A method and device for measuring the number of spherical branches of long-stem grape fern

Abstract

This application discloses a method and device for measuring the number of spherical branchlets in long-stem grape ferns, relating to the technical field of long-stem grape ferns detection technology. The method comprises: using a small animal in vivo microCT imager to collect three-dimensional point cloud data of each long-stem grape ferns sample; constructing a dataset by annotating the spherical branchlets in each 3D point cloud data; using the dataset to train a point cloud segmentation network to obtain a target segmentation model; and using the target segmentation model to detect the number of spherical branchlets in the long-stem grape ferns to be detected. This application can achieve non-destructive, high-throughput measurement of the number of spherical branchlets.

Description

Technical Field

[0001] The present application relates to the technical field of long-stem grape ferns detection, and in particular to a method and device for measuring the number of spherical branches of long-stem grape ferns. Background Art

[0002] The measurement of long-stemmed grape fern algae is generally done manually, that is, using tweezers to remove the spherical branches one by one and count them. Manual measurement of the characteristics of upright branches is time-consuming and labor-intensive, and may also cause the upright branches to wilt and die. In addition, due to factors such as human subjective errors and complex data management, the data accuracy is low. In addition, the algae itself has a special grape morphological structure. Extracting the phenotypic characteristics of upright branches from two-dimensional images results in inaccurate data and slow processing speed due to the mutual occlusion between the spherical branches. Summary of the Invention

[0003] The purpose of this application is to provide a method and device for measuring the number of spherical branchlets of long-stem grape fern, which can improve the accuracy of the statistics of spherical upright branches.

[0004] To achieve the above objectives, this application provides the following solutions:

[0005] In a first aspect, the present application provides a method for measuring the number of spherical branchlets of long-stem grape fern algae, the method comprising:

[0006] Small animal in vivo microCT imaging was used to collect three-dimensional point cloud data of each long-stem grape fern algae sample;

[0007] The dataset is constructed by annotating spherical twigs on each 3D point cloud data;

[0008] Use the dataset to train the point cloud segmentation network to obtain the target segmentation model;

[0009] The target segmentation model is used to detect the number of spherical branches of the long-stem grape fern to be detected.

[0010] Optionally, a small animal in vivo microCT imager is used to collect three-dimensional point cloud data of each long-stem grape fern algae sample, specifically including:

[0011] Starting from one end of the sample bed groove, a polyethylene film is wrapped along the radial contour of the sample bed groove, and the polyethylene film is stretched to form a support plane, wherein the support plane prevents the long-stem grape fern algae sample from contacting the bottom of the sample bed groove, thereby obtaining a processed long-stem grape fern algae sample;

[0012] After placing the processed long-stem grape fern algae sample into the small animal in vivo microCT imaging chamber, setting the operating parameters of the small animal in vivo microCT imaging chamber according to preset operating parameter conditions, wherein the preset operating parameter conditions include that the voltage and current meet the set conditions, the field of view is adjusted to the minimum, the camera speed is the set speed, and the camera rotation angle is the set angle;

[0013] The three-dimensional point cloud data is obtained after the small animal living microCT imager performs CT scanning on the long-stem grape fern algae sample.

[0014] Optionally, the voltage and current meet the set conditions that the voltage The current is ≤8W, the set speed is 21° / s, and the set angle is 384°.

[0015] Optionally, the preset working parameter conditions also include the diameter of the X-ray scanning area being set to 86 mm and the voxel size being set to 172 μm.

[0016] Optionally, the point cloud segmentation network adopts a PointNet++ network.

[0017] Optionally, the target segmentation model is used to detect the number of spherical branches of the long-stem grape fern to be detected, specifically including:

[0018] Small animal in vivo microCT imaging was used to collect three-dimensional point cloud data of the long-stemmed grape fern to be tested;

[0019] The three-dimensional point cloud data of the long-stem grape cauliflower to be detected is input into the target segmentation model to obtain the total number of spherical branches of the long-stem grape cauliflower to be detected.

[0020] Optionally, the three-dimensional point cloud data is three-dimensional point cloud data in ply format.

[0021] Optionally, a data set is constructed by annotating spherical twigs on each 3D point cloud data, specifically including:

[0022] The boundaries of the spherical branches in each 3D point cloud data are annotated, and the bounding box of each spherical branch and the total number of spherical branches are used as label data.

[0023] Optionally, the boundary of each spherical branch of the three-dimensional point cloud data is marked, specifically including:

[0024] CloudCompare software was used to annotate the boundaries of the spherical branches in each 3D point cloud data.

[0025] In a second aspect, the present application provides a device for measuring the number of spherical branchlets of a long-stem grape ferns algae, characterized in that the device for measuring the number of spherical branchlets of a long-stem grape ferns algae applies any of the above-described methods for measuring the number of spherical branchlets of a long-stem grape ferns algae, and the device for measuring the number of spherical branchlets of a long-stem grape ferns algae comprises:

[0026] A three-dimensional point cloud data acquisition module is used to acquire three-dimensional point cloud data of each long-stem grape fern algae sample using a small animal in vivo microCT imager;

[0027] A data set construction module, used for constructing a data set by annotating spherical twigs on each three-dimensional point cloud data;

[0028] The model training module is used to train the point cloud segmentation network using the data set to obtain the target segmentation model;

[0029] The detection module is used to detect the number of spherical branches of the long-stem grape fern to be detected by using the target segmentation model.

[0030] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0031] The present application provides a method and device for measuring the number of spherical branchlets of long-stem grape fern. A small animal in vivo microCT imager (Quantum GX2) is used to collect three-dimensional (3D) point cloud data of each long-stem grape fern sample. The obtained point cloud data has high resolution and fast scanning speed. A target segmentation model is used to detect the number of upright branches (spherical branchlets) of the long-stem grape fern to be detected. Compared with manual counting, the present application improves the accuracy and efficiency of spherical upright branch counting based on CT-3D imaging and target segmentation modules, and realizes non-destructive high-throughput measurement of upright branches, providing an effective tool for measuring the morphological traits of long-stem grape fern and a scientific method for variety selection and improvement. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0033] Figure 1 A schematic flow chart of a method for measuring the number of spherical branchlets of a long-stem grape fern provided in one embodiment of the present application.

[0034] Figure 2 Schematic diagram of the positional relationship between the long-stem grape fern algae sample and the sample bed provided in one embodiment of the present application.

[0035] Figure 3 A schematic diagram of an interface for separating spherical branchlets during the point cloud annotation process provided in one embodiment of the present application.

[0036] Figure 4 A schematic diagram of an interface for selecting a segmented point cloud during point cloud annotation provided in an embodiment of the present application.

[0037] Figure 5 A schematic diagram of an interface for annotating objects in a point cloud during the point cloud annotation process provided in one embodiment of the present application.

[0038] Figure 6 A schematic diagram of the interface after saving during the point cloud annotation process provided in one embodiment of the present application.

[0039] Figure 7 A schematic diagram of manual threshing counting provided in one embodiment of the present application.

[0040] Figure 8 This is a schematic diagram of the functional modules of a device for measuring the number of spherical branchlets of a long-stem grape fern provided in one embodiment of the present application. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0042] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0043] In an exemplary embodiment, the present application provides a method for measuring the number of spherical branches of long-stem grape fern, such as Figure 1 As shown, the method for measuring the number of spherical branchlets of long-stem grape fern includes steps 101 to 104.

[0044] Step 101: using a small animal in vivo microCT imager to collect three-dimensional point cloud data of each long-stem grape fern algae sample.

[0045] Step 102: Construct a data set by annotating spherical twigs on each 3D point cloud data.

[0046] Step 103: Use the dataset to train the point cloud segmentation network to obtain a target segmentation model.

[0047] Step 104: using the target segmentation model to detect the number of spherical branches of the long-stem grape fern to be detected.

[0048] This application uses a small animal in vivo microCT imager to collect three-dimensional point cloud data of each long-stem grape fern sample. The obtained point cloud data has high resolution and fast scanning speed. The target segmentation model is used to detect the number of spherical branchlets (upright branches) of the long-stem grape fern to be detected. Compared with manual statistics, the accuracy and efficiency of the spherical upright branch statistics are improved.

[0049] In an exemplary embodiment, step 101 specifically includes steps 201 and 202 .

[0050] Step 201: Starting from one end of the sample bed groove, a polyethylene film is used to wrap along the radial contour of the sample bed groove, and the polyethylene film is stretched to form a supporting plane. The supporting plane prevents the long-stem grape fern sample from contacting the bottom of the sample bed groove, thereby obtaining a processed long-stem grape fern sample.

[0051] Step 202: After placing the processed long-stem grape fern algae sample into the small animal in vivo microCT imaging chamber, the operating parameters of the small animal in vivo microCT imaging chamber are set according to preset operating parameter conditions, wherein the preset operating parameter conditions include the voltage and current meeting the set conditions, the field of view being adjusted to the minimum, the camera speed being the set speed, and the camera rotation angle being the set angle.

[0052] Step 203: Acquire the three-dimensional point cloud data obtained by performing CT scanning on the long-stem grape fern algae sample using the small animal living microCT imager.

[0053] The voltage and current meet the set conditions for voltage The current is ≤8W, the set speed is 21° / s, and the set angle is 384°.

[0054] The preset working parameter conditions also include that the diameter of the X-ray scanning area is set to 86 mm and the voxel size is set to 172 μm.

[0055] In an exemplary embodiment, step 104 specifically includes: using a small animal in vivo microCT imager to collect three-dimensional point cloud data of the long-stem grape ferns to be detected; inputting the three-dimensional point cloud data of the long-stem grape ferns to be detected into the target segmentation model to obtain the total number of spherical branches of the long-stem grape ferns to be detected.

[0056] The three-dimensional point cloud data is three-dimensional point cloud data in ply format.

[0057] Each long-stem grape fern sample includes long-stem grape fern samples of multiple accessions. Before measurement using a small animal live microCT imager, relatively complete upright branches of each accession were selected, dried, and wrapped with plastic wrap to prevent the spherical branches from sticking to each other and interfering with each other.

[0058] In an exemplary embodiment, the operation of a small animal in vivo microCT imager includes the following.

[0059] (1) Start the instrument and preheat. After preheating, select the appropriate sample chamber and sample bed, and wrap the middle of the sample bed with plastic wrap to form a support surface, separating the sample from the sample bed groove to facilitate subsequent image processing and data processing. During the operation of the instrument, close and lock the X-ray room door. It is strictly forbidden to open the X-ray room door during operation to avoid X-ray radiation leakage.

[0060] Specifically, the density of the long-stemmed grape fern algae sample and the stage (grooved shape) are similar, and the difference in their linear attenuation coefficient (μ value) for X-rays is minimal, resulting in reduced CT image contrast and difficulty in distinguishing the boundary. The stage is the sample bed. Therefore, the long-stemmed grape fern algae sample and the stage are physically separated. The material selection is: a low-attenuation coefficient material - polyethylene plastic wrap - is used to make the diaphragm plate to reduce its density similarity with the sample. Starting from one end of the groove, the polyethylene plastic wrap is slowly wrapped along the groove contour. During the process, the plastic wrap is appropriately stretched to keep it tightly attached to the groove, forming a support plane. The bracket is then suspended and fixed to avoid direct contact. This separation between the sample and the sample bed groove facilitates subsequent image and data processing. Figure 2 Part A shows the imaging effect of the long-stem grape fern algae sample and the stage before using polyethylene film, and part B shows the imaging effect of the long-stem grape fern algae sample and the stage after using polyethylene film. It can be seen that the polyethylene preservative film can fully separate the sample bed groove and the long-stem grape fern algae sample.

[0061] (2) Place the processed sample in the chamber and close the door.

[0062] (3) Establish data storage path and sample description. Establish database (Database) and create sample description and set parameters. Set voltage 90V and current 88uA in the shooting interface (to ensure voltage Current ≤ 8W is sufficient), adjust the field of view to the minimum (FOV: 86mm; Voxel Size: 172μm), rotate the camera 384°, and adjust the camera speed to 21° / s. FOV represents the field of view, and Voxel Size represents the voxel resolution.

[0063] (4) Data acquisition.

[0064] After receiving the vox file output from the Quantum GX2, a script is used to convert it to the ply format. After receiving the vox file, a specially designed script is used to perform the conversion. This script efficiently converts the vox file to the ply format, facilitating subsequent processing and analysis. The ply format is widely used in 3D modeling and visualization due to its flexibility and wide support. This conversion allows vox data, which was previously difficult to use directly, to be read and manipulated by various 3D processing software, greatly expanding the data's application scope and possibilities.

[0065] Quantum GX2 can process 600 samples per hour, far exceeding the current speed of manual counting.

[0066] Specifically, parameter settings included adjusting the position of the stage and X-ray device based on the width and height of the upright branches of the long-stemmed grape fern. In CT imaging experiments, the long-stemmed grape fern, as a special research specimen, possesses unique physical and chemical properties, posing specific requirements for CT scanning imaging. The algae has an extremely high water content, reaching up to 97%, and during sample preparation, it is difficult to completely remove the water adhering to its upright branches.

[0067] 1. Given the unique absorption properties of water for X-rays, especially low-energy X-rays, its photoelectric absorption effect is significant, resulting in strong attenuation. When performing CT scans of long-stemmed grape fern, given the sample thickness exceeding 1 cm, using an excessively low tube voltage (e.g., less than 70 kV) to excite the X-rays will result in rapid energy attenuation during penetration due to water's high absorption of low-energy X-rays. This will likely result in ineffective penetration, leading to image loss or severe artifacts, compromising accurate observation of the sample's internal structure. Conversely, moderately increasing the tube voltage to a higher level ensures sufficient X-ray energy to penetrate samples containing significant amounts of water and a certain thickness, ensuring complete acquisition of the projection data required for imaging. Furthermore, the morphology of long-stemmed grape fern is unique, consisting of numerous spherical branchlets, often with significant water accumulation between them. A higher tube voltage helps mitigate the negative impact of this water on imaging. By enhancing X-ray penetration and reducing scattering and attenuation caused by the accumulation of water layers between spherical branchlets, image clarity and accuracy are further improved, avoiding image blurring caused by interference from localized water layers. Therefore, to ensure both X-ray penetration of the test specimen and clear imaging, a voltage of 90kV and a current of 88μA are set. The relatively high voltage of 90kV produces X-rays with higher energy and better penetration, while the current of 88μA ensures X-ray intensity. Combined with the voltage, a clear image is obtained at the appropriate dose.

[0068] 2. The X-ray scanning area was set to 86 mm in diameter and the voxel size to 172 μm. The camera was rotated to scan and acquire a 3D model file of the upright branch in VOX format. This took approximately 18 seconds, with a rotation step of 384° and a camera speed of 21° / s. The 3D-Soft filter smoothed the reconstructed image, reducing noise and facilitating better visualization of the upright branch morphology and boundaries. The Cu 0.06+Al 0.5 X-ray filter filtered out low-energy X-rays, optimizing the X-ray energy spectrum and improving image contrast and clarity. This filter also reduced unnecessary radiation dose, minimizing radiation hazards to the animal and operator.

[0069] 3. During the CT scan, to ensure image quality, attention must be paid to specimen fixation and positioning. The upright branches of the long-stemmed grape fern should be securely placed on the stage to prevent movement or deformation during scanning, which could affect the final imaging quality. During positioning, ensure that the center of the upright branch coincides with the center of the X-ray beam to maintain the integrity of the scanning area and image symmetry.

[0070] In an exemplary embodiment, step 102 specifically includes: annotating the boundaries of the spherical branchlets in each three-dimensional point cloud data, and using the bounding box of each spherical branchlet and the total number of spherical branchlets as label data.

[0071] Each sample in the data set includes input data and label data, the input data is three-dimensional point cloud data, and the label data is the bounding box of each spherical branchlet and the total number of spherical branchlets.

[0072] The point cloud segmentation network adopts the PointNet++ network.

[0073] In an exemplary embodiment, the boundary annotation of the spherical branchlets of each three-dimensional point cloud data specifically includes: using CloudCompare software to perform boundary annotation on the spherical branchlets of each three-dimensional point cloud data.

[0074] More specifically, we first segment the 3D point cloud and remove the grooves. Then, we label the 3D point cloud dataset similarly as follows, giving each target (spherical branch) a label such as "1" and other backgrounds 0. Finally, we convert the point cloud data .ply into .txt files one by one and divide them into training and test sets.

[0075] (1) Open the open source software Cloudcompare and click the open button (“Open”) to select the point cloud to be annotated. In the display window, click the left mouse button to drag the point cloud, and right mouse button to drag the view angle to adjust the point cloud to the appropriate view angle for annotation.

[0076] (2) Select the point cloud file to be annotated, such as Figure 3 As shown, use the scissors button to use the polygonal selection method to select and separate the part of the twig that needs to be spherical; Figure 4 As shown in the figure, cancel the previous point cloud selection and select the segmented point cloud; Figure 5 As shown in the figure, enter "1" in the value category option box of the spherical twig part.

[0077] (3) After marking all targets, enter “0” in the value category option box for the remaining upright branch axis (without spherical branchlets).

[0078] (4) Click the merge icon and click "generate a scalar field with the original cloud index" to mark the result. After the adjustment is completed, click the "Save" button to save the result. Figure 6 As shown, save it in ply format.

[0079] In an exemplary embodiment, the three-dimensional point cloud data of the upright branches of the long-stem grape fern are converted into txt format and divided into semantic segmentation training set, validation set and test set in a ratio of 7:2:1, which are used for subsequent training and evaluation of the semantic segmentation model.

[0080] (1) Data export format preparation.

[0081] (2) Dataset structure creation.

[0082] First, write a list of dataset partition files.

[0083] Secondly, the data of this application is rearranged according to the structure of its KITTI dataset.

[0084] Finally, write the dataset loading .yaml file according to the input method of the PointNet++ network.

[0085] (3) Divide the training set, validation set, and test set into a ratio of 7:2:1.

[0086] In an exemplary embodiment, the training environment is set up as follows: (1) Install basic operating software: Pycharm+ software and Anaconda software.

[0087] (2) Install the Graphics Processing Unit (GPU) operating environment: Compute Unified Device Architecture (CUDA) and CUDA Deep Neural Network library (cuDNN).

[0088] (3) Using Pycharm as a platform, we built the Anaconda virtual operating environment, which includes the deep learning library PyTorch and other basic libraries. The Python interpreter in the Anaconda environment uses Interpreter.

[0089] As shown in Table 1, the algorithm running task of this application is executed on a Windows 10 system, with Python version 3.11, PyTorch 1.6.0 framework, CUDA 10.1, and the server uses an Intel (R) Core (TM) i7-9750H CPU @2.60GHz processor and NVIDIA GeForce RTX 2060 graphics card.

[0090] Table 1 Training environment

[0091]

[0092] This application downloads the PointNet++ network code, opens it in Pycharm, and modifies the training file path and dataset type.

[0093] During the training process, the PointNet++ network receives raw 3D point cloud data as input. Then, the network begins to process the data layer by layer. Through a series of transformations and feature extraction operations, it gradually converts the raw 3D point cloud data into higher-level feature representations, learning and identifying points belonging to spherical twigs. After processing by the network, each point is assigned a semantic label, so that it knows which points belong to spherical twigs and which points belong to other parts. Based on the semantic segmentation results of the PointNet++ network, the points belonging to the spherical twigs are extracted, and then the extracted points are clustered and segmented using Euclidean distance.

[0094] Use the PointNet++ network for 3D point cloud object detection. After data preprocessing (converting to a txt file and dividing the dataset), you can perform 3D object detection training and adjust the hyperparameters. After hundreds of rounds of training, the trained PointNet++ network model reaches the optimal effect and obtains the object segmentation model.

[0095] Load the trained pth file and successfully infer and detect the number of sea grapes.

[0096] The verification results for this application show that the trained PointNet++ network achieved a segmentation accuracy of approximately 96.68%, demonstrating that the object segmentation model meets the segmentation requirements of this application. A higher accuracy indicates that the object segmentation model is less likely to make misjudgments. A higher average intersection-over-union (IoU) ratio indicates that the object segmentation model performs well in labeling the spherical branchlets of long-stemmed grape fern. For scenes with multiple similar objects, the model's ability to distinguish and segment each instance can be more accurately evaluated.

[0097] This application innovatively applies small animal non-invasive in vivo imaging technology to algal trait determination. The Quantum GX2 allows sample processing speeds far exceeding current manual counting speeds. Experiments have shown that the 3D imaging function of the Quantum GX2 system can be used to quickly and non-destructively obtain three-dimensional data of upright branches and convert it into point cloud data, thereby enabling the counting of spherical branchlets and the determination of upright branch phenotypes. The Quantum GX2 imaging system uses an extremely low X-ray dose to ensure true long-term microCT imaging, while still providing high-resolution images and faster scanning speeds. This allows for high-throughput counting of spherical branchlets on upright branches, providing an ideal platform for the evaluation of upright branch traits.

[0098] In order to evaluate the accuracy of the object segmentation model after training the PointNet++ network, 20 upright branches were used as samples in this experiment. The number of spherical branches was obtained by manual counting and the object segmentation model. After the CT system completed the scan, the spherical branches of all upright branches were counted manually. The counting method was to use tweezers to remove the spherical branches on the upright branches one by one, as shown in the following figure. Figure 7 The time taken by each of them is recorded separately. The commonly used indicators for evaluating measurement accuracy are the coefficient of determination (R 2 ), mean absolute percentage error (MAPE), and root mean square error (RMSE).

[0099] The R value of the manual value and the model calculation value of the spherical branchlet number statistics of Miqiu No. 1 2The value of R² is 0.9276. The closer the R² value is to 1, the stronger the target segmentation model's ability to explain the data and the better the fitting effect. The root mean square error (RMSE) and mean absolute error (MAPE) are important indicators for evaluating the prediction error of the target segmentation model. In this experiment, the RMSE is 4.23 and the MAPE is 6.37%. It can be seen that the target segmentation model of the dense ball No. 1 has a good fitting effect.

[0100] Based on the same inventive concept, embodiments of the present application also provide a device for measuring the number of spherical branchlets of long-stem grape ferns, used to implement the aforementioned method for measuring the number of spherical branchlets of long-stem grape ferns. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more embodiments of the device for measuring the number of spherical branchlets of long-stem grape ferns provided below can be found in the aforementioned method for measuring the number of spherical branchlets of long-stem grape ferns, and will not be repeated here.

[0101] In an exemplary embodiment, Figure 8 As shown, a device for measuring the number of spherical branchlets of long-stem grape ferns is provided. The device for measuring the number of spherical branchlets of long-stem grape ferns applies any of the methods for measuring the number of spherical branchlets of long-stem grape ferns, and the device for measuring the number of spherical branchlets of long-stem grape ferns includes the following modules.

[0102] The three-dimensional point cloud data acquisition module is used to acquire three-dimensional point cloud data of each long-stem grape fern algae sample using a small animal in vivo microCT imager.

[0103] The dataset construction module is used to construct a dataset by annotating spherical twigs on each three-dimensional point cloud data.

[0104] The model training module is used to train the point cloud segmentation network using the data set to obtain the target segmentation model.

[0105] The detection module is used to detect the number of spherical branches of the long-stem grape fern to be detected by using the target segmentation model.

[0106] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0107] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for measuring the number of spherical branchlets of long-stem grape fern, characterized in that: The method for measuring the number of spherical branchlets of the long-stem grape fern comprises: Small animal in vivo microCT imaging was used to collect three-dimensional point cloud data of each long-stem grape fern algae sample; The dataset is constructed by annotating spherical twigs on each 3D point cloud data; Use the dataset to train the point cloud segmentation network to obtain the target segmentation model; The target segmentation model is used to detect the number of spherical branches of the long-stem grape fern to be detected; The three-dimensional point cloud data of each long-stem grape fern algae sample was collected using a small animal in vivo microCT imager, including: Starting from one end of the sample bed groove, a polyethylene film is wrapped along the radial contour of the sample bed groove, and the polyethylene film is stretched to form a support plane, wherein the support plane prevents the long-stem grape fern algae sample from contacting the bottom of the sample bed groove, thereby obtaining a processed long-stem grape fern algae sample; After placing the processed long-stem grape fern algae sample into the small animal in vivo microCT imaging chamber, setting the operating parameters of the small animal in vivo microCT imaging chamber according to preset operating parameter conditions; the preset operating parameter conditions include that the voltage and current meet the set conditions, the field of view is adjusted to the minimum, the camera speed is the set speed, and the camera rotation angle is the set angle; The three-dimensional point cloud data is obtained after the small animal living microCT imager performs CT scanning on the long-stem grape fern algae sample.

2. The method for measuring the number of spherical branchlets of long-stem grape fern according to claim 1, characterized in that: The voltage and current meet the set conditions for voltage The current is ≤8W, the set speed is 21° / s, and the set angle is 384°.

3. The method for measuring the number of spherical branchlets of long-stem grape fern according to claim 1, characterized in that: The preset working parameter conditions also include that the diameter of the X-ray scanning area is set to 86 mm and the voxel size is set to 172 μm.

4. The method for measuring the number of spherical branchlets of long-stem grape fern according to claim 1, characterized in that: The point cloud segmentation network adopts the PointNet++ network.

5. The method for measuring the number of spherical branchlets of long-stem grape fern according to claim 1, characterized in that: The target segmentation model is used to detect the number of spherical branches of the long-stem grape fern to be detected, specifically including: Small animal in vivo microCT imaging was used to collect three-dimensional point cloud data of the long-stemmed grape fern to be tested; The three-dimensional point cloud data of the long-stem grape cauliflower to be detected is input into the target segmentation model to obtain the total number of spherical branches of the long-stem grape cauliflower to be detected.

6. The method for measuring the number of spherical branchlets of long-stem grape fern according to claim 1, characterized in that: The three-dimensional point cloud data is three-dimensional point cloud data in ply format.

7. The method for measuring the number of spherical branchlets of long-stem grape fern according to claim 1, characterized in that: The dataset is constructed by annotating spherical twigs on each 3D point cloud data, specifically including: The boundaries of the spherical branches in each 3D point cloud data are annotated, and the bounding box of each spherical branch and the total number of spherical branches are used as label data.

8. The method for measuring the number of spherical branchlets of long-stem grape fern according to claim 7, characterized in that: The boundary annotation of the spherical branches of each 3D point cloud data includes: CloudCompare software was used to annotate the boundaries of the spherical branches in each 3D point cloud data.

9. A device for measuring the number of spherical branches of long-stem grape fern, characterized in that: The device for measuring the number of spherical branchlets of the long-stem grape ferns adopts the method for measuring the number of spherical branchlets of the long-stem grape ferns according to any one of claims 1 to 8, and the device for measuring the number of spherical branchlets of the long-stem grape ferns comprises: A three-dimensional point cloud data acquisition module is used to acquire three-dimensional point cloud data of each long-stem grape fern algae sample using a small animal in vivo microCT imager; A data set construction module, used for constructing a data set by annotating spherical twigs on each three-dimensional point cloud data; The model training module is used to train the point cloud segmentation network using the data set to obtain the target segmentation model; The detection module is used to detect the number of spherical branches of the long-stem grape fern to be detected by using the target segmentation model.

Images