Root system dynamic analysis method and observation device

Abstract

The invention discloses a root system dynamic analysis method and an observation device, and relates to the field of crop root system analysis. The method comprises the following steps: planting to-be-analyzed seeds in a culture medium of a culture bottle, culturing until the seeds are white, acquiring image data of the culture bottle at multiple shooting angles by utilizing a shooting device at a shooting frequency at a fixed time interval lambda, acquiring multi-view root system image data of the to-be-analyzed seeds with angle information, and analyzing the to-be-analyzed seeds by utilizing each acquired root system image. The method comprises the following steps: constructing a three-dimensional model of a to-be-analyzed seed root system, exporting the three-dimensional model in a Las format, reading an LAS file by using a point cloud processing library Open3D and obtaining point cloud data in the LAS file, carrying out denoising and filtering processing on multi-point cloud data, segmenting a root system region, extracting a skeleton of point cloud by using a three-dimensional skeleton extraction algorithm, obtaining a center line of the root system region, and carrying out root system analysis. Simplifying the point cloud into a skeleton representing a root system trunk and branches, and extracting root system morphological features in the skeleton representing the root system trunk and branches;

Description

Technical Field

[0001] The invention relates to the field of crop root system analysis, and in particular to a root system dynamic analysis method and an observation device. Background Art

[0002] The root system is a key organ for crop growth and development. Studying the root system architecture and root system development under different conditions is of great significance to the study of variety characteristics, resistance mechanisms and related supporting technologies. Since the root system grows in the soil, it is impossible to directly observe and analyze it. Therefore, how to observe and analyze root system development has always been the focus and difficulty of research.

[0003] The commonly used root observation and analysis techniques include excavation, root column, root box, hydroponic method, etc., but all of them have certain problems. The excavation method is the most economical and practical method, but the excavation method is time-consuming and labor-intensive, and it is easy to damage the root system and cannot obtain a complete root system; the root column method and the root box method are common root observation methods, but the technology is costly and complex to install, and it is not possible to observe and analyze the root system easily and quickly; although the hydroponic method can be used to observe and analyze the root system easily and quickly, it cannot fix the morphology of the root system and cannot meet the research on root configuration, root angle, etc. In the research, there is an urgent need for a device that can observe the root system easily and quickly and the corresponding analysis method. To this end, we propose a root system dynamic analysis method and observation device. Summary of the invention

[0004] The main purpose of the present invention is to provide a root system dynamic analysis method and observation device, which can effectively solve the problems in the background technology.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] A root system dynamic analysis method, comprising:

[0007] A culture matrix is ​​arranged in a culture bottle, and seeds to be analyzed are planted in the culture matrix of the culture bottle, and the bottle is placed in an incubator for cultivation until the seeds to be analyzed turn white, and a camera is used to obtain a plurality of image data of the culture bottle at multiple shooting angles at a fixed time interval λ;

[0008] Preprocessing the acquired image data to obtain multi-view root system image data of the seeds to be analyzed with angle information;

[0009] Using the acquired root system images, construct a three-dimensional model of the root system of the seed to be analyzed, and export the three-dimensional model in Las format;

[0010] Use the point cloud processing library Open3D to read the LAS file and obtain the point cloud data therein, perform denoising and filtering on the point cloud data, segment the root area, use the three-dimensional skeleton extraction algorithm to extract the skeleton of the point cloud, obtain the center line of the root area, simplify the point cloud into a skeleton representing the root trunk and branches, and extract the root morphological characteristics in the skeleton representing the root trunk and branches.

[0011] The method further comprises:

[0012] A root system evaluation model is constructed, and according to the obtained root system morphological characteristics, the root system evaluation model is used to calculate the growth index Gr of the seed root to be analyzed, wherein the expression of the root system evaluation model is: Gr = αGb + βGq; wherein Gb is the basic growth evaluation index; Gq is the growth trend evaluation index; α and β are weight coefficients, and α and β∈(0,1), α+β=1;

[0013] The calculation formula of the basic growth evaluation index Gb is: In the formula, Mf i It is expressed as the root morphological characteristic value of the i-th item; It is expressed as the coefficient of the i-th root morphological characteristic; N represents the type of root morphological characteristics;

[0014] The calculation formula of the growth trend evaluation index Gq is: In the formula, Mf it Mf is the root morphological characteristic value of the ith item at the tth shooting; i(t+1) It is expressed as the i-th root morphological characteristic value at the t+1-th shooting; T is the observation period.

[0015] The process of acquiring multi-view root system image data includes the following steps:

[0016] Step S21: using the camera to obtain image data of the plant root system in the culture bottle at multiple angles, and then performing image preprocessing by denoising, contrast enhancement and reflection elimination;

[0017] Step S22: extracting multi-view root system image data of the seeds to be analyzed in the culture bottle by using any image segmentation algorithm such as color segmentation, threshold segmentation or GrabCut algorithm.

[0018] The process of constructing a three-dimensional model of the seed root system to be analyzed includes the following steps:

[0019] Step S31: extracting local feature points from each of the root system images, and calculating feature descriptors for image matching and point cloud generation for each of the local feature points through descriptor calculation;

[0020] Step S32: performing feature point matching on images taken from different viewing angles, finding the local feature points representing the features of the same object in different images, identifying and removing erroneous matching point pairs, and ensuring that the matching results meet the multi-view geometric consistency;

[0021] Step S33: using the matched feature points and the motion recovery structure to derive the camera device external parameters and camera device internal parameters corresponding to each image, estimate the camera device parameters, calculate the camera device posture, and use the matching relationship of the feature points to perform triangulation to generate a sparse three-dimensional point cloud; wherein the camera device external parameters include a rotation matrix and a displacement vector; and the camera device internal parameters include a focal length and a distortion parameter;

[0022] Step S34: by performing in-depth calculations on the feature points of the objects in the adjacent images, the depth information of the three-dimensional points in the point cloud is derived using a voxel-based method, and inserted into the sparse point cloud to form a dense point cloud;

[0023] Step S35: The dense point cloud is converted into a triangular mesh structure through the Poisson surface reconstruction algorithm, and the noise and redundant data are eliminated through smoothing, simplification and denoising, and the existing holes are filled to make the model surface smoother and more continuous, thereby obtaining an optimized three-dimensional mesh model;

[0024] Step S36: Project the initial root system images back onto the constructed three-dimensional grid model, generate texture using the color information of each root system image, and integrate image information from different angles through texture fusion technology to generate seamless high-resolution texture.

[0025] The extraction process of root morphological features in the skeleton representing the root trunk and branches includes the following steps:

[0026] Step S41: Use the point cloud processing library Open3D to read the LAS file and obtain the point cloud data therein, and then use the filter to perform denoising and filtering, while retaining the set features to remove isolated points and low-quality points, and retain high-quality point cloud data;

[0027] Step S42: performing cluster analysis on the point cloud, identifying the point set belonging to the root system, separating the root system from other objects, and segmenting the root system area;

[0028] Step S43: extract the skeleton of the point cloud, obtain the center line of the root system, use the three-dimensional skeleton extraction algorithm to simplify the point cloud into a skeleton representing the trunk and branches of the root system, and extract the root morphological characteristics in the skeleton representing the trunk and branches of the root system.

[0029] The root morphological characteristics include: root length, number of branch points, branching order, root volume, and root density.

[0030] The root length is approximately calculated by using the skeleton line segments of the skeleton to calculate the length of each branch, and the length is accumulated and obtained;

[0031] The branch point is obtained by calculating the number of connections of each root node and finding the branch point and the leaf node;

[0032] The branching level is determined by using the connection relationship of the branching points to determine the branching level of the root system and analyzing the topological structure of the root system to obtain;

[0033] The root system volume is obtained by measuring the total volume of the root system using a point cloud volume estimation algorithm;

[0034] The root density is estimated by volume and point density.

[0035] A root system dynamic observation device, comprising:

[0036] A culture bottle for cultivating the seeds to be analyzed, wherein the culture bottle contains a culture matrix;

[0037] A turntable for placing the culture bottle, wherein the turntable rotates in a circle with the center line of the culture bottle as the axis;

[0038] A photographing device for photographing the culture bottle, wherein the photographing device acquires a plurality of image data of the culture bottle at multiple shooting angles at a shooting frequency of a fixed time interval λ; and the photographing device performs a photographing operation every time the turntable rotates 6 degrees;

[0039] At least one fill light for filling light for the culture bottle during shooting, and the fill lights are evenly distributed on the outside of the culture bottle;

[0040] An image processing module for preprocessing the acquired image data to obtain multi-view root system image data of the seeds to be analyzed with angle information;

[0041] A three-dimensional model construction module for constructing a three-dimensional model of the root system of the seed to be analyzed using the acquired root system images, and exporting the three-dimensional model in Las format;

[0042] A morphological feature extraction module for extracting root morphological features in a skeleton representing a root trunk and branches, wherein the morphological feature extraction module reads a LAS file using a point cloud processing library Open3D and obtains point cloud data therein, performs denoising and filtering on the point cloud data, segments the root area, extracts the skeleton of the point cloud using a three-dimensional skeleton extraction algorithm, obtains the center line of the root area, simplifies the point cloud into a skeleton representing a root trunk and branches, and thereby extracts the root morphological features in the skeleton representing the root trunk and branches;

[0043] A growth status assessment module for assessing the growth status of the root system of the seed to be analyzed, wherein the growth status assessment module constructs a root system assessment model, and based on the acquired root morphological characteristics, utilizes the root system assessment model to obtain a growth index of the root system of the seed to be analyzed for characterizing the growth status of the root system.

[0044] The system further includes a memory, a processor, and an electronic program stored in the memory and capable of running on the processor.

[0045] The present invention has the following beneficial effects:

[0046] Traditional digging method, root column method, root box method, hydroponic method, etc. all have certain problems. For example, the simple hydroponic method cannot fix the root morphology, the field planting digging method is time-consuming and labor-intensive, and the root system is easily damaged, and the root column and root box method is costly and complex to install. Compared with the prior art, the technical solution of the present invention can fix the root morphology and observe it at any time by using transparent culture medium for cultivation, and it is simple, low-cost and highly operable.

[0047] Traditional root analysis methods are mostly to take samples and then scan, and then obtain root-related morphological parameters. The technical solution of the present invention adopts the method of taking pictures, and in order to eliminate the problems of mirror images and reflections when taking pictures with glassware, it adopts fill light on both sides and opposite sides and 360-degree photography, so that the images obtained by the camera can minimize light reflection and observe the plant root system completely and clearly. The root morphology and development dynamics can be observed in real time, without loss and without sampling.

[0048] Compared with the prior art, the technical solution of the present invention can prepare the culture matrix according to the research needs. On this basis, according to the research purpose, an appropriate amount of exogenous substances can be added to simulate the stress of abiotic factors. It has various forms, a wide range of applications, and is simple and easy to implement.

[0049] Compared with the prior art, the technical solution of the present invention constructs a three-dimensional model of the root system based on analysis and processing after taking pictures, and obtains the dynamic change process of the plant root system through multi-time series graphics acquisition and analysis.

[0050] Compared with the prior art, the observation device proposed by the present invention has low cost, simple structure, good applicability, convenient use, and batch operation. The root system configuration obtained by this method is fixed, the root system is clearly visible, and the root system can be observed and analyzed quickly and non-destructively. Through the combination of two-dimensional and three-dimensional, the development and structural characteristics of the root system can be obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A flow chart of a root system dynamic analysis method of the present invention;

[0052] Figure 2 The figure is a schematic structural diagram of a root system dynamic observation device of the present invention. DETAILED DESCRIPTION

[0053] The present invention will be further described below in conjunction with specific implementation methods, wherein the accompanying drawings are only used for exemplary descriptions and represent only schematic diagrams rather than actual drawings, and should not be understood as limiting the present invention. In order to better illustrate the specific implementation methods of the present invention, some parts of the accompanying drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product.

[0054] The specific implementation process of the technical solution of the present invention includes the following steps:

[0055] Step 1: After the culture matrix is ​​prepared, pour it into a cylindrical culture bottle and plant the seeds in the culture matrix.

[0056] Step 2: Place the culture bottle with the seeds in the incubator for cultivation and observe regularly. When the seeds turn white, start taking pictures.

[0057] Step 3: Place the culture bottle on the turntable, set the turntable rotation angle and photo shooting time, and turn on the fill light, turntable and camera. Take a photo every time the turntable rotates 6 degrees. In this embodiment, the number of fill lights is 3, and their distribution is as follows: Figure 2 As shown, the rotation direction of the turntable can be along Figure 2 The rotation direction 1 or the rotation direction 2 in the embodiment is periodically rotated.

[0058] Step 4: Image Processing

[0059] Step 41: Image Preprocessing

[0060] The developed camera is used to obtain image data of plant roots in culture bottles from multiple angles, and then image preprocessing is performed by denoising, contrast enhancement, and reflection elimination.

[0061] Denoising: The transparency of the glass container can cause reflection and refraction of light, which can introduce noise. Use Gaussian filtering to remove noise.

[0062] Contrast enhancement: Due to the transparency of glass, the edges of objects will be blurred. Adaptive Histogram Equalization (CLAHE) is used to enhance the contrast of the image and improve the difference between the object and the background.

[0063] Remove reflections: Light reflections from glass surfaces are one of the main distractions. Use a polarizing filter to reduce reflections during the shot, or use a reflection removal algorithm to post-process the image.

[0064] Step 42: Background separation and object detection

[0065] The plant root images in the container are extracted through color segmentation, threshold segmentation and GrabCut algorithm.

[0066] Color segmentation: Because the color of the object in the container is significantly different from the glass container and the background, the object can be segmented through color space conversion (such as RGB to HSV, Lab, etc.).

[0067] Threshold-based segmentation: By setting a suitable threshold value for the grayscale image, the glass container and the internal objects can be distinguished. The adaptive threshold method selects the threshold value based on the brightness change of the local area, which is suitable for dealing with uneven lighting conditions.

[0068] Using the GrabCut algorithm, by defining the outside of the glass container as the background and the inside as the foreground, the plant roots in the container can be effectively extracted.

[0069] Through the above process, multi-view plant root image data without background and with angle information is obtained.

[0070] Step 5: 3D reconstruction

[0071] Step 51: Feature point extraction

[0072] The key local feature points are extracted from each image using an image feature extraction algorithm. Through descriptor calculation, feature descriptors are calculated for each feature point for subsequent image matching and point cloud generation.

[0073] Step 52: Feature point matching

[0074] The images taken from different perspectives are matched with feature points, and the Brute-Force matching algorithm is used to find the points representing the features of the same object in different images. Then the geometrically constrained RANSAC algorithm is used to identify and remove incorrect matching point pairs to ensure that the matching results meet the multi-perspective geometric consistency.

[0075] Step 53: Camera pose estimation

[0076] Through the matched feature points, the camera parameters are estimated by deriving the camera extrinsic parameters (rotation matrix and displacement vector) and intrinsic parameters (focal length, distortion parameters, etc.) corresponding to each image using motion recovery structure. After calculating the camera's pose, the matching relationship of the feature points can be used for triangulation to generate a sparse 3D point cloud.

[0077] Step 54: Dense reconstruction

[0078] By deeply calculating the feature points of objects in adjacent images, the depth information of more 3D points is derived using a voxel-based method. More 3D points are inserted into the sparse point cloud to form a dense point cloud.

[0079] Step 55: Mesh reconstruction

[0080] The dense point cloud is converted into a triangular mesh structure through the Poisson surface reconstruction algorithm. The mesh model is further optimized through smoothing, simplification and denoising. Noise and redundant data are eliminated, possible holes are filled, and the model surface is made smoother and more continuous.

[0081] Step 56: Texture mapping

[0082] The initial image is projected back onto the 3D mesh model, and the color information of each image is used to generate textures. Through texture fusion technology, image information from different angles is integrated to generate seamless high-resolution textures.

[0083] Step 57: Export the 3D model

[0084] Finally, the three-dimensional model is exported in Las format for the next calculation.

[0085] Step 6: Automatic extraction of root morphology

[0086] Step 61: Loading LAS files and preprocessing

[0087] Use the point cloud processing library Open3D to read the LAS file and obtain the point cloud data therein, and then use the filter to denoise and filter, while retaining the set features, to remove isolated points and low-quality points and retain high-quality point cloud data.

[0088] Step 62: Root Area Segmentation

[0089] The root area is segmented by clustering algorithm and deep learning-based segmentation method. The DBSCAN clustering algorithm is used to perform cluster analysis on the point cloud to identify the point set belonging to the root system and separate the root system from other objects.

[0090] Step 63: Skeleton extraction

[0091] By extracting the skeleton of the point cloud, the center line of the root system is obtained to analyze the length, branch structure and other characteristics. Using the 3D skeleton extraction algorithm, the point cloud is simplified into a skeleton representing the trunk and branches of the root system.

[0092] Step 64: Root morphological feature extraction

[0093] Root length: The length of each branch is approximately calculated using the skeleton segments of the skeleton, and the total length of the root system is accumulated.

[0094] Branching points: Count the number of connections of each root node and find branching points and leaf nodes.

[0095] Branching level: Use the connection relationship of branch points to determine the branching level of the root system and analyze the topological structure of the root system.

[0096] Root volume: The total volume of the root system is measured using a point cloud volume estimation algorithm.

[0097] Root Density: Estimate root density distribution by volume and point density.

[0098] Step 7: construct a root system evaluation model, and use the root system evaluation model to calculate the growth index Gr of the seed root to be analyzed according to the acquired root morphological characteristics, wherein the expression of the root system evaluation model is: Gr = αGb + βGq; wherein Gb is the basic growth evaluation index; Gq is the growth trend evaluation index; α and β are weight coefficients, and α and β∈(0,1), α+β=1;

[0099] The calculation formula of the basic growth evaluation index Gb is: In the formula, Mf i It is expressed as the root morphological characteristic value of the i-th item; It is expressed as the coefficient of the i-th root morphological characteristic; N represents the type of root morphological characteristics;

[0100] The calculation formula of the growth trend evaluation index Gq is: In the formula, Mf it Mf is the root morphological characteristic value of the ith item at the tth shooting; i(t+1) It is expressed as the i-th root morphological characteristic value at the t+1-th shooting; T is the observation period.

[0101] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A root system dynamic analysis method, characterized in that: include: A culture matrix is ​​arranged in a culture bottle, and seeds to be analyzed are planted in the culture matrix of the culture bottle, and the bottle is placed in an incubator for cultivation until the seeds to be analyzed turn white, and a camera is used to obtain a plurality of image data of the culture bottle at multiple shooting angles at a fixed time interval λ; Preprocessing the acquired image data to obtain multi-view root system image data of the seeds to be analyzed with angle information; Using the acquired root system images, construct a three-dimensional model of the root system of the seed to be analyzed, and export the three-dimensional model in Las format; Use the point cloud processing library Open3D to read the LAS file and obtain the point cloud data therein, perform denoising and filtering on the point cloud data, segment the root area, use the three-dimensional skeleton extraction algorithm to extract the skeleton of the point cloud, obtain the center line of the root area, simplify the point cloud into a skeleton representing the root trunk and branches, and extract the root morphological characteristics in the skeleton representing the root trunk and branches.

2. A root system dynamic analysis method according to claim 1, characterized in that: The method further comprises: A root system evaluation model is constructed, and according to the obtained root system morphological characteristics, the root system evaluation model is used to calculate the growth index Gr of the seed root to be analyzed, wherein the expression of the root system evaluation model is: Gr = αGb + βGq; wherein Gb is the basic growth evaluation index; Gq is the growth trend evaluation index; α and β are weight coefficients, and α and β∈(0,1), α+β=1; The calculation formula of the basic growth evaluation index Gb is: In the formula, Mf i It is expressed as the root morphological characteristic value of the i-th item; It is expressed as the coefficient of the i-th root morphological characteristic; N represents the type of root morphological characteristics; The calculation formula of the growth trend evaluation index Gq is: In the formula, Mf it Mf is the root morphological characteristic value of the ith item at the tth shooting; i(t+1) It is expressed as the i-th root morphological characteristic value at the t+1-th shooting; T is the observation period.

3. A root system dynamic analysis method according to claim 1, characterized in that: The process of acquiring multi-view root system image data includes the following steps: Step S21: using the camera to obtain image data of the plant root system in the culture bottle at multiple angles, and then performing image preprocessing by denoising, contrast enhancement and reflection elimination; Step S22: extracting multi-view root system image data of the seeds to be analyzed in the culture bottle by using any image segmentation algorithm such as color segmentation, threshold segmentation or GrabCut algorithm.

4. A root system dynamic analysis method according to claim 1, characterized in that: The process of constructing a three-dimensional model of the seed root system to be analyzed includes the following steps: Step S31: extracting local feature points from each of the root system images, and calculating feature descriptors for image matching and point cloud generation for each of the local feature points through descriptor calculation; Step S32: performing feature point matching on images taken from different viewing angles, finding the local feature points representing the features of the same object in different images, identifying and removing erroneous matching point pairs, and ensuring that the matching results meet the multi-view geometric consistency; Step S33: using the matched feature points and the motion recovery structure to derive the camera device external parameters and camera device internal parameters corresponding to each image, estimate the camera device parameters, calculate the camera device posture, and use the matching relationship of the feature points to perform triangulation to generate a sparse three-dimensional point cloud; wherein the camera device external parameters include a rotation matrix and a displacement vector; and the camera device internal parameters include a focal length and a distortion parameter; Step S34: by performing in-depth calculations on the feature points of the objects in the adjacent images, the depth information of the three-dimensional points in the point cloud is derived using a voxel-based method, and inserted into the sparse point cloud to form a dense point cloud; Step S35: The dense point cloud is converted into a triangular mesh structure through the Poisson surface reconstruction algorithm, and the noise and redundant data are eliminated through smoothing, simplification and denoising, and the existing holes are filled to make the model surface smoother and more continuous, thereby obtaining an optimized three-dimensional mesh model; Step S36: Project the initial root system images back onto the constructed three-dimensional grid model, generate texture using the color information of each root system image, and integrate image information from different angles through texture fusion technology to generate seamless high-resolution texture.

5. A root system dynamic analysis method according to claim 1, characterized in that: The extraction process of root morphological features in the skeleton representing the root trunk and branches includes the following steps: Step S41: Use the point cloud processing library Open3D to read the LAS file and obtain the point cloud data therein, and then use the filter to perform denoising and filtering, while retaining the set features to remove isolated points and low-quality points, and retain high-quality point cloud data; Step S42: performing cluster analysis on the point cloud, identifying the point set belonging to the root system, separating the root system from other objects, and segmenting the root system area; Step S43: extract the skeleton of the point cloud, obtain the center line of the root system, use the three-dimensional skeleton extraction algorithm to simplify the point cloud into a skeleton representing the trunk and branches of the root system, and extract the root morphological characteristics in the skeleton representing the trunk and branches of the root system.

6. A root system dynamic analysis method according to any one of claims 1 or 5, characterized in that: The root morphological characteristics include: root length, number of branch points, branching order, root volume, and root density.

7. A root system dynamic analysis method according to claim 6, characterized in that: The root length is approximately calculated by using the skeleton line segments of the skeleton to calculate the length of each branch, and the length is accumulated and obtained; The branch point is obtained by calculating the number of connections of each root node and finding the branch point and the leaf node; The branching level is determined by using the connection relationship of the branching points to determine the branching level of the root system and analyzing the topological structure of the root system to obtain; The root system volume is obtained by measuring the total volume of the root system using a point cloud volume estimation algorithm; The root density is estimated by volume and point density.

8. A root system dynamic observation device, characterized in that: The observation device is used to implement the steps of a root system dynamic analysis method according to any one of claims 1 to 7, including: A culture bottle for cultivating the seeds to be analyzed, wherein the culture bottle contains a culture matrix; A turntable for placing the culture bottle, wherein the turntable rotates in a circle with the center line of the culture bottle as the axis; A photographing device for photographing the culture bottle, wherein the photographing device acquires a plurality of image data of the culture bottle at multiple shooting angles at a shooting frequency of a fixed time interval λ; and the photographing device performs a photographing operation every time the turntable rotates 6 degrees; At least one fill light for filling light for the culture bottle during shooting, and the fill lights are evenly distributed on the outside of the culture bottle; An image processing module for preprocessing the acquired image data to obtain multi-view root system image data of the seeds to be analyzed with angle information; A three-dimensional model construction module for constructing a three-dimensional model of the root system of the seed to be analyzed using the acquired root system images, and exporting the three-dimensional model in Las format; A morphological feature extraction module for extracting root morphological features in a skeleton representing a root trunk and branches, wherein the morphological feature extraction module reads a LAS file using a point cloud processing library Open3D and obtains point cloud data therein, performs denoising and filtering on the point cloud data, segments the root area, extracts the skeleton of the point cloud using a three-dimensional skeleton extraction algorithm, obtains the center line of the root area, simplifies the point cloud into a skeleton representing a root trunk and branches, and thereby extracts the root morphological features in the skeleton representing the root trunk and branches; A growth status assessment module for assessing the growth status of the root system of the seed to be analyzed, wherein the growth status assessment module constructs a root system assessment model, and based on the acquired root morphological characteristics, utilizes the root system assessment model to obtain a growth index of the root system of the seed to be analyzed for characterizing the growth status of the root system.

9. A root system dynamic observation device according to claim 8, characterized in that: The system also includes a memory, a processor, and an electronic program stored in the memory and capable of running on the processor, wherein the processor can implement the steps of a root system dynamic analysis method as described in any one of claims 1-7 when running the electronic program.

Images