. In contrast, the orbital phenotype platform is closer to the crop canopy, can more stably obtain high-resolution data, and can flexibly adjust the sensor height according to the growth height of the crop. Ground-based LiDAR or backpack radar can also obtain the three-dimensional point cloud of the crop canopy, but there are requirements for planting density and plot space.

[0022] Currently, although there are various ways to obtain the three-dimensional point cloud data of crop canopies, the problem of data missing is common, especially the data missing in the middle and lower parts of the canopy is serious. In addition, the existing point cloud completion methods mainly target crop leaves and cannot effectively solve the problem of completing the point cloud of the crop canopy.

[0023] Figure 1 Schematic flow chart of the method for completing and reconstructing the three-dimensional point cloud of the crop population canopy provided by the present invention. As Figure 1 shown, the present invention provides a method for completing and reconstructing the three-dimensional point cloud of the crop population canopy, including: Step 101: Obtain the three-dimensional point cloud data of the canopy population to be reconstructed corresponding to the target crop population and the seedling-stage crop population data. Among them, the three-dimensional point cloud data of the canopy to be reconstructed is the three-dimensional point cloud data of the canopy population of the target crop population at other growth stages except the seedling stage; the seedling-stage crop population data is the seedling-stage crop population image or the seedling-stage crop population point cloud data.

[0024] In the present invention, in crop phenotype research or precision agriculture applications, in order to comprehensively understand the growth status and canopy structure of the target crop population, it is necessary to obtain the three-dimensional point cloud data of crops at different growth stages. Among them, the three-dimensional point cloud data of the canopy population to be reconstructed is the three-dimensional point cloud data of the canopy population of the target crop population at other growth stages (such as the growth stage, maturity stage, etc.) except the seedling stage, which is obtained by using an unmanned aerial vehicle or an orbital phenotype platform. These data contain detailed three-dimensional information of the crop canopy, such as height, density, and shape, etc., and are important bases for analyzing crop growth status, evaluating yield potential, and conducting pest and disease control, etc.

[0025] At the same time, the present invention also obtains the seedling-stage crop population data, which is usually taken by a ground camera, a camera mounted on an unmanned aerial vehicle, or other image acquisition devices at the seedling stage of the crop, and contains information such as the growth status, leaf distribution, and plant spacing of the seedling-stage crops. These image data can be combined with the three-dimensional point cloud data of the canopy to be reconstructed to provide reference and basis for the subsequent completion and reconstruction of the three-dimensional point cloud of the canopy. In the present invention, the seedling-stage crop population data can be the image data of the top of the seedling-stage crop canopy (i.e., the seedling-stage crop population image) or the three-dimensional point cloud data (i.e., the seedling-stage crop population point cloud data), which is mainly used for the positioning of each plant in the crop population during the development period.

[0026] Step 102: Determine the central coordinate information and azimuth angle information corresponding to each seedling-stage crop plant in the seedling-stage crop population data.

[0027] In the present invention, based on the seedling-stage crop population data, the position of each seedling-stage crop in the image can be accurately identified and located, as well as its azimuth angle relative to a certain reference direction. Specifically, first, through image processing techniques, such as image segmentation and feature extraction, etc., each crop in the seedling-stage crop population image is separately identified. Then, for each identified seedling-stage crop, it is necessary to determine its central coordinate information in the image. The central coordinate usually refers to the geometric center of the crop on the image plane, which can be obtained by calculating the centroid of the crop contour or using other geometric methods. In addition, the present invention also needs to determine the azimuth angle information of each crop. The azimuth angle refers to the angle of the crop relative to a certain reference direction. In order to obtain this information, known reference objects in the image (such as landmarks, sun position, etc.) or by combining with the attitude information of the image acquisition device can be used for calculation.

[0028] Step 103: According to the center coordinate information and the azimuth information, crop the three-dimensional point cloud data of the canopy population to be reconstructed, and obtain the target point cloud regions corresponding to the crop plants at each growth stage in the three-dimensional point cloud data of the canopy population to be reconstructed.

[0029] In the present invention, based on the center coordinate information and the azimuth information of each seedling-stage crop plant in the obtained seedling-stage crop population data, the three-dimensional point cloud data of the canopy population to be reconstructed is cropped to obtain the target point cloud regions corresponding to the crop plants at each growth stage.

[0030] Specifically, in the present invention, the center coordinate information provides the precise position of each crop plant on the two-dimensional image plane. In three-dimensional space, although the point cloud data is discrete, through a certain mapping or registration method, the center coordinates on the two-dimensional image can be associated with the corresponding positions in the three-dimensional point cloud data to ensure the spatial consistency between the two. The azimuth information provides the angle of each crop plant relative to a certain reference direction. In the three-dimensional point cloud data, this angle can determine the growth direction of the crop in space. By combining the center coordinates and the azimuth, a spatial cropping box can be defined, and this spatial cropping box can be used to enclose the point cloud data of each crop plant in three-dimensional space.

[0031] Furthermore, use the spatial cropping box to crop the three-dimensional point cloud data of the canopy population to be reconstructed. The cropping process is to extract the point cloud subsets corresponding to each crop plant from the original point cloud data, that is, the target point cloud regions, which contain the canopy structure information of each crop plant at a specific growth stage.

[0032] Step 104: Based on the alternative plant point cloud data corresponding to the current target point cloud region, perform interpolation processing on the point cloud plant data within the current target point cloud region. After determining that the interpolation processing of all the point cloud plant data within the target point cloud regions in the three-dimensional point cloud data of the canopy population to be reconstructed is completed, obtain the completed reconstruction result of the three-dimensional point cloud of the target crop population; wherein, the alternative plant point cloud data is obtained based on the plant point cloud data corresponding to the crop plants at the growth stage within a preset region in the three-dimensional point cloud data of the canopy population to be reconstructed.

[0033] Through the above embodiments, after obtaining the target point cloud regions corresponding to the crop plants at each growth stage, further perform interpolation processing on the plant data within these target point cloud regions to complement the possibly missing or sparse point cloud data. This step is completed based on the alternative plant point cloud data corresponding to the current target point cloud region.

[0034] In the present invention, the alternative plant point cloud data is the point cloud data corresponding to the crop plants in the growth period selected from the three-dimensional point cloud data of the canopy population to be reconstructed within a preset area. The preset area can be determined according to experience, prior knowledge or statistical methods, and it contains the point cloud data of representative or typical crop plants, and these alternative data will be used as a reference or template for interpolation processing. Optionally, in the present invention, the alternative plant point cloud data is the point cloud data of the plants at non-edge positions within the planting area corresponding to the three-dimensional point cloud data of the canopy population to be reconstructed.

[0035] Further, the current target point cloud area is compared and analyzed with the alternative plant point cloud data. By comparing the characteristics such as the shape, structure, density, etc. of the two, the missing or sparse parts that may exist in the target point cloud area can be identified. Then, using an interpolation algorithm or model, according to the characteristics of the alternative plant point cloud data, the missing or sparse parts in the target point cloud area are filled or enhanced. The process of interpolation processing involves a variety of techniques and methods, such as linear interpolation, non-linear interpolation, and interpolation based on machine learning, etc. Through interpolation processing, the point cloud data within the target point cloud area can be made more complete, dense and accurate.

[0036] Finally, after determining that the interpolation processing of the point cloud plant data in all target point cloud areas within the three-dimensional point cloud data of the canopy population to be reconstructed is completed, the three-dimensional point cloud completion and reconstruction result corresponding to the target crop population is obtained. This result is a more complete, accurate and detailed three-dimensional point cloud model of the canopy, which can be used in various agricultural application and research fields such as subsequent crop growth analysis, yield prediction, pest and disease control, etc. For example, in one embodiment, based on the above three-dimensional point cloud completion and reconstruction method for the crop population canopy, the three-dimensional dense point cloud data of the crop population canopy obtained by completion and reconstruction can be used to calculate indexes such as canopy occupation volume (COV), canopy coverage (CC), and canopy vertical profile (CVP).

[0037] The three-dimensional point cloud completion and reconstruction method for the crop population canopy provided by the present invention extracts the central coordinates and azimuth angles of each plant from the seedling stage crop population data, and then uses this information to crop the non-seedling stage canopy point cloud data to obtain the point cloud areas of the crop plants at each growth stage. Then, based on the plant point cloud data within the preset area as an alternative, the point cloud data within the current target point cloud area is interpolated to obtain a more accurate three-dimensional point cloud completion and reconstruction result of the crop population canopy.

[0038] On the basis of the above embodiment, the alternative plant point cloud data is constructed through the following steps: Based on the crop variety types corresponding to the target crop population, determine the crop type plots corresponding to each of the crop variety types in the three-dimensional point cloud data of the canopy population to be reconstructed; Take any non-edge area within each of the crop type plots as the preset area, and obtain the point cloud data of a preset number of individual plants within the preset area; Perform a first preprocessing on the point cloud data of the individual plants to obtain the preprocessed point cloud data of the individual plants, where the first preprocessing at least includes size reduction processing, point cloud direction correction processing, and noise point removal processing; Based on the preprocessed point cloud data of the individual plants and the keyword information corresponding to the preprocessed point cloud data of the individual plants, construct the candidate plant point cloud data, where the keyword information at least includes the crop variety, planting density, growth period information, number of leaves, and plant height of the crop plants.

[0039] In the present invention, based on the crop varieties corresponding to the target crop population, the crop type plots corresponding to each of the crop variety types in the three-dimensional point cloud data of the canopy population to be reconstructed are determined. The crop type plot refers to a specific area divided according to the crop varieties in the three-dimensional point cloud data, and the crop plants within these areas belong to the same variety type. By identifying and analyzing the crop variety types, the present invention can orderly divide the three-dimensional point cloud data of the canopy population to be reconstructed into multiple crop type plots, providing convenience for subsequent processing and analysis.

[0040] Further, within each crop type plot, the present invention selects a non-edge area as the preset area. The non-edge area refers to an area far from the plot boundary and less affected by the boundary effect, and the growth conditions of the crop plants within these areas are relatively stable, and the quality of the point cloud data is relatively high. The present invention selects the non-edge area as the preset area to ensure that the obtained point cloud data of the individual plants is representative and reliable.

[0041] Within the preset area, obtain the point cloud data of a preset number of individual plants. The point cloud data of the individual plants refers to the point cloud data corresponding to a single crop plant, which contains detailed information such as the shape, structure, and density of the plant. By obtaining the point cloud data of a preset number of individual plants, sufficient data support can be provided for subsequent interpolation processing, feature extraction, and other tasks.

[0042] In the present invention, the three-dimensional point cloud data of the canopy population to be reconstructed can be obtained by performing three-dimensional point cloud reconstruction on the sequence images of the top of the target crop population canopy, or the three-dimensional point cloud of the top of the target crop population canopy can be directly obtained by lidar. Then, for the three-dimensional point cloud data of the canopy population to be reconstructed, multiple (such as 3) representative plants (such as plants in non-edge regions) of different crop type plots are selected and transplanted into flowerpots, and multi-view imaging, lidar, three-dimensional scanners, etc. are used to obtain the single-plant point cloud data corresponding to these plants. Further, size reduction is performed through calibration or scaling factors, and then principal component analysis (PCA) is used to correct the direction of the single-plant point cloud data to align the main axis of the plant with the Z-axis. Finally, the lowest point of the plant is translated to the origin, and noise points are removed to ensure the accuracy and consistency of the point cloud data.

[0043] In the present invention, after obtaining the single-plant point cloud data, it needs to be subjected to a first preprocessing. The first preprocessing at least includes size reduction processing, point cloud direction correction processing, and noise point removal processing. Among them, size reduction processing refers to restoring the point cloud data to the actual size ratio to ensure the accuracy of the data; point cloud direction correction processing refers to adjusting the direction of the point cloud data to make it consistent with the actual crop growth direction; noise point removal processing refers to removing noise points or abnormal points in the point cloud data to improve the quality of the data. Through the first preprocessing, the preprocessed single-plant point cloud data can be obtained, and these data are more accurate, reliable, and easy to process.

[0044] Finally, based on the preprocessed single-plant point cloud data and its corresponding keyword information, alternative plant point cloud data is constructed. In the present invention, the keyword information at least includes the crop variety, planting density, growth stage information, number of leaves, and plant height of the crop plant, etc. These keyword information describe the main characteristics and attributes of the single-plant point cloud data.

[0045] Based on the above embodiments, determining the central coordinate information and azimuth information corresponding to each seedling crop plant in the seedling crop population data includes: Based on the Ground Density Quickshift++ algorithm, obtaining the seedling crop plant point cloud data corresponding to each seedling crop plant in the seedling crop population point cloud data; or, based on an image detection model, obtaining the seedling crop plant point cloud data corresponding to each seedling crop plant in the seedling crop population image; According to the central point of the seedling crop plant in the seedling crop plant point cloud data, obtaining the central coordinate information corresponding to each seedling crop plant in the seedling crop population data; Based on the principal component analysis algorithm, obtain the main direction of each seedling crop plant corresponding to the point cloud data of the seedling crop plants, and determine the azimuth information according to the angle between the main direction of the plant and the horizontal axis in the two-dimensional plane.

[0046] In the present invention, through the ground projection density fast segmentation algorithm, the point cloud data of the seedling crop population is processed, so as to segment each individual seedling crop plant from the population and form independent plant point cloud data. For the parts that cannot be automatically segmented by the algorithm, a small amount of manual interaction can be used to complete. The ground projection density fast segmentation algorithm adopted in the present invention is an improved version of the Quickshift algorithm. Compared with the Quickshift algorithm, the Ground DensityQuickshift++ algorithm has lower sensitivity to input parameters while maintaining the same algorithm time complexity, improving the accuracy and robustness of segmentation.

[0047] For the seedling crop population image, the present invention uses an image detection model (such as Plant U-Net) for processing, and each individual seedling crop plant can be segmented from the seedling crop population image to form independent plant point cloud data.

[0048] After segmentation, the point cloud data corresponding to each seedling crop plant is obtained, and these data contain detailed information such as the three-dimensional shape and structure of the plant. In the plant point cloud data, a point representing the overall position of the plant, that is, the center point of the seedling crop plant, can be calculated or identified. This point is usually the geometric center or centroid of the plant in three-dimensional space. Further, projecting the center point of the plant onto a two-dimensional plane (such as the XY plane) can obtain the position coordinates of the plant in the two-dimensional space, that is, the center coordinate information.

[0049] In the present invention, based on the principal component analysis algorithm, the main growth direction or main direction of the plant is determined. Specifically, by processing the plant point cloud data with the principal component analysis algorithm, a vector representing the main direction of the plant can be obtained, and this vector indicates the main growth or extension direction of the plant in three-dimensional space. On the two-dimensional plane, a direction is selected as the horizontal axis (such as the X axis) for reference and comparison. Then, by calculating the angle between the main direction of the plant and the horizontal axis in the two-dimensional plane, the azimuth information of the plant can be obtained, and the azimuth information describes the direction of the plant relative to the horizontal axis.

[0050] On the basis of the above embodiments, before cropping the three-dimensional point cloud data of the to-be-reconstructed canopy population according to the center coordinate information and the azimuth information, the method further includes: When it is determined that the three-dimensional point cloud data of the to-be-reconstructed canopy population is obtained by an orbital phenotyping platform, based on the rotation transformation matrix, register and align the three-dimensional point cloud data of the to-be-reconstructed canopy population with the point cloud data of each seedling crop plant in the seedling crop population data; When it is determined that the three-dimensional point cloud data of the to-be-reconstructed canopy population is obtained by a drone platform, based on a preset reference object, register and align the three-dimensional point cloud data of the to-be-reconstructed canopy population with the point cloud data of each seedling crop plant in the seedling crop population data; According to the registration and alignment result, use the central coordinate information and the azimuth information corresponding to each seedling crop plant in the seedling crop population data as the central coordinate information and the azimuth information corresponding to each crop plant at each growth stage in the three-dimensional point cloud data of the to-be-reconstructed canopy population; The step of cropping the three-dimensional point cloud data of the to-be-reconstructed canopy population according to the central coordinate information and the azimuth information includes: Based on the central coordinate information and the azimuth information corresponding to each crop plant at each growth stage in the three-dimensional point cloud data of the to-be-reconstructed canopy population, construct a cropping rectangular area for each crop plant at each growth stage; According to the cropping rectangular area, crop the three-dimensional point cloud data of the to-be-reconstructed canopy population to obtain the target point cloud area corresponding to each crop plant at each growth stage in the three-dimensional point cloud data of the to-be-reconstructed canopy population.

[0051] In the present invention, when it is determined that the three-dimensional point cloud data of the to-be-reconstructed canopy population is obtained by an orbital phenotyping platform, since the obtained crop population time-series point cloud data is in a unified coordinate system, the point cloud data at different times can be aligned through a unified rotation transformation matrix, that is, register the three-dimensional point cloud data of the to-be-reconstructed canopy population with the point cloud data corresponding to each seedling crop plant in the seedling crop population data to make them aligned in the same coordinate system.

[0052] When it is determined that the three-dimensional point cloud data of the to-be-reconstructed canopy population is obtained by a drone platform, since the point cloud data obtained by the drone platform may have deviations from the seedling data due to flight altitude, angle, etc., it is necessary to use a preset calibration rod or marker as a reference for registering and aligning the point cloud data. By identifying and using these reference objects, the three-dimensional point cloud data of the to-be-reconstructed canopy population can be aligned with the point cloud data in the seedling crop population data.

[0053] After completing the registration and alignment of the point cloud data, the central coordinate information and azimuth angle information of each seedling-stage crop plant in the seedling-stage crop population data are used as the central coordinate information and azimuth angle information of each crop plant at different growth stages in the 3D point cloud data of the canopy population to be reconstructed. This is because during the growth process of the crop, its central position and azimuth angle remain unchanged or change slightly. Therefore, the information of the seedling stage can be used for the processing of the point cloud data in subsequent growth stages.

[0054] Furthermore, based on the central coordinate information and azimuth angle information of each crop plant at different growth stages in the 3D point cloud data of the canopy population to be reconstructed, a cropping rectangular area for each crop plant at different growth stages is constructed. This cropping rectangular area is determined based on the center point of the crop plant and its azimuth angle information, aiming to roughly cover the growth range of the crop plant. Then, according to the constructed cropping rectangular area, the 3D point cloud data of the canopy population to be reconstructed is cropped. During the cropping process, only the point cloud data within the cropping rectangular area (i.e., the target point cloud area) is retained, and these data are considered to belong to the corresponding crop plant.

[0055] Based on the above embodiments, the interpolation processing of the point cloud plant data within the current target point cloud area based on the alternative plant point cloud data corresponding to the current target point cloud area includes: Based on the crop variety, planting density, growth stage information, number of leaves, and plant height of the crop plants at different growth stages within the current target point cloud area, a preset number of undetermined alternative plant point cloud data are determined from multiple alternative plant point cloud data; Based on the plant height of the crop plants at different growth stages within the current target point cloud area, the undetermined alternative plant point cloud data are scaled to obtain the scaled undetermined alternative plant point cloud data; Based on the azimuth angle information of the corresponding seedling-stage crop plants of the crop plants at different growth stages within the current target point cloud area, the scaled undetermined alternative plant point cloud data are rotated to obtain the rotated undetermined alternative plant point cloud data; Calculate the Chamfer distance between each rotated undetermined alternative plant point cloud data and the point cloud plant data of the crop plants at different growth stages within the current target point cloud area, and use the rotated undetermined alternative plant point cloud data corresponding to the minimum Chamfer distance as the target alternative plant point cloud data; Based on the target alternative plant point cloud data, the point cloud plant data of the crop plants at different growth stages within the current target point cloud area are interpolated to obtain the interpolated point cloud plant data, so as to obtain the complementary reconstruction result of the 3D point cloud of the canopy corresponding to the target crop population according to the interpolated point cloud plant data.

[0056] In the present invention, based on the crop variety, planting density, growth stage information, number of leaves, and plant height of the crop plants in the current target point cloud region, a preset number of undetermined alternative plant point cloud data are selected from multiple alternative plant point cloud data. The purpose of this step is to find several alternative plants in the database (i.e., the database of alternative plant point cloud data) that are most similar to the current plant in multiple characteristics, including crop variety, planting density (reflecting the density or planting density between plants), growth stage (reflecting the growth stage of the plants), number of leaves (affecting the morphology and canopy structure of the plants), and plant height, etc. By matching these characteristics, the range of alternative plants can be narrowed, and the accuracy of subsequent interpolation reconstruction can be improved.

[0057] Furthermore, based on the plant height of the crop plants in the current target point cloud region, the undetermined alternative plant point cloud data are scaled so that the scaled plant height is the same as the current plant height. And based on the azimuth information of the seedling-stage crop plants corresponding to the crop plants in the current target point cloud region, the scaled undetermined alternative plant point cloud data are rotated around the z-axis (i.e., the stem direction) so that the azimuth of the rotated plants is the same as the azimuth of the current plant at the seedling stage. The azimuth is an important feature of the plant growth direction. Through the rotation process, it can be ensured that the alternative plants match the current plant in the growth direction, further improving the accuracy of point cloud interpolation.

[0058] Furthermore, the Chamfer distance between each rotated undetermined alternative plant point cloud data and the point cloud plant data of the crop plants in the current target point cloud region is calculated. In the present invention, the Chamfer distance is calculated by an improved Chamfer algorithm, and the formula of the improved Chamfer algorithm is as follows: ; where is the point cloud of the plant corresponding block obtained after population segmentation, that is, the point cloud plant data of the crop plants in the current target point cloud region; is the plant template point cloud matched in the database, that is, the rotated undetermined alternative plant point cloud data; is the point in the point cloud ; is the point in the point cloud ; is the point in the point cloud and

[0059] In the present invention, the rotated candidate plant point cloud data corresponding to the minimum Chamfer distance is selected as the target candidate plant point cloud data, so as to find a plant in the candidate plants that is most similar to the current plant in terms of point cloud morphology as a template for subsequent interpolation. Further, interpolation processing is performed on the point cloud plant data of the crop plants in the target point cloud region. The target candidate plant point cloud data is translated to the corresponding position according to the calculated central point coordinates, and the morphology and structure information of the target candidate plant point cloud are used to supplement and complete the missing or incomplete parts in the current plant point cloud. Through the interpolation processing, more complete and accurate plant point cloud data can be obtained. The interpolated plant point cloud data is combined together to form the three-dimensional point cloud of the canopy of the entire crop population, and more complete and real morphological and structural information of the crop population canopy can be obtained.

[0060] Based on the above embodiments, the method further includes: Performing a second preprocessing on the three-dimensional point cloud data of the canopy population to be reconstructed to obtain the preprocessed three-dimensional point cloud data of the canopy population to be reconstructed, wherein the second preprocessing includes coordinate system calibration processing, point cloud denoising processing, ground point removal processing, and crop type plot segmentation processing; The step of cropping the three-dimensional point cloud data of the canopy population to be reconstructed according to the central coordinate information and the azimuth information to obtain the target point cloud regions corresponding to the crop plants at each growth stage in the three-dimensional point cloud data of the canopy population to be reconstructed further includes: Cropping the preprocessed three-dimensional point cloud data of the canopy population to be reconstructed according to the central coordinate information and the azimuth information to obtain the target point cloud regions corresponding to the crop plants at each growth stage in the preprocessed three-dimensional point cloud data of the canopy population to be reconstructed.

[0061] In the present invention, the second preprocessing includes coordinate system calibration processing, point cloud denoising processing, ground point removal processing, and crop type plot segmentation processing. The specific processing process is as follows: Coordinate system calibration processing: Since the lidar mounted on the imaging unit is placed vertically downward during data acquisition, it is necessary to adjust the point cloud to be vertically upward for subsequent analysis. In the present invention, by calculating the angle between the horizontal X direction of the point cloud and the crop row direction, a rotation transformation is performed on the point cloud so that the row direction is parallel to the X axis.

[0062] Point cloud denoising processing: The original point cloud obtained by using a drone or a phenotyping platform has noise points outside the crop population, including road surfaces and the running trajectories of sensors. Since the coordinate system of the population point cloud has been calibrated, the point cloud data required for analysis can be cut out according to the coordinate range of the crop population.

[0063] Ground point removal processing: The cloth simulation filter (CSF) algorithm is used to detect and remove the ground points from the point cloud in the target area. The CSF algorithm can effectively separate the ground points from the non-ground points in the point cloud, especially suitable for distinguishing non-ground points such as buildings and vegetation on the terrain from the point cloud data.

[0064] Crop type plot segmentation processing: The point cloud of the crop rows is segmented according to the crop planting plan. For plots with obvious gaps between them, density clustering can be used for segmentation; for those without obvious intervals, calibration poles can be set before data acquisition, and the coordinates of the calibration poles are used to determine the density range of different areas to cut out each crop type plot.

[0065] Furthermore, the preprocessed three-dimensional point cloud data of the canopy population to be reconstructed is cropped. During the cropping process, the central coordinate information is used to determine the approximate position of each crop plant. Then, combined with the azimuth information, the growth range and morphology of each plant can be more accurately defined, and the point cloud area corresponding to each crop plant at each growth stage can be accurately cropped from the three-dimensional point cloud data of the canopy population to be reconstructed.

[0066] The crop population canopy three-dimensional point cloud completion and reconstruction system provided by the present invention will be described below. The crop population canopy three-dimensional point cloud completion and reconstruction system described below can be mutually referred to with the crop population canopy three-dimensional point cloud completion and reconstruction method described above.

[0067] Figure 2 is a schematic structural diagram of the crop population canopy three-dimensional point cloud completion and reconstruction system provided by the present invention, as Figure 2As shown in the figure, the present invention provides a three-dimensional point cloud completion and reconstruction system for a crop population canopy, including a point cloud data acquisition module 201, a seedling-stage plant positioning module 202, a point cloud data registration and segmentation module 203, and a point cloud completion and reconstruction module 204. Among them, the point cloud data acquisition module 201 is used to obtain the three-dimensional point cloud data of the canopy population to be reconstructed corresponding to the target crop population and the data of the crop population at the seedling stage. Among them, the three-dimensional point cloud data of the canopy to be reconstructed is the three-dimensional point cloud data of the canopy population of the target crop population at other growth stages except the seedling stage of the crop; the data of the crop population at the seedling stage is the image of the crop population at the seedling stage or the three-dimensional point cloud data of the crop population at the seedling stage; the seedling-stage plant positioning module 202 is used to determine the central coordinate information and azimuth information corresponding to each seedling-stage crop plant in the data of the crop population at the seedling stage; the point cloud data registration and segmentation module 203 is used to crop the three-dimensional point cloud data of the canopy population to be reconstructed according to the central coordinate information and the azimuth information, and obtain the target point cloud region corresponding to each growth-stage crop plant in the three-dimensional point cloud data of the canopy population to be reconstructed; the point cloud completion and reconstruction module 204 is used to perform interpolation processing on the point cloud plant data in the current target point cloud region based on the alternative plant point cloud data corresponding to the current target point cloud region. After determining that the interpolation processing of the point cloud plant data in all the target point cloud regions in the three-dimensional point cloud data of the canopy population to be reconstructed is completed, the three-dimensional point cloud completion and reconstruction result corresponding to the target crop population is obtained; among them, the alternative plant point cloud data is obtained based on the plant point cloud data corresponding to the growth-stage crop plants in the preset region in the three-dimensional point cloud data of the canopy population to be reconstructed.

[0068] The three-dimensional point cloud completion and reconstruction system for a crop population canopy provided by the present invention extracts the central coordinates and azimuth angles of each plant from the data of the crop population at the seedling stage, and then uses this information to crop the non-seedling-stage canopy point cloud data to obtain the point cloud regions of the crop plants at each growth stage. Then, based on the plant point cloud data in the preset region as an alternative, interpolation processing is performed on the point cloud data in the current target point cloud region to obtain a more accurate three-dimensional point cloud completion and reconstruction result of the crop population canopy.

[0069] The system provided in the embodiments of the present invention is used to execute the above method embodiments. For the specific process and detailed content, please refer to the above embodiments and will not be elaborated here.

[0070] Figure 3 It is a schematic structural diagram of the electronic device provided by the present invention, as Figure 3As shown in the figure, the electronic device may include: a processor 301, a communications interface 302, a memory 303, and a communication bus 304. Among them, the processor 301, the communications interface 302, and the memory 303 complete communication with each other through the communication bus 304. The processor 301 may call logic instructions in the memory 303 to execute a method for three-dimensional point cloud completion and reconstruction of a crop population canopy. The method includes: obtaining three-dimensional point cloud data of a canopy population to be reconstructed corresponding to a target crop population and three-dimensional point cloud data of a seedling crop population. Among them, the three-dimensional point cloud data of the canopy population to be reconstructed is the three-dimensional point cloud data of the canopy population of the target crop population at other growth stages except the seedling stage of the crop; the three-dimensional point cloud data of the seedling crop population is an image of the seedling crop population or three-dimensional point cloud data of the seedling crop population; determining the central coordinate information and azimuth angle information corresponding to each seedling crop plant in the three-dimensional point cloud data of the seedling crop population; according to the central coordinate information and the azimuth angle information, cropping the three-dimensional point cloud data of the canopy population to be reconstructed to obtain a target point cloud region corresponding to each crop plant at each growth stage in the three-dimensional point cloud data of the canopy population to be reconstructed; based on the alternative plant point cloud data corresponding to the current target point cloud region, performing interpolation processing on the point cloud plant data in the current target point cloud region. After determining that the interpolation processing of the point cloud plant data in all the target point cloud regions in the three-dimensional point cloud data of the canopy population to be reconstructed is completed, obtaining a three-dimensional point cloud completion and reconstruction result of the canopy corresponding to the target crop population; among them, the alternative plant point cloud data is obtained based on the plant point cloud data corresponding to the crop plants at the growth stage in a preset region in the three-dimensional point cloud data of the canopy population to be reconstructed.

[0071] In addition, when the logic instructions in the above-mentioned memory 303 are implemented in the form of a software functional unit and sold or used as an independent product, they may be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, may be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc that can store program codes.

[0072] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, and when the program instructions are executed by a computer, the computer can execute the crop population canopy three-dimensional point cloud completion and reconstruction method provided by each of the above methods. The method includes: obtaining the three-dimensional point cloud data of the canopy population to be reconstructed corresponding to the target crop population and the three-dimensional point cloud data of the seedling-stage crop population, wherein the three-dimensional point cloud data of the canopy population to be reconstructed is the three-dimensional point cloud data of the canopy population of the target crop population at other growth stages except the seedling stage; the three-dimensional point cloud data of the seedling-stage crop population is the image of the seedling-stage crop population or the three-dimensional point cloud data of the seedling-stage crop population; determining the central coordinate information and azimuth information corresponding to each seedling-stage crop plant in the three-dimensional point cloud data of the seedling-stage crop population; according to the central coordinate information and the azimuth information, cropping the three-dimensional point cloud data of the canopy population to be reconstructed to obtain the target point cloud region corresponding to each crop plant at each growth stage in the three-dimensional point cloud data of the canopy population to be reconstructed; based on the alternative plant point cloud data corresponding to the current target point cloud region, performing interpolation processing on the point cloud plant data in the current target point cloud region, and after determining that the interpolation processing of all the point cloud plant data in the target point cloud regions in the three-dimensional point cloud data of the canopy population to be reconstructed is completed, obtaining the result of the three-dimensional point cloud completion and reconstruction of the canopy corresponding to the target crop population; wherein the alternative plant point cloud data is obtained based on the plant point cloud data corresponding to the crop plants at the growth stage in the preset region in the three-dimensional point cloud data of the canopy population to be reconstructed.

[0073] In another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is configured to execute the three-dimensional point cloud completion and reconstruction method for a crop population canopy provided in the above-mentioned various embodiments. The method includes: obtaining the three-dimensional point cloud data of the canopy population to be reconstructed corresponding to the target crop population and the three-dimensional point cloud data of the seedling-stage crop population. Wherein, the three-dimensional point cloud data of the canopy population to be reconstructed is the three-dimensional point cloud data of the canopy population of the target crop population at other growth stages except the seedling stage; the three-dimensional point cloud data of the seedling-stage crop population is the image of the seedling-stage crop population or the three-dimensional point cloud data of the seedling-stage crop population; determining the central coordinate information and azimuth angle information corresponding to each seedling-stage crop plant in the three-dimensional point cloud data of the seedling-stage crop population; according to the central coordinate information and the azimuth angle information, cropping the three-dimensional point cloud data of the canopy population to be reconstructed to obtain the target point cloud regions corresponding to the crop plants at each growth stage in the three-dimensional point cloud data of the canopy population to be reconstructed; based on the alternative plant point cloud data corresponding to the current target point cloud region, performing interpolation processing on the point cloud plant data in the current target point cloud region. After determining that the interpolation processing of the point cloud plant data in all the target point cloud regions in the three-dimensional point cloud data of the canopy population to be reconstructed is completed, obtaining the three-dimensional point cloud completion and reconstruction result corresponding to the target crop population; wherein, the alternative plant point cloud data is obtained based on the plant point cloud data corresponding to the crop plants at the growth stage in the preset region in the three-dimensional point cloud data of the canopy population to be reconstructed.

[0074] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative effort.

[0075] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, also by hardware. Based on such an understanding, the above technical solution, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0076] 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 them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and 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 three-dimensional point cloud completion and reconstruction method for a crop population canopy, characterized in that, Including: Obtaining the three-dimensional point cloud data of the canopy population to be reconstructed corresponding to the target crop population and the seedling-stage crop population data, where the three-dimensional point cloud data of the canopy to be reconstructed is the three-dimensional point cloud data of the canopy population of the target crop population at other growth stages except the seedling stage; the seedling-stage crop population data is the seedling-stage crop population image or the seedling-stage crop population point cloud data; Determining the central coordinate information and azimuth information corresponding to each seedling-stage crop plant in the seedling-stage crop population data; According to the central coordinate information and the azimuth information, cropping the three-dimensional point cloud data of the canopy population to be reconstructed to obtain the target point cloud regions corresponding to the crop plants at each growth stage in the three-dimensional point cloud data of the canopy population to be reconstructed; Based on the alternative plant point cloud data corresponding to the current target point cloud region, performing interpolation processing on the point cloud plant data in the current target point cloud region. After determining that the interpolation processing of all the point cloud plant data in the target point cloud regions in the three-dimensional point cloud data of the canopy population to be reconstructed is completed, obtaining the completed reconstruction result of the three-dimensional point cloud of the canopy corresponding to the target crop population; where the alternative plant point cloud data is obtained based on the plant point cloud data corresponding to the crop plants at the growth stage in the preset region in the three-dimensional point cloud data of the canopy population to be reconstructed.

2. The three-dimensional point cloud completion and reconstruction method for crop population canopy according to claim 1, wherein The alternative plant point cloud data is constructed through the following steps: Based on the crop variety type corresponding to the target crop population, determining the crop type plots corresponding to each crop variety type in the three-dimensional point cloud data of the canopy population to be reconstructed; Taking any non-edge region in each crop type plot as the preset region, and obtaining the preset number of single-plant point cloud data in the preset region; Performing a first preprocessing on the single-plant point cloud data to obtain the preprocessed single-plant point cloud data, where the first preprocessing at least includes size reduction processing, point cloud direction correction processing, and noise point removal processing; According to the preprocessed single-plant point cloud data and the keyword information corresponding to the preprocessed single-plant point cloud data, constructing the alternative plant point cloud data, where the keyword information at least includes the crop variety, planting density, growth stage information, number of leaves, and plant height of the crop plant.

3. The three-dimensional point cloud completion and reconstruction method for crop population canopy according to claim 1, characterized in that The determining the central coordinate information and azimuth information corresponding to each seedling-stage crop plant in the seedling-stage crop population data includes: Based on the ground projection density fast segmentation algorithm, obtaining the seedling-stage crop plant point cloud data corresponding to each seedling-stage crop plant in the seedling-stage crop population point cloud data; or, based on the image detection model, obtaining the seedling-stage crop plant point cloud data corresponding to each seedling-stage crop plant in the seedling-stage crop population image; According to the central points of the seedling-stage crop plants in the seedling-stage crop plant point cloud data, obtaining the central coordinate information corresponding to each seedling-stage crop plant in the seedling-stage crop population data; Based on the principal component analysis algorithm, obtaining the main direction of each seedling-stage crop plant in the seedling-stage crop plant point cloud data, and determining the azimuth information according to the angle between the main direction of the plant and the horizontal axis in the two-dimensional plane.

4. The method for three-dimensional point cloud completion and reconstruction of a crop population canopy according to claim 3, characterized in that, Before cropping the three-dimensional point cloud data of the to-be-reconstructed canopy population according to the central coordinate information and the azimuth information, the method further includes: When it is determined that the three-dimensional point cloud data of the to-be-reconstructed canopy population is obtained through an orbital phenotyping platform, based on the rotation transformation matrix, registering and aligning the three-dimensional point cloud data of the to-be-reconstructed canopy population with the seedling crop plant point cloud data corresponding to each seedling crop plant in the seedling crop population data; When it is determined that the three-dimensional point cloud data of the to-be-reconstructed canopy population is obtained through a drone platform, based on a preset reference object, registering and aligning the three-dimensional point cloud data of the to-be-reconstructed canopy population with the seedling crop plant point cloud data corresponding to each seedling crop plant in the seedling crop population data; According to the registration and alignment result, taking the central coordinate information and the azimuth information corresponding to each seedling crop plant in the seedling crop population data as the central coordinate information and the azimuth information corresponding to each crop plant at each growth stage in the three-dimensional point cloud data of the to-be-reconstructed canopy population; The cropping of the three-dimensional point cloud data of the to-be-reconstructed canopy population according to the central coordinate information and the azimuth information includes: Based on the central coordinate information and the azimuth information corresponding to each crop plant at each growth stage in the three-dimensional point cloud data of the to-be-reconstructed canopy population, constructing a cropping rectangular area for each crop plant at each growth stage; According to the cropping rectangular area, cropping the three-dimensional point cloud data of the to-be-reconstructed canopy population to obtain the target point cloud area corresponding to each crop plant at each growth stage in the three-dimensional point cloud data of the to-be-reconstructed canopy population.

5. The method for complementing and reconstructing the three-dimensional point cloud of the crop population canopy according to claim 2, wherein The interpolation processing of the point cloud plant data in the current target point cloud area based on the alternative plant point cloud data corresponding to the current target point cloud area includes: Based on the crop variety, the planting density, the growth stage information, the number of leaves, and the plant height of the crop plants at the growth stage in the current target point cloud area, determining a preset number of to-be-determined alternative plant point cloud data from multiple alternative plant point cloud data; Based on the plant height of the crop plants at the growth stage in the current target point cloud area, performing a scaling process on the to-be-determined alternative plant point cloud data to obtain the scaled to-be-determined alternative plant point cloud data; Based on the azimuth information of the seedling crop plants corresponding to the crop plants at the growth stage in the current target point cloud area, rotating the scaled to-be-determined alternative plant point cloud data to obtain the rotated to-be-determined alternative plant point cloud data; Calculating the Chamfer distance between each rotated to-be-determined alternative plant point cloud data and the point cloud plant data of the crop plants at the growth stage in the current target point cloud area, and taking the rotated to-be-determined alternative plant point cloud data corresponding to the minimum Chamfer distance as the target alternative plant point cloud data; Based on the point cloud data of the target alternative plants, interpolate the point cloud plant data of the crop plants in the current target point cloud area during the growth period to obtain the interpolated point cloud plant data, so as to obtain the completed three-dimensional point cloud reconstruction result of the canopy corresponding to the target crop population according to the interpolated point cloud plant data.

6. The method for three-dimensional point cloud completion and reconstruction of a crop population canopy according to claim 1 or 4, characterized in that The method further includes: Perform a second preprocessing on the three-dimensional point cloud data of the canopy population to be reconstructed to obtain the preprocessed three-dimensional point cloud data of the canopy population to be reconstructed, where the second preprocessing includes coordinate system calibration processing, point cloud denoising processing, ground point removal processing, and crop type plot segmentation processing; The step of, according to the central coordinate information and the azimuth information, cropping the three-dimensional point cloud data of the canopy population to be reconstructed to obtain the target point cloud area corresponding to each crop plant during the growth period in the three-dimensional point cloud data of the canopy population to be reconstructed, further includes: According to the central coordinate information and the azimuth information, crop the preprocessed three-dimensional point cloud data of the canopy population to be reconstructed to obtain the target point cloud area corresponding to each crop plant during the growth period in the preprocessed three-dimensional point cloud data of the canopy population to be reconstructed.

7. A three-dimensional point cloud completion and reconstruction system for a crop population canopy, characterized in that, It includes: A point cloud data acquisition module, configured to acquire the three-dimensional point cloud data of the canopy population to be reconstructed corresponding to the target crop population and the data of the seedling crop population, where the three-dimensional point cloud data of the canopy to be reconstructed is the three-dimensional point cloud data of the canopy population of the target crop population during other growth periods except the seedling stage of the crop; the data of the seedling crop population is the image of the seedling crop population or the point cloud data of the seedling crop population; A seedling plant positioning module, configured to determine the central coordinate information and the azimuth information corresponding to each seedling crop plant in the data of the seedling crop population; A point cloud data registration and segmentation module, configured to crop the three-dimensional point cloud data of the canopy population to be reconstructed according to the central coordinate information and the azimuth information to obtain the target point cloud area corresponding to each crop plant during the growth period in the three-dimensional point cloud data of the canopy population to be reconstructed; A point cloud completion and reconstruction module, configured to interpolate the point cloud plant data in the current target point cloud area based on the alternative plant point cloud data corresponding to the current target point cloud area. After determining that the interpolation processing of the point cloud plant data in all the target point cloud areas in the three-dimensional point cloud data of the canopy population to be reconstructed is completed, obtain the completed three-dimensional point cloud reconstruction result of the canopy corresponding to the target crop population; where the alternative plant point cloud data is obtained based on the plant point cloud data corresponding to the crop plants during the growth period in the preset area in the three-dimensional point cloud data of the canopy population to be reconstructed.

8. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the method for completing and reconstructing the three-dimensional point cloud of the crop population canopy according to any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for completing and reconstructing the three-dimensional point cloud of the crop population canopy according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for completing and reconstructing the three-dimensional point cloud of the crop population canopy according to any one of claims 1 to 6.

Images