Three-dimensional modeling-based plant pruning method and system

Abstract

Disclosed in the present application are a three-dimensional modeling-based plant pruning method and system. The method comprises: collecting three-dimensional point cloud data and multi-view image data of a target plant; using a sensing device to monitor the target plant, and collecting environmental sensing data; using a three-dimensional reconstruction technology to generate a three-dimensional model of the target plant; by means of a pre-constructed polymorphic recognition model, extracting from the three-dimensional model and the environmental sensing data a morphological feature, a physiological feature and an environmental feature of the target plant; on the basis of the three-dimensional model, recognizing the spatial distribution of branches and buds; in light of time series data, predicting a growth trend of the target plant; on the basis of a growth prediction result, generating a pruning suggestion for the branches of the target plant; and displaying the growth prediction result and / or the pruning suggestion via a visual interface, so as to assist a worker in completing pruning work.

Description

A plant pruning method and system based on 3D modeling Technical Field

[0001] This application relates to the field of agricultural technology, and more specifically, to a plant pruning method and system based on three-dimensional modeling. Background Technology

[0002] Currently, fruit tree management and pruning in agricultural production still largely rely on manual experience. Traditional pruning methods are not only inefficient but also prone to inconsistencies and errors due to a lack of scientific data support, making large-scale, standardized operations difficult. Furthermore, while advancements in agricultural technology have led to the application of automated tools and techniques, these still have limitations, particularly in complex and changing natural environments, hindering precise and effective pruning. These issues limit fruit growers' effective control over fruit yield and quality, and also impact the overall efficiency and development speed of modern agricultural production. Summary of the Invention

[0003] To address the aforementioned technical problems, this application discloses a plant pruning method and system based on 3D modeling. It utilizes a polymorphic recognition model to predict the growth trend of the target plant and generates pruning suggestions for the plant's branches based on the growth prediction results. By fusing multidimensional feature data, the accuracy of the model is improved, and 3D visualization is achieved, allowing users to intuitively see the comparison before and after pruning and its predictive effect on future growth, thus assisting manual plant pruning. Specifically, the technical solution of this application is as follows:

[0004] In a first aspect, this application discloses a plant pruning method based on three-dimensional modeling, comprising the following steps:

[0005] Collect three-dimensional point cloud data and multi-view image data of the target plant; use sensing devices to monitor the target plant and collect environmental sensing data;

[0006] A three-dimensional model of the target plant is generated using three-dimensional reconstruction technology;

[0007] Using a pre-constructed polymorphic recognition model, the morphological, physiological, and environmental characteristics of the target plant are extracted from the 3D model and the environmental sensing data; the spatial distribution of branches and buds is identified based on the 3D model; the growth trend of the target plant is predicted by combining time-series data; and pruning suggestions for the branches of the target plant are generated based on the growth prediction results.

[0008] The growth prediction results and / or pruning suggestions are displayed through a visual interface to assist manual pruning.

[0009] In some embodiments, the plant pruning method based on three-dimensional modeling further includes:

[0010] The polymorphic recognition model is constructed by combining the Transformer structure with a three-dimensional convolutional neural network (3D CNN).

[0011] The historical growth records of several types of plants were selected and collected according to specific rules as a training set to train the polymorphic recognition model.

[0012] An expert knowledge base is constructed so that the polymorphic recognition model can use machine learning algorithms to integrate the expert knowledge base to provide dynamic feedback on pruning suggestions to the user.

[0013] In some embodiments, generating a three-dimensional model of the target plant using three-dimensional reconstruction technology specifically includes:

[0014] Data preprocessing is performed on the three-dimensional point cloud data and the multi-view image data, including filtering and denoising, and data normalization;

[0015] By using point cloud registration technology, feature points are extracted from point cloud data and matched. Then, point cloud registration algorithms are used to align multi-view point cloud data to the same coordinate system.

[0016] The registered point cloud data is fused using gridding processing technology, and the fused point cloud data is converted into a grid model using a gridding algorithm.

[0017] Texture information is extracted from the multi-view image data using texture mapping technology, and the extracted texture is mapped onto the mesh model to generate a three-dimensional model, ensuring that the generated three-dimensional model can restore the spatial distribution of the branches and the position of the buds of the target plant.

[0018] In some implementations, the prediction of the growth trend of the target plant by combining time-series data specifically includes:

[0019] Physiological and environmental characteristic data of the target plant extracted at different times are arranged in chronological order to form time-series data of the target plant's growth. The growth trend of the target plant is obtained by analyzing the time-series data and the growth pattern of the target plant using the polymorphic recognition model.

[0020] In some implementations, the generation of pruning suggestions for the branches of the target plant based on growth prediction results specifically includes:

[0021] Virtual pruning operations are performed on the three-dimensional model to simulate the pruning effects under different pruning schemes.

[0022] Based on the results of the virtual pruning simulation and the predicted growth trends before and after pruning, pruning suggestions are generated, including the location and angle of branch pruning and the length of branches to be retained.

[0023] In some implementations, displaying the growth prediction results and / or pruning suggestions through a visual interface specifically includes:

[0024] Using augmented reality devices, the pruning suggestions are overlaid in real time on the three-dimensional model, including displaying the pruning position, angle, and length of the branches to be retained.

[0025] In some embodiments, displaying the growth prediction results and / or the pruning suggestions through a visual interface further includes:

[0026] Using augmented reality equipment, the growth prediction results of the target plant before and after pruning are displayed on the basis of the three-dimensional model, including the predicted direction and length of branch growth and the predicted fruit yield.

[0027] Secondly, this application also discloses a plant pruning system based on three-dimensional modeling, comprising:

[0028] The data acquisition module is used to acquire three-dimensional point cloud data and multi-view image data of the target plant; and to monitor the target plant using sensing devices and acquire environmental sensing data.

[0029] A 3D modeling module is used to generate a 3D model of the target plant using 3D reconstruction technology;

[0030] The feature extraction module is used to extract the morphological features, physiological features and environmental features of the target plant from the three-dimensional model and the environmental sensing data through a pre-constructed polymorphic recognition model.

[0031] The intelligent algorithm module is used to identify the spatial distribution of branches and buds based on the three-dimensional model using the polymorphic recognition model; predict the growth trend of the target plant by combining time-series data; and generate pruning suggestions for the branches of the target plant based on the growth prediction results.

[0032] The AR display module is used to display the growth prediction results and / or pruning suggestions through a visual interface to assist manual pruning.

[0033] In some implementations, the intelligent algorithm module specifically includes:

[0034] The identification submodule is used to identify the spatial distribution of branches and buds based on the three-dimensional model using the polymorphic identification model.

[0035] The prediction submodule is used to arrange the physiological and environmental characteristic data of the target plant extracted at different times in chronological order to form time-series data of the target plant's growth; and to use the polymorphic recognition model to analyze the growth trend of the target plant by combining the time-series data with the growth pattern of the target plant.

[0036] The simulated pruning submodule is used to perform virtual pruning operations based on the three-dimensional model, simulating the pruning effects under different pruning schemes; and generating pruning suggestions based on the results of the virtual pruning simulation and the predicted growth trends before and after pruning, including the pruning position, angle, and length of the branches to be retained.

[0037] In some implementations, the AR display module is specifically used to: utilize an augmented reality device to overlay the pruning suggestions on the three-dimensional model in real time, including displaying the pruning position, angle, and length of the branches to be retained;

[0038] The AR display module is also used to display the growth prediction results of the target plant before and after pruning based on the three-dimensional model using augmented reality devices, including the predicted direction and length of branch growth and the predicted fruit yield.

[0039] Compared with the prior art, this application has at least one of the following beneficial effects:

[0040] 1. This application employs deep learning algorithms to accurately identify the structure and growth status of plants by analyzing multi-source data, predicting the growth trend of target plants and generating scientific pruning suggestions. This provides precise guidance for pruning, improving pruning efficiency and reducing human error. It ensures that neither over-pruning nor neglecting necessary thinning is achieved, contributing to the scaling up and standardization of modern agricultural production.

[0041] 2. This application utilizes AR devices, allowing users to directly see virtual pruning suggestions overlaid on the actual plants in the real-world environment. This intuitive visualization helps users perform pruning operations more accurately. AR technology can display the location, angle, and length of branches to be pruned in real time, reducing errors from human judgment and thus improving the accuracy and consistency of pruning. AR devices can quickly provide pruning suggestions, eliminating the need for users to frequently consult paper guides or rely on experience, thereby significantly improving the efficiency of pruning work.

[0042] 3. The plant pruning method described in this application reduces the skill requirements for operators. Even beginners can quickly learn to use AR devices and complete the pruning work by following virtual prompts, thereby expanding the pool of people who can perform pruning tasks. Combined with robotic technology, automated pruning is achieved, significantly reducing labor costs. Attached Figure Description

[0043] The preferred embodiments will now be described in a clear and easy-to-understand manner, in conjunction with the accompanying drawings, to further explain the above-mentioned characteristics, technical features, advantages, and implementation methods of this application.

[0044] Figure 1 is a flowchart of the steps of an embodiment of a plant pruning method based on three-dimensional modeling according to this application;

[0045] Figure 2 is a structural block diagram of an embodiment of a plant pruning system based on three-dimensional modeling according to this application. Detailed Implementation

[0046] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0047] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or sets.

[0048] To keep the drawings concise, each figure only schematically shows the parts relevant to the invention, and these do not represent the actual structure of the product. Furthermore, to facilitate understanding, in some figures, only one of components with the same structure or function is schematically depicted, or only one is labeled. In this document, "one" not only means "only one," but can also mean "more than one."

[0049] It should also be further understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0050] In specific implementations, the terminal devices described in the embodiments of this application include, but are not limited to, other portable devices such as mobile phones, laptops, educational computers, or tablet computers with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads). It should also be understood that in some embodiments, the terminal device is not a portable communication device, but a desktop computer with touch-sensitive surfaces (e.g., touchscreen displays and / or touchpads).

[0051] Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0052] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the specific implementation methods of this application will be described below with reference to the accompanying drawings. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings and other implementation methods can be obtained based on these drawings without creative effort.

[0053] Existing fruit tree pruning techniques mainly include the following:

[0054] 1. Traditional manual pruning: This method relies entirely on the worker's personal experience and skills. While it offers high flexibility, it is inefficient, labor-intensive, and the pruning quality is greatly affected by individual differences.

[0055] 2. Automatic recognition technology based on two-dimensional image processing: This type of technology attempts to determine the best pruning position by taking pictures of the tree canopy. However, since it only considers planar information, it cannot accurately reflect the real spatial structure of the tree. In particular, when the leaves fall, the occlusion relationship between the branches becomes more complex, and the effectiveness of this method is even more limited.

[0056] 3. Automated Pruning Robots: In recent years, researchers have developed some robotic devices that can move autonomously in the field and perform pruning tasks. Although they can improve work efficiency and reduce labor requirements, their high cost makes them less popular. At the same time, these robots are not adaptable and flexible enough, and often fail to meet different garden conditions.

[0057] Each approach has its applicable scope and specific advantages, but also comes with obvious drawbacks. Currently available fruit tree pruning techniques each have their merits, but in practical application, they generally suffer from the following shortcomings:

[0058] For fruit trees that have already lost their leaves, traditional visual assessment methods struggle to accurately determine which branches should be removed because the layering and density variations between branches are difficult to discern without leaf cover. Existing two-dimensional image-based recognition methods cannot capture the three-dimensional structural information of fruit trees, making it difficult to meet the needs of precise pruning.

[0059] Using robotic arms or fixed pruning devices for large-scale operations may cause unnecessary damage or miss important growth stages, thus affecting the subsequent growth and development of the plants.

[0060] Although some advanced equipment integrates multiple sensing elements and can better perceive changes in the surrounding environment, it is difficult to promote due to its high level of specialization, and ordinary farmers often cannot afford such expensive investments.

[0061] This application aims to provide a more efficient, low-cost, and easy-to-use method to assist farmers in properly and effectively pruning fruit trees during winter or other seasons, ensuring that neither over-pruning nor neglecting necessary thinning is achieved. By integrating multi-dimensional feature data, the accuracy of the model is improved, overcoming the bias caused by single-dimensional data. Combining fruit tree growth prediction with augmented reality technology, a three-dimensional visualization is achieved, allowing users to intuitively see the before-and-after comparison of pruning effects and its predicted impact on future growth.

[0062] This application discloses a plant pruning method and system based on 3D modeling, which can be used not only for fruit tree pruning but also for pruning shrubs, lawns, potted plants, green belts, vines, and other plants, playing a role in horticultural design, agriculture, and art. This application describes the method using the pruning of deciduous fruit trees in winter as an example. Winter is the main period for fruit tree pruning, usually carried out after the fruit trees enter dormancy and before spring budding. At this time, the physiological activity of the fruit trees is relatively weak, and the pruning causes less damage to the tree. The main purpose is to remove diseased branches, insect-infested branches, dead branches, crossing branches, and overly dense branches, reducing overwintering sites for pests and diseases, while adjusting the canopy structure and increasing the fruit-bearing rate. Of course, the choice of other seasons or other plant species does not affect the implementation of the technical solution in this application. It should be understood that although preferred embodiments of this application have been described, those skilled in the art, once they understand the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0063] Referring to Figure 1 in the specification, an embodiment of a plant pruning method based on three-dimensional modeling according to this application specifically includes the following steps:

[0064] S100: Collect three-dimensional point cloud data and multi-view image data of the target plant; use sensing devices to monitor the target plant and collect environmental sensing data.

[0065] Specifically, LiDAR and multi-view cameras are used to collect three-dimensional point cloud data and multi-view image data of deciduous fruit trees, and environmental data such as light, temperature, humidity and soil conditions in the orchard are collected through environmental sensors.

[0066] S200, using 3D reconstruction technology to generate a 3D model of the target plant.

[0067] Specifically, based on the collected 3D point cloud data and multi-view image data, a high-precision 3D model of deciduous fruit trees is constructed using 3D reconstruction technology to reflect the spatial location of branches and buds in the model.

[0068] S300 uses a pre-built polymorphic recognition model to extract the morphological, physiological, and environmental characteristics of the target plant from the three-dimensional model and the environmental sensing data.

[0069] Specifically, based on deep learning algorithms, the morphological, physiological, and environmental characteristics of the fruit trees are integrated to extract polymorphic feature data. The morphological characteristics include branch distribution and bud location, the physiological characteristics include historical growth data and nutritional status, and the environmental characteristics include sunshine duration, light intensity, temperature and humidity, and soil conditions.

[0070] S400: Based on the three-dimensional model, identify the spatial distribution of branches and buds; predict the growth trend of the target plant by combining time-series data; and generate pruning suggestions for the branches of the target plant based on the growth prediction results.

[0071] Specifically, virtual pruning operations are performed based on the aforementioned 3D model to simulate the pruning effects of different pruning schemes. The post-pruning growth trend is predicted by combining this with the fruit tree's growth patterns, including branch growth direction, length changes, and fruit yield estimates. The optimal pruning scheme is selected as the recommended pruning plan from among the different schemes. The pruning recommendations and prediction results are displayed through a visual interface to assist fruit growers in their decision-making.

[0072] S500 displays the growth prediction results and / or pruning suggestions through a visual interface to assist manual pruning.

[0073] Specifically, the visual interface is an AR device interface that uses AR devices to overlay pruning suggestions in real time, guiding fruit growers to perform precise pruning. Combined with an expert knowledge base, it provides dynamic feedback, adjusting pruning suggestions in real time to reduce the error rate.

[0074] This application employs deep learning algorithms to accurately identify the structure and growth status of plants by analyzing multi-source data, predicts the growth trend of target plants, and generates scientific pruning suggestions, thus providing precise guidance for pruning. This method not only improves pruning efficiency but also reduces errors from manual operation, ensuring that neither over-pruning nor neglecting necessary thinning is achieved.

[0075] This application discloses another embodiment of a plant pruning method based on three-dimensional modeling. Based on the above embodiment, step S100 specifically includes the following sub-steps:

[0076] S110, use lidar to collect the three-dimensional point cloud data of the target plant and the terrain within the target area where the target plant is located.

[0077] Specifically, a lidar scanner is used to acquire topographic information of the target area. Alternatively, a depth camera can be used instead of lidar to acquire 3D point cloud data.

[0078] S120, using an image acquisition device to photograph the target plant from different angles to acquire the multi-view image data of the target plant.

[0079] Specifically, the image acquisition equipment includes multi-view or single-view cameras. Position the image acquisition equipment at a suitable height and slowly walk around the fruit tree to be pruned, ensuring that all parts are captured.

[0080] S130 uses soil sensors and meteorological sensors to collect environmental characteristic data of the target plant, including: soil moisture, temperature, pH value; air temperature and humidity, light intensity, wind speed, and rainfall.

[0081] Specifically, soil sensors are used to monitor parameters such as soil moisture, temperature, pH, and electrical conductivity. These parameters help to understand soil fertility and moisture conditions, thereby optimizing fertilization and irrigation strategies.

[0082] Meteorological sensors: used to monitor meteorological parameters such as air temperature and humidity, light intensity, wind speed, and rainfall. This data helps assess the impact of orchard climate conditions on fruit tree growth.

[0083] S140 uses plant physiological sensors to collect physiological characteristic data of the target plant, including: canopy coverage, tree height, fruit load, and nutritional status.

[0084] Specifically, plant physiological sensors are used to monitor growth indicators of fruit trees, such as leaf surface temperature, leaf surface humidity, fruit enlargement, leaf thickness, and leaf area. These data can reflect the growth status and health condition of the fruit trees.

[0085] This application utilizes a LiDAR scanner to acquire topographical information within the target area, and combines this with multi-angle image data captured by a camera to construct a high-precision 3D model of the fruit trees. Specific operations include: turning on the device power, starting the multi-state recognition program, and confirming that the sensors are operating normally. During data acquisition, weather conditions can be considered to avoid the influence of sunlight; data can be collected separately for sunny, cloudy, and overcast days for the fruit trees to be pruned. After data collection is complete, the system switches to the 3D modeling interface and waits for the modeling system to load.

[0086] The process of establishing a 3D model relies on 3D reconstruction techniques, including point cloud registration, meshing, and texture mapping. This application provides another embodiment of a plant pruning method based on 3D modeling. Based on any embodiment of the above method, step S200 specifically includes the following sub-steps:

[0087] S210, perform data preprocessing on the three-dimensional point cloud data and the multi-view image data, including filtering and denoising, and data normalization.

[0088] Specifically, point cloud data is processed by denoising, filtering, and cropping, while image data is processed by correction and feature extraction.

[0089] S220 uses point cloud registration technology to extract feature points from point cloud data and match them, and uses a point cloud registration algorithm to align multi-view point cloud data to the same coordinate system.

[0090] Specifically, feature points (such as SIFT or SURF features) are extracted from the point cloud data and matched. Iterative nearest point (ICP) or feature-based registration algorithms (such as RANSAC) are used to align the multi-view point cloud data to the same coordinate system. Global optimization algorithms (such as the Gauss-Newton method or the Levenberg-Marquardt method) are then used to further improve the registration accuracy.

[0091] The S230 uses gridding technology to fuse registered point cloud data and then uses a gridding algorithm to convert the fused point cloud data into a grid model.

[0092] Specifically, the registered point cloud data are fused to generate a uniform, dense point cloud. A meshing algorithm (such as Marching Cubes or Poisson reconstruction) is then used to convert the point cloud data into a triangular mesh model. The generated mesh is then simplified, smoothed, and repaired to improve the model's quality.

[0093] S240, texture information is extracted from the multi-view image data using texture mapping technology, and the extracted texture is mapped onto the mesh model to generate a three-dimensional model, so as to ensure that the generated three-dimensional model can restore the spatial distribution of the branches and the position of the buds of the target plant.

[0094] Specifically, texture information is extracted from multi-view images, and the extracted texture is mapped onto a mesh model to generate a realistic 3D model.

[0095] Optionally, this also includes further optimization of the generated 3D model, addressing texture overlap and inconsistencies between multi-view images to ensure texture continuity and consistency. This includes reducing the number of polygons and improving texture quality.

[0096] This embodiment provides another implementation method. Based on this embodiment, in step S300, the morphological features, physiological features, and environmental features of the target plant are extracted from the three-dimensional model and the environmental sensing data using a pre-constructed polymorphic recognition model. Specifically, the polymorphic recognition model adopts a deep learning model that combines a Transformer structure with a three-dimensional convolutional neural network (3D CNN) to extract the morphology, physiology (historical growth data, nutritional status, etc.), and environmental features (duration of sunshine, light intensity, temperature and humidity, soil conditions, etc.) of the fruit tree.

[0097] Based on this, in another embodiment of this example, before using the polymorphic recognition model, the method further includes: S301, constructing the polymorphic recognition model by combining a Transformer structure with a three-dimensional convolutional neural network (3D CNN).

[0098] Specifically, by integrating multi-dimensional information such as the morphological characteristics of fruit trees (branch distribution, bud location), physiological characteristics (historical growth data, nutritional status), and external environmental characteristics (duration of sunshine, light intensity, temperature and humidity, soil conditions), a multi-morphic recognition model is constructed to improve the recognition accuracy of the branch structure of deciduous fruit trees.

[0099] S302, Select and collect historical growth records of several types of plants according to specific rules as a training set, and train the polymorphic recognition model.

[0100] S303, Construct an expert knowledge base so that the polymorphic recognition model can use machine learning algorithms to integrate the expert knowledge base to provide dynamic feedback on pruning suggestions to the user.

[0101] Specifically, expert systems store professional knowledge through a knowledge base, which contains factual knowledge (known information) and heuristic knowledge (rules based on experience). Polymorphic recognition models improve their problem-solving capabilities by learning this knowledge.

[0102] This application's polymorphic recognition model can utilize machine learning algorithms to analyze user behavior, automatically update knowledge base content, mark outdated information, and generate new content based on user needs. Furthermore, through user feedback and system self-learning, the knowledge base can be continuously optimized, improving the accuracy and relevance of information.

[0103] This application provides another embodiment of a plant pruning method based on three-dimensional modeling. Based on any embodiment of the above method, step S400 specifically includes the following sub-steps:

[0104] S410, using the polymorphic recognition model, the spatial distribution of branches and buds of the target plant is identified based on the three-dimensional model.

[0105] Specifically, the polymorphic recognition model uses a 3D model of the target plant to accurately and dynamically identify the morphological features (branch distribution, bud location) of deciduous fruit trees. The spatial locations of branches and buds are then marked on the 3D model of the fruit tree.

[0106] S420, the physiological and environmental characteristics data of the target plant collected at different times are arranged in chronological order to form time-series data of the target plant's growth; the growth trend of the target plant is obtained by analyzing the time-series data using the polymorphic recognition model.

[0107] Specifically, the collected multi-source data is fused and analyzed to obtain time-series data on fruit tree growth. For example, a polymorphic recognition model integrates multi-dimensional information such as the physiological characteristics of fruit trees (historical growth data, nutritional status) and external environmental characteristics (duration of sunshine, light intensity, temperature and humidity, soil conditions) to form comprehensive fruit tree growth monitoring data, which is then arranged in chronological order to form time-series data. Using polymorphic recognition model data analysis tools, trend analysis and correlation analysis are performed on the time-series data to assess the growth trend and environmental adaptability of the fruit trees. Combining growth data and historical yield data, machine learning models are used to predict future fruit yield.

[0108] S430, based on the three-dimensional model, perform virtual pruning operations to simulate the pruning effects under different pruning schemes; generate pruning suggestions based on the results of the virtual pruning simulation and the predicted growth trends before and after pruning, including the pruning position, angle, and length of the branches to be retained.

[0109] Specifically, the spatial locations of branches and buds are marked in the model. Virtual pruning is then performed based on this 3D model to simulate the effects of different pruning schemes. The post-pruning growth trend is predicted by combining this simulation with the tree's growth patterns, including branch growth direction, length changes, and fruit yield estimates. Simulating pruning effects in a virtual environment and predicting future growth trends based on the tree's growth patterns provides a scientific basis for pruning decisions. Based on the results of the virtual pruning simulation, pruning suggestions are generated, including information such as the location of pruning, the length to retain, and the pruning angle.

[0110] This application provides another embodiment of a plant pruning method based on three-dimensional modeling. Based on any embodiment of the above method, step S500 specifically includes the following sub-steps:

[0111] S510, using augmented reality equipment, overlays the pruning suggestions on the three-dimensional model in real time, including displaying the pruning position, angle, and length of the branches to be retained.

[0112] S520, using augmented reality equipment, displays the growth prediction results of the target plant before and after pruning based on the three-dimensional model, including the predicted direction and length of branch growth and the predicted fruit yield.

[0113] Specifically, augmented reality (AR) devices (such as smart glasses) are used to overlay trimming suggestions in real time. Combined with an expert knowledge base and dynamic feedback mechanisms, the trimming suggestions are adjusted in real time, and the conclusions are presented to the user in the form of a transparent layer, guiding them to act accordingly. The specific operation can be reflected in the following steps: wearing the AR smart device and initially adjusting the viewing angle until satisfied; activating the virtual trimming mode and selecting the initial trimming path; following the arrows on the screen to perform each action sequentially; after completing one round of trimming, returning to view the trimming effect or the expected effect, and repeating some of the above steps until perfect is achieved if further adjustments are needed.

[0114] AR devices can quickly provide pruning suggestions, eliminating the need for users to frequently consult paper guides or rely on experience, thus significantly improving the efficiency of pruning. Furthermore, combined with deep learning algorithms, AR devices can dynamically adjust pruning suggestions based on the real-time pruning results of the plants, further optimizing the pruning process.

[0115] This application provides another embodiment of a plant pruning method based on three-dimensional modeling. Referring to Figure 2 of the specification, based on any embodiment of the above method, it further includes: after the actual pruning is completed, evaluating the pruning effect, including the branch removal rate, the accuracy of the retained length, and the impact on the growth of the fruit tree, so as to further optimize the pruning suggestion generation algorithm.

[0116] Based on the same concept, this application also discloses a plant pruning system based on three-dimensional modeling. The system is used to implement the steps described in any of the above method embodiments. Specifically, one embodiment of the plant pruning system based on three-dimensional modeling in this application includes:

[0117] The data acquisition module is used to acquire three-dimensional point cloud data and multi-view image data of the target plant; and to monitor the target plant using sensing devices and acquire environmental sensing data.

[0118] The 3D modeling module is used to generate a 3D model of the target plant using 3D reconstruction technology.

[0119] The feature extraction module is used to extract the morphological, physiological and environmental features of the target plant from the three-dimensional model and the environmental sensing data through a pre-constructed polymorphic recognition model.

[0120] The intelligent algorithm module is used to identify the spatial distribution of branches and buds based on the three-dimensional model using the polymorphic recognition model; predict the growth trend of the target plant by combining time-series data; and generate pruning suggestions for the branches of the target plant based on the growth prediction results.

[0121] The AR display module presents growth prediction results and pruning suggestions intuitively through a visual interface, providing precise guidance for manual pruning and ensuring the efficient and accurate execution of pruning work.

[0122] Based on the above embodiments, this application discloses another embodiment of a plant pruning system based on three-dimensional modeling, wherein the data acquisition module specifically includes the following sub-modules:

[0123] A lidar is used to collect the three-dimensional point cloud data of the target plant and the terrain within the target area where the target plant is located.

[0124] An image acquisition device is used to photograph the target plant from different angles to acquire multi-view image data of the target plant.

[0125] Sensing devices include soil sensors, weather sensors, and plant physiological sensors.

[0126] Specifically, soil sensors are used to monitor parameters such as soil moisture, temperature, pH, and electrical conductivity. These parameters help to understand soil fertility and moisture conditions, thereby optimizing fertilization and irrigation strategies.

[0127] Meteorological sensors: used to monitor meteorological parameters such as air temperature and humidity, light intensity, wind speed, and rainfall. This data helps assess the impact of orchard climate conditions on fruit tree growth.

[0128] Plant physiological sensors: used to monitor growth indicators of fruit trees, such as leaf surface temperature, leaf surface humidity, fruit enlargement, leaf thickness, and leaf area. These data can reflect the growth status and health of the fruit trees.

[0129] In another embodiment of this application, the three-dimensional modeling module specifically includes the following sub-modules:

[0130] The data preprocessing submodule is used to perform data preprocessing on the 3D point cloud data and the multi-view image data, including filtering and denoising, and data normalization.

[0131] The point cloud registration submodule is used to extract feature points from point cloud data and match them using point cloud registration technology, and to align multi-view point cloud data to the same coordinate system using point cloud registration algorithms.

[0132] The meshing processing submodule is used to fuse the registered point cloud data using meshing processing technology, and to convert the fused point cloud data into a mesh model using a meshing algorithm.

[0133] The texture mapping submodule is used to extract texture information from the multi-view image data using texture mapping technology, and to map the extracted texture onto the mesh model to generate a three-dimensional model.

[0134] Based on the above embodiments, this application discloses another embodiment of a plant pruning system based on three-dimensional modeling, wherein the intelligent algorithm module specifically includes the following sub-modules:

[0135] The identification submodule is used to identify the spatial distribution of branches and buds based on the three-dimensional model using the polymorphic identification model.

[0136] The prediction submodule is used to arrange the physiological and environmental characteristic data of the target plant extracted at different times in chronological order to form time-series data of the target plant's growth; and to use the polymorphic recognition model to analyze the growth trend of the target plant by combining the time-series data with the growth pattern of the target plant.

[0137] The simulated pruning submodule is used to perform virtual pruning operations based on the three-dimensional model, simulating the pruning effects under different pruning schemes; and generating pruning suggestions based on the results of the virtual pruning simulation and the predicted growth trends before and after pruning, including the pruning position, angle, and length of the branches to be retained.

[0138] This application provides another embodiment of a plant pruning system based on three-dimensional modeling, which, in addition to any embodiment of the above system, further includes:

[0139] The AR display module is specifically used to: utilize augmented reality devices to overlay the pruning suggestions on the three-dimensional model in real time, including displaying the pruning position, angle, and length of the branches to be retained.

[0140] The AR display module is also used to display the growth prediction results of the target plant before and after pruning based on the three-dimensional model using augmented reality devices, including the predicted direction and length of branch growth and the predicted fruit yield.

[0141] More preferably, this application discloses another embodiment of a plant pruning system based on three-dimensional modeling. Based on any of the above system embodiments, this embodiment further includes the following modules:

[0142] The model building module is used to construct the polymorphic recognition model by combining the Transformer structure with a three-dimensional convolutional neural network (3D CNN); to select and collect historical growth records of several types of plants according to specific rules as a training set to train the polymorphic recognition model; and to build an expert knowledge base so that the polymorphic recognition model can use machine learning algorithms to integrate the expert knowledge base to provide dynamic feedback on pruning suggestions to users.

[0143] The model optimization module is used to continuously optimize and adjust the polymorphic recognition model based on feedback from actual pruning results, so as to improve the model's recognition accuracy and adaptability.

[0144] The pruning effect evaluation module is used to evaluate the pruning effect after the actual pruning is completed, including the branch removal rate, the accuracy of the retained length, and the impact on the growth of fruit trees, so as to further optimize the pruning suggestion generation algorithm.

[0145] The plant pruning method and system based on three-dimensional modeling in this application have the same technical concept, and the technical details of the embodiments of the two are applicable to each other. In order to reduce repetition, they will not be repeated here.

[0146] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of program modules is merely an example. In practical applications, the above functions can be assigned to different program modules as needed, that is, the internal structure of the device can be divided into different program units or modules to complete all or part of the functions described above. The program modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one processing unit. The integrated unit can be implemented in hardware or as a software program unit. Furthermore, the specific names of the program modules are only for easy differentiation and are not intended to limit the scope of protection of this application.

[0147] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A plant pruning method based on three-dimensional modeling, characterized in that, Includes the following steps: Collect three-dimensional point cloud data and multi-view image data of the target plant; use sensing devices to monitor the target plant and collect environmental sensing data; A three-dimensional model of the target plant was generated using three-dimensional reconstruction technology; The morphological, physiological, and environmental characteristics of the target plant are extracted from the three-dimensional model and the environmental sensing data using a pre-constructed polymorphic recognition model. Based on the three-dimensional model, the spatial distribution of branches and buds is identified; and the growth trend of the target plant is predicted by combining time-series data. Based on the growth prediction results, pruning suggestions for the branches and trunks of the target plant are generated; The growth prediction results and / or pruning suggestions are displayed through a visual interface to assist manual pruning.

2. The plant pruning method based on three-dimensional modeling as described in claim 1, characterized in that, Also includes: The polymorphic recognition model is constructed by combining the Transformer structure with a three-dimensional convolutional neural network (3D CNN). The historical growth records of several types of plants were selected and collected according to specific rules as a training set to train the polymorphic recognition model. An expert knowledge base is constructed so that the polymorphic recognition model can use machine learning algorithms to integrate the expert knowledge base to provide dynamic feedback on pruning suggestions to the user.

3. The plant pruning method based on three-dimensional modeling as described in claim 1, characterized in that, The method of generating a three-dimensional model of the target plant using three-dimensional reconstruction technology specifically includes: Data preprocessing is performed on the three-dimensional point cloud data and the multi-view image data, including filtering and denoising, and data normalization; By using point cloud registration technology, feature points are extracted from point cloud data and matched. Then, point cloud registration algorithms are used to align multi-view point cloud data to the same coordinate system. The registered point cloud data is fused using gridding processing technology, and the fused point cloud data is converted into a grid model using a gridding algorithm. Texture information is extracted from the multi-view image data using texture mapping technology, and the extracted texture is mapped onto the mesh model to generate a three-dimensional model, ensuring that the generated three-dimensional model can restore the spatial distribution of the branches and the position of the buds of the target plant.

4. The plant pruning method based on three-dimensional modeling as described in claim 1, characterized in that, The prediction of the growth trend of the target plant by combining time-series data specifically includes: Physiological and environmental characteristic data of the target plant extracted at different times are arranged in chronological order to form time-series data of the target plant's growth. The growth trend of the target plant is obtained by combining the time-series data with the growth pattern of the target plant using the polymorphic recognition model.

5. A plant pruning method based on three-dimensional modeling as described in claim 1 or 4, characterized in that, The aforementioned pruning suggestions for the target plant's branches and trunks, generated based on growth prediction results, specifically include: Virtual pruning operations are performed on the three-dimensional model to simulate the pruning effects under different pruning schemes. Based on the results of the virtual pruning simulation and the predicted growth trends before and after pruning, pruning suggestions are generated, including the location and angle of branch pruning and the length of branches to be retained.

6. The plant pruning method based on three-dimensional modeling as described in claim 1, characterized in that, The provision of a visual interface to display the growth prediction results and / or the pruning suggestions specifically includes: Using augmented reality devices, the pruning suggestions are overlaid in real time on the three-dimensional model, including displaying the pruning position, angle, and length of the branches to be retained.

7. A plant pruning method based on three-dimensional modeling as described in claim 1 or 6, characterized in that, The method of displaying the growth prediction results and / or the pruning suggestions through a visual interface also includes: Using augmented reality equipment, the growth prediction results of the target plant before and after pruning are displayed on the basis of the three-dimensional model, including the predicted direction and length of branch growth and the predicted fruit yield.

8. A plant pruning system based on three-dimensional modeling, characterized in that, include: The data acquisition module is used to acquire three-dimensional point cloud data and multi-view image data of the target plant. The target plant is monitored using sensing devices to collect environmental sensing data. A 3D modeling module is used to generate a 3D model of the target plant using 3D reconstruction technology; The feature extraction module is used to extract the morphological features, physiological features and environmental features of the target plant from the three-dimensional model and the environmental sensing data through a pre-constructed polymorphic recognition model. The intelligent algorithm module is used to identify the spatial distribution of branches and buds based on the three-dimensional model using the polymorphic recognition model; and to predict the growth trend of the target plant by combining time-series data. Based on the growth prediction results, pruning suggestions for the branches and trunks of the target plant are generated; The AR display module is used to display the growth prediction results and / or pruning suggestions through a visual interface to assist manual pruning.

9. A plant pruning system based on three-dimensional modeling as described in claim 8, characterized in that, The intelligent algorithm module specifically includes: The identification submodule is used to identify the spatial distribution of branches and buds based on the three-dimensional model using the polymorphic identification model. The prediction submodule is used to arrange the physiological and environmental characteristic data of the target plant extracted at different times in chronological order to form the time-series data of the target plant's growth; and to use the polymorphic recognition model to analyze the growth trend of the target plant by combining the time-series data with the growth pattern of the target plant. The simulated pruning submodule is used to perform virtual pruning operations based on the three-dimensional model, simulating the pruning effects under different pruning schemes; and generating pruning suggestions based on the results of the virtual pruning simulation and the predicted growth trends before and after pruning, including the pruning position, angle, and length of the branches to be retained.

10. A plant pruning system based on three-dimensional modeling as described in claim 8, characterized in that, The AR display module is specifically used to: utilize augmented reality devices to overlay the pruning suggestions on the three-dimensional model in real time, including displaying the pruning position, angle, and length of the branches to be retained; The AR display module is also used to display the growth prediction results of the target plant before and after pruning based on the three-dimensional model using augmented reality devices, including the predicted direction and length of branch growth and the predicted fruit yield.

Images