Crop root system 3D microscopic phenotype detection system and detection method

Abstract

The invention discloses a crop root system 3D microscopic phenotype detection system and detection method. The electric three-dimensional module is adopted to drive the high-resolution microscopic imaging system to move, and super-large-range and multi-angle crop root microscopic imaging is achieved. Through combination of depth-of-field synthesis, image depth estimation and an automatic focusing technology, various root system microscopic phenotypes such as root hair and root epidermis cell morphology can be rapidly obtained. The device adopts a mode of combining synchronous dark field light supplement with other various lighting, and enhances the microstructure characteristics of the root system, so as to realize high-quality microscopic imaging of the crop root system. The result shows that the root system microscopic image obtained by using the system and the method is high in quality. Compared with the prior art, the invention provides a rapid, efficient and low-cost crop root system microscopic phenotype detection technology. According to the system, microscopic imaging can be carried out on a super-large-range area under the condition that a sample is fixed, the crop root system microscopic phenotype collection efficiency can be remarkably improved, and the cost is reduced.

Description

Technical Field

[0001] The present invention belongs to the fields of smart agricultural equipment technology, computer vision technology and biometric identification technology, and specifically relates to a crop root 3D microscopic phenotype detection system and detection method. Background Art

[0002] Plant roots are key organs for crops to absorb water and nutrients. The analysis of their microstructural characteristics is of great significance for revealing plant physiological mechanisms and guiding precision breeding. Traditional research methods such as excavation and soil core drilling require destructive sampling, which not only makes it difficult to achieve in-situ dynamic observations, but also fails to meet the needs of modern plant phenomics for high-throughput and high-precision detection. With the development of non-destructive imaging technology, although the relevant technical system has made certain breakthroughs, there are still significant technical bottlenecks in dimensions such as microscopic phenotypic analysis and dynamic monitoring.

[0003] In terms of spatial resolution, existing three-dimensional imaging technology is difficult to balance the detection requirements of macroscopic scales and microscopic features. A typical example is the multi-view image reconstruction technology proposed in patent CN104897575A. Although it realizes submillimeter-level observation through the coordinated scanning of a high-precision electrically controlled rotating table and a vertical lifting table, due to the limitation of the fixed objective lens design, the minimum imaging unit of its optical path magnification system is only 50μm, which cannot effectively capture the root hair structure with a diameter of 5-15μm and more subtle epidermal cell morphology. This directly leads to data missing in the existing three-dimensional model when characterizing key phenotypic parameters such as root hair density and meristem cell arrangement.

[0004] In terms of dynamic monitoring capabilities, the current system architecture generally lacks coordination between the mechanical structure and the imaging module. Taking the multimodal 3D reconstruction system of patent CN107392956B as an example, the mechanical arm rotation acquisition mode it adopts requires manual intervention, and the scanning of a single plant takes more than 30 minutes. In addition, the microscopic imaging module is not integrated, which makes it impossible to achieve continuous observation of the growth process of living roots. This dual limitation of temporal resolution and spatial resolution has seriously restricted the research on dynamic biological processes such as plant stress response.

[0005] The environmental adaptability challenge is concentrated on the elimination of optical interference in complex media. Although patent CN119437076A improves transparent medium imaging by combining colloidal culture with an optical calibration tank, its three-dimensional reconstruction algorithm based on silhouette images is easily interfered by impurity particles in real soil environments, resulting in distortion in the extraction of root edge features. More importantly, the existing optical system lacks a refractive index correction mechanism for microscopic imaging, making it difficult to penetrate heterogeneous media while maintaining high resolution, affecting the accurate measurement of biomechanical characteristics such as epidermal cell wall thickness.

[0006] From the perspective of technology application, existing solutions have a significant contradiction between efficiency and cost. Although equipment such as X-ray CT can achieve non-destructive testing of the root system of the entire plant, the cost of a single micro-CT scan exceeds 10,000 yuan, and a single plant scan takes up to 2-3 hours. This high-cost and low-efficiency feature makes it difficult to support the sample throughput required for large-scale plant phenomics research, which seriously restricts the effective transformation of technological achievements into breeding practice.

[0007] The combined effect of the above technical bottlenecks has resulted in the existing root analysis system still being unable to meet the urgent needs of smart agriculture for digital analysis of crop root phenotypes in terms of key indicators such as microscopic phenotype analysis accuracy, dynamic process capture capability, adaptability to complex environments, and large-scale detection efficiency. Summary of the invention

[0008] 1. Technical issues to be resolved Existing microscopic imaging equipment is limited by its fixed optical path structure and fixed light source, which makes its imaging area small and fixed, and cannot be used for large-scale non-destructive imaging of crop roots. The purpose of the present invention is to address the problems existing in the existing crop root microscopic phenotype observation technology and means, and propose an efficient crop root microscopic phenotype extraction scheme to achieve real-time, non-destructive collection and analysis of crop root microscopic phenotypes, such as root hairs, root epidermal cells, and rhizosphere microorganisms.

[0009] (II) Technical solution In order to solve the above problems, the present invention provides the following technical solutions, and proposes a crop root 3D microscopic phenotype detection system and detection method, as follows.

[0010] A crop root 3D microscopic phenotype detection system is used to obtain the microscopic phenotype of the crop root system. The detection system comprises: An optical platform (1), placed horizontally, for accommodating other components of the detection system; The three-dimensional module (2) is fixed on the optical platform, and is controlled by a servo motor to drive the microscopic imaging system (3) to move in three dimensions, thereby achieving microscopic automatic focusing and multi-angle shooting of the root system of the crop to be tested; A microscopic imaging system (3) is installed on the three-dimensional module (2), the microscopic imaging system comprising an industrial camera (3-1) and a microscope lens (3-2) capable of automatically changing magnification, and is used for performing microscopic imaging of crop roots; A rotating platform (4) is fixed on the optical platform (1), located on the right side of the three-dimensional module (2), and is used as a platform for placing the plant culture box (5); the rotating platform is controlled by a stepping motor and can drive the plant culture box to rotate in both directions; The plant cultivation box (5) is rectangular in shape as a whole, has a square bottom surface, and is made of highly transparent material; the plant cultivation box is composed of an outer box (5-1) and an inner box (5-2), wherein a fixed gap (5-3) is provided between the inner wall of the outer box and the outer wall of the inner box, forming four narrow flat surface spaces, wherein a nutrient solution supporting plant growth is contained inside, and a crop plant is planted in each flat surface space so that the plant root system grows in the flat surface space; the plant cultivation box is placed on the rotating platform (4) and can rotate bidirectionally with the rotating platform; The microscope supplementary light system (6) is fixedly connected to the microscopic imaging system (3) and is used to supplement the light for the root system of the crop to be tested during microscopic photography.

[0011] Preferably, the three-dimensional module (2) is composed of three linear motion modules (2-1) (2-2) (2-3) and one rotation module (2-4), wherein the two linear motion modules (2-2) (2-3) are responsible for driving the microscopic imaging system to move forward and backward and up and down, the one linear motion module (2-1) is responsible for driving the microscopic imaging system to focus, and the rotation module is responsible for adjusting the imaging angle of the microscopic imaging system. The above four modules cooperate together to realize three-dimensional multi-angle shooting of the microscopic imaging system.

[0012] Preferably, the automatic zoom microscope head of the microscopic imaging system (3) uses a 4K high-resolution zoom lens and is used in conjunction with the industrial camera, with a maximum resolution of 2.8 microns.

[0013] Preferably, the rotating platform (4) is provided with four infrared limit points for rotational positioning, the four infrared limit points corresponding to the root system photography of the four surfaces of the plant culture box, and when the rotating platform rotates to a specific infrared limit point, it will automatically pause, so that the flat surface space of the plant culture box, i.e., the root system surface to be measured, faces the microscopic imaging system.

[0014] Preferably, the top of the inner box of the plant culture box is provided with four inclined surfaces (5-4) inclined toward the inner cavity, which respectively form angles with the four flat surface spaces, and the angles are used to fix the base of the plant so that the root system of the plant is immersed in the flat surface space.

[0015] Preferably, the microscope fill-light system comprises four fill-light modules, namely a synchronous dark field fill-light module (6-1), a coaxial light source fill-light module (6-2), an annular front external light source fill-light module (6-3) and a back light source fill-light module (6-4).

[0016] Preferably, the synchronous dark-field light supplement module of the microscope light supplement system includes a linkage bracket (6-1-1) and a light source assembly (6-1-2); the linkage bracket is in a "Ji" shape, with one end rigidly connected to the micro-imaging system and the other end rigidly connected to the light source assembly; the light source assembly is suspended in the inner cavity of the inner box (5-2) of the plant incubator through the linkage bracket and can move synchronously with the movement of the micro-imaging system; the light source assembly consists of a cavity (6-1-4) in the middle and symmetric light sources (6-1-3) on both sides. The plane of the cavity in the middle is perpendicular to the imaging optical axis of the micro-imaging system (3), providing a dark-field background during micro-imaging, and the symmetric light sources on both sides symmetrically supplement light to the root system to be measured from the side and rear.

[0017] Preferably, the optical axes of the symmetric light sources (6-1-3) of the light source assembly (6-1-2) face the root system to be measured, and the included angle between the light supplement optical axis and the imaging optical axis of the micro lens (3-2) is adjustable, which is used to enhance the Rayleigh scattering effect of transparent root hairs. Preferably, the included angle formed by the light supplement optical axis of the symmetric light source and the imaging optical axis of the micro lens (3-2) is 30-45°.

[0018] A method for detecting the 3D microscopic phenotype of crop roots uses the aforementioned crop root 3D microscopic phenotype detection system for detection. The detection method includes the following steps: S1: Place the plant incubator (5) on the rotary platform (4); S2: The rotary platform (4) rotates to drive the plant incubator (5) to rotate and stops after reaching the limit, so that the micro-imaging system is aligned with a root system surface to be measured; S3, turn on the microscope light supplement system (6), select a suitable light supplement scheme according to the root system development situation, and supplement light to the root system to be measured through the microscope light supplement system; S4: The three-dimensional module (2) drives the micro-imaging system (3) to move, and uses an industrial camera to take large-range root system images in the low magnification mode, obtain the global root system picture, and calculate and mark the specific position coordinates of the root system in the incubator; S5: The three-dimensional module (2) drives the micro-imaging system (3) to move, uses the micro lens to take pictures of the entire root system plane, and at the same time completes the panoramic stitching of the root system of the current surface to be measured in the homogeneous coordinate system; S6: The rotary platform (4) rotates to the next limit point to take pictures of the next root system surface to be measured; S7, repeat steps S4 to S6 until the shooting of the crop roots on the 4 root system surfaces to be measured in the entire plant incubator (4) is completed, and the task ends.

[0019] Preferably, the phenotypic trait extraction in step S5 specifically uses the Symphonies deep learning model to segment multiple parts including the crop taproot, lateral roots, root hairs and root tip meristem, and uses computer vision technology to remove impurities.

[0020] Preferably, the root phenotypic traits extracted in step S5 include root biomass, root structure, number of lateral roots, root hair density and root tip meristem length.

[0021] (III) Beneficial effects Compared with the prior art, the present invention has at least the following positive technical effects.

[0022] (1) Existing root microscopy imaging technology is limited by fixed optical paths and static light sources. The imaging area is limited to a small range, making it difficult to cover a large range of crop root structures. In addition, traditional bright field illumination has poor imaging effects on translucent tissues (such as root hairs). The present invention breaks through the imaging boundaries of traditional microscopes through the design of synchronous and coordinated movement of a three-dimensional motion module and a microscopy imaging system, combined with autofocus technology, to achieve efficient scanning of roots at multiple angles and multiple regions. By integrating image stitching algorithms with three-dimensional reconstruction technology, the system can complete global three-dimensional modeling of the root system at a micron-level resolution (such as 0.5 μm / pixel), supporting quantitative analysis of microscopic phenotypes such as root hair density and epidermal cell morphology.

[0023] (2) The existing microscopic imaging lighting scheme is difficult to adapt to the imaging requirements of the root microscopic morphology. The present invention innovatively adopts synchronous microscopic dark field lighting technology, uses a linkage bracket to realize the rigid connection between the light source and the microscope lens to achieve synchronous movement and lighting, and uses a "J"-shaped bracket to hang the light source behind the root system to be measured to achieve lighting from the side and rear. The symmetrical light source is used to symmetrically light the root system to be measured from the side and rear. At the same time, the background facing the imaging axis is designed to be a cavity dark field to enhance the detail contrast of the root system to be measured. The above design can effectively enhance the overall capture ability of the scattered light signal of translucent biological tissues (such as root hairs and root apical meristem). Experiments show that compared with traditional bright field imaging, the edge clarity of root hairs in dark field mode is improved by more than 3 times, providing a high-precision data basis for automated phenotypic parameter extraction (such as root hair length and density).

[0024] (3) Existing microscopic imaging solutions require moving or inverting samples, which cannot be applied to non-destructive microscopic phenotypic observation of plants. The present invention uses a three-dimensional motion module to drive the microscopic imaging system to actively move, so that the sample remains stationary during the shooting process, and combines it with a non-contact optical scanning strategy to achieve in-situ non-destructive detection of the root system. The system can adapt to a variety of culture modes such as soil culture and hydroponics, support dynamic monitoring of living roots (such as root tip growth rate), and avoid damage to the root microenvironment by traditional sampling methods. And through the design of a plant culture box that can be planted on all four sides and a rotating platform, combined with three-dimensional motion module control, continuous microscopic focusing and imaging of crop roots can be achieved, so that the system's single scanning time is 75% shorter than traditional single plant detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is the overall design scheme diagram of the present invention.

[0026] Figure 2 It is a schematic diagram of the structure of the plant culture box of the present invention.

[0027] Figure 3 It is a schematic diagram of the structure of the microscope fill light system of the present invention.

[0028] Figure 4 It is a schematic diagram of the synchronous dark field fill light module of the present invention.

[0029] Figure 5 The present invention is a flow chart of the crop root 3D microscopic phenotype detection method.

[0030] Figure 6 This is a comparison diagram of the photography effects of the annular front fill light and the mobile dark field fill light for rice root hairs of the present invention. DETAILED DESCRIPTION

[0031] The present invention is further described below in conjunction with the accompanying drawings and implementation examples.

[0032] The present invention discloses a crop root 3D microscopic phenotype detection system and detection method. Figure 1The figure shows the overall design scheme of the crop root 3D microscopic phenotype detection system of the present invention. The system includes an optical platform 1, a three-dimensional module 2, a microscopic imaging system 3, a rotating platform 4, a plant culture box 5 and a microscope supplementary light system 6. The optical platform 1 is placed horizontally, and a shockproof structure is provided on the surface for fixing the three-dimensional module 2. The three-dimensional module 2 consists of three linear motion modules 2-1, 2-2, 2-3 and one rotating module 2-4, which are driven by a servo motor. The linear motion module 2-1 is responsible for the focus (Z-axis movement) of the microscopic imaging system 3, and the linear motion modules 2-2 and 2-3 control the front and back (X-axis) and up and down (Y-axis) movement respectively; the rotating module 2-4 adjusts the shooting angle of the microscopic imaging system 3. The module movement accuracy is ±0.01mm, and the travel range is: 0-300mm for the X / Y axis and 0-50mm for the Z axis. The pixel resolution of industrial camera 3-1 is 4112*3000, and the maximum shooting frame rate is 30 frames; microscope lens 3-2 is a 4K automatic zoom lens with a maximum resolution of 2.8μm / pixel. The imaging system moves through the three-dimensional module 2 to achieve autofocus (based on contrast detection algorithm) and multi-angle shooting. The rotating platform 4 is installed on the right side of the optical platform 1 and is driven by a stepper motor. The rotation angle is 0-360° and the positioning accuracy is ±0.1°. The platform is equipped with 4 infrared limit points (90° apart), corresponding to the 4 flat surface spaces of the incubator 5.

[0033] like Figure 2 The figure shows the structure of the plant cultivation box 5 of the present invention. Four flat surface gaps 5-3 (thickness 5 mm) are formed between the outer box 5-1 and the inner box 5-2 of the plant cultivation box 5, and the nutrient solution is poured inside. Four inclined surfaces 5-4 are provided on the top of the inner box to fix the base of the plant so that the root system grows along the flat surface space.

[0034] like Figure 3 , Figure 4As shown, the microscope fill-in light system 6 of the present invention includes four modules, namely, a synchronous dark field fill-in light module 6-1, a coaxial light source fill-in light module 6-2, an annular external light source 6-3 and a back light source 6-4. Among them, the light source assembly 6-1-2 of the synchronous dark field fill-in light module 6-1 is suspended in the inner cavity of the incubator, and the optical axis of the bilaterally symmetrical light source 6-1-3 (power 10W) ​​and the imaging optical axis of the microscope lens 3-2 are at an angle of 30-45°, and the angle is adjustable, which improves the clarity and contrast of dark field imaging by enhancing the Rayleigh scattering of root hairs; the coaxial light source fill-in light module 6-2 provides uniform front illumination, and the adjustable brightness range is 0-1000Lux, which ensures uniform illumination of the sample surface and reduces shadows and light spots; the annular external light source 6-3 is used for lateral fill-in light, providing powerful surface illumination, which is suitable for the imaging needs of complex samples; the back light source 6-4 is used for transmission imaging, providing stable backlight illumination, and ensuring the clear presentation of the internal structure of the sample. Through the coordinated work of four supplementary illumination modules, the system can adapt to different sample types and imaging conditions to achieve high-quality, high-throughput root microscopic phenotype group detection.

[0035] like Figure 5 As shown, the specific method steps of an embodiment of the detection system of the present invention when performing detection are as follows.

[0036] (1) System startup and self-test: Turn on the image acquisition system, run the self-test program, and check the connection status and power stability of hardware devices such as the microscope imaging system 3, rotating platform 4, and three-dimensional module 2 one by one; simultaneously verify the loading status of software modules such as image capture, path planning, and stitching algorithm to ensure that there are no hardware failures or software errors.

[0037] (2) Shooting parameter setting: In the system operation interface, complete the shooting parameter configuration: set the exposure time according to the shooting requirements; select the corresponding fill light scheme (such as dark field fill light, coaxial light source, etc.); calibrate the white balance parameters to restore the true color; specify the storage path of the image and data, and complete the parameter initialization before shooting.

[0038] (3) Placement and number input of plant incubator 5: Place the plant incubator firmly in the designated position of the turntable, click the "Start Collection" function on the system interface, and enter the incubator number in the pop-up input box (such as using the naming convention of date + serial number) to complete the collection preparation work.

[0039] (4) Low-magnification photography and root positioning: Trigger the microscope imaging system to first capture a large-scale root system image in low-magnification mode to obtain a global root system image; analyze the image through an image recognition algorithm, calculate and mark the specific location coordinates of the roots in the incubator, and provide data basis for refined photography.

[0040] (5) Path planning, shooting and image naming: The computer plans a detailed shooting path (such as X / Y axis movement trajectory, focus point, etc.) based on the root position coordinates, and controls the 3D module to move along the path; triggers shooting at each point, and simultaneously names the collected images with "incubator number + shooting order" to ensure that the image data is traceable.

[0041] (6) Turntable rotation and loop shooting: After completing the current angle shooting, control the turntable to rotate to the next infrared limit point, repeat steps (4) and (5), and continue to shoot the next plant root system; after all shooting is completed, the image is globally stitched, and the stitched panoramic image and original data are saved according to the preset path, and the crop root phenotypic traits are extracted based on the above images and data.

[0042] The phenotypic trait extraction specifically uses the Symphonies deep learning model to segment multiple parts including the crop taproot, lateral roots, root hairs and root apical meristem, and uses computer vision technology to remove impurities. The root phenotypic traits extracted by the present invention include root biomass, root structure, lateral root number, root hair density and root apical meristem length.

[0043] like Figure 6 The figure shows the effect comparison of the common annular front fill light method and the mobile dark field fill light method on the rice root hair. The advantage of the fill light method of the present invention is verified by the fill light method effect comparison experiment.

[0044] The experiment used rice seedling roots (cultivated for 7 days) as samples, and used the same microscopic imaging system (industrial camera + 4K microscope lens) to shoot using traditional annular front fill light (the light source is located around the lens, and the optical axis is parallel to the imaging optical axis) and synchronous dark field fill light (the light source component is rigidly connected to the microscope lens, and the optical axis angle is 40°). The experimental results show that under traditional fill light, the edge of the root hair is blurred, the grayscale gradient value is 15-20, the contrast between the background and the root hair is low, it is difficult to distinguish the translucent tissue, and the details of the root hair bifurcation are lost, and the recognition accuracy is only 65%; while under synchronous dark field fill light, the clarity of the root hair edge is significantly improved, the grayscale gradient value reaches 50-60, the background has a dark field effect, the contrast with the root hair is significantly enhanced, the root hair boundary is clearly visible, and the recognition accuracy is increased to 95%. In addition, the synchronous dark field fill light method does not require multiple adjustments to the light source position, and the single shooting time is shortened by 20%, further improving the imaging efficiency. In summary, the synchronous dark field fill-in lighting method of the present invention is significantly superior to the traditional fill-in lighting method in terms of root hair clarity, background contrast, detail capture capability and imaging efficiency.

[0045] The specific examples described in the application are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific examples described in the present invention, or replace them in a similar manner, but they will not deviate from the spirit of the present invention or exceed the scope defined by the attached claims.

Claims

1. A crop root 3D microscopic phenotype detection system for obtaining crop root microscopic phenotypes, characterized in that: The detection system includes: An optical platform (1), placed horizontally, for accommodating other components of the detection system; The three-dimensional module (2) is fixed on the optical platform, and is controlled by a servo motor to drive the microscopic imaging system (3) to move in three dimensions, thereby achieving microscopic automatic focusing and multi-angle shooting of the root system of the crop to be tested; A microscopic imaging system (3) is installed on the three-dimensional module (2), the microscopic imaging system comprising an industrial camera (3-1) and a microscope lens (3-2) capable of automatically changing magnification, and is used for performing microscopic imaging of crop roots; A rotating platform (4) is fixed on the optical platform (1), located on the right side of the three-dimensional module (2), and is used as a platform for placing the plant culture box (5); the rotating platform is controlled by a stepping motor and can drive the plant culture box to rotate in both directions; The plant cultivation box (5) is rectangular in shape as a whole, has a square bottom surface, and is made of highly transparent material; the plant cultivation box is composed of an outer box (5-1) and an inner box (5-2), wherein a fixed gap (5-3) is provided between the inner wall of the outer box and the outer wall of the inner box, forming four narrow flat surface spaces, wherein a nutrient solution supporting plant growth is contained inside, and a crop plant is planted in each flat surface space so that the plant root system grows in the flat surface space; the plant cultivation box is placed on the rotating platform (4) and can rotate bidirectionally with the rotating platform; The microscope light-filling system (6) is fixedly connected to the microscopic imaging system (3) and is used to fill in the light for the root system of the crop to be tested during microscopic photography.

2. The crop root 3D microscopic phenotype detection system according to claim 1, characterized in that: The three-dimensional module (2) is composed of three linear motion modules (2-1) (2-2) (2-3) and one rotation module (2-4), wherein the two linear motion modules (2-2) (2-3) are responsible for driving the microscopic imaging system to move forward and backward and up and down, the one linear motion module (2-1) is responsible for driving the microscopic imaging system to focus, and the rotation module is responsible for adjusting the imaging angle of the microscopic imaging system. The above four modules work together to realize three-dimensional multi-angle shooting of the microscopic imaging system.

3. The crop root 3D microscopic phenotype detection system according to claim 1, characterized in that: The automatic zoom microscope head of the microscopic imaging system (3) adopts a 4K high-resolution zoom lens and is used in conjunction with the industrial camera, with a maximum resolution of 2.8 microns.

4. The crop root 3D microscopic phenotype detection system according to claim 1, characterized in that: The rotating platform (4) is provided with four infrared limit points for rotational positioning, the four infrared limit points corresponding to the root system photography of the four surfaces of the plant culture box, and when the rotating platform rotates to a specific infrared limit point, it will automatically pause, so that the flat surface space of the plant culture box, i.e., the root system surface to be measured, faces the photography of the microscopic imaging system.

5. The crop root 3D microscopic phenotype detection system according to claim 1, characterized in that: The top of the inner box of the plant culture box is provided with four inclined surfaces (5-4) inclined toward the inner cavity, which respectively form angles with the four flat surface spaces, and the angles are used to fix the base of the plant so that the root system of the plant is immersed in the flat surface space.

6. The crop root 3D microscopic phenotype detection system according to claim 1, characterized in that: The microscope fill-in light system comprises four fill-in light modules, namely a synchronous dark field fill-in light module (6-1), a coaxial light source fill-in light module (6-2), a ring-shaped front external light source fill-in light module (6-3) and a back light source fill-in light module (6-4).

7. The crop root 3D microscopic phenotype detection system according to claim 6, characterized in that: The synchronous dark field fill light module of the microscope fill light system comprises a linkage bracket (6-1-1) and a light source assembly (6-1-2); the linkage bracket is in the shape of a "J", one end of which is rigidly connected to the microscopic imaging system, and the other end of which is rigidly connected to the light source assembly; the light source assembly is suspended in the inner cavity space of the plant culture box inner box (5-2) through the linkage bracket, and can move synchronously with the movement of the microscopic imaging system; the light source assembly consists of a middle cavity (6-1-4) and symmetrical light sources (6-1-3) located on both sides, wherein the middle cavity plane is perpendicular to the imaging optical axis of the microscopic imaging system (3), providing a dark field background during microscopic imaging, and the symmetrical light sources on both sides symmetrically fill light the root system to be measured from the side and rear.

8. The crop root 3D microscopic phenotype detection system according to claim 7, characterized in that: The fill light axis of the symmetrical light source (6-1-3) of the light source assembly (6-1-2) faces the root system to be measured, and forms an adjustable angle with the imaging axis of the microscope lens (3-2) to enhance the Rayleigh scattering effect of the transparent root hairs.

9. The crop root 3D microscopic phenotype detection system according to claim 8, characterized in that: The angle formed by the fill light axis of the symmetrical light source and the imaging optical axis of the microscope lens (3-2) is 30-45°.

10. A method for detecting 3D microscopic phenotypes of crop roots, which uses any one of the crop root 3D microscopic phenotype detection systems described in claims 1 to 9 for detection, and the detection method comprises the following steps: S1: placing the plant culture box (5) on the rotating platform (4); S2: The rotating platform (4) rotates to drive the plant culture box (5) to rotate, and stops after reaching a limit position, so that the microscopic imaging system is aligned with a root surface to be measured; S3, turning on the microscope supplementary lighting system (6), selecting a suitable supplementary lighting scheme according to the root system development, and supplementing the light of the root system to be tested through the microscope supplementary lighting system; S4: The three-dimensional module (2) drives the microscopic imaging system (3) to move, and uses an industrial camera to capture a large-scale root system image in a low-magnification mode to obtain a global root system image, and calculate and mark the specific position coordinates of the root system in the incubator; S5: The three-dimensional module (2) drives the microscopic imaging system (3) to move, and uses a microscope lens to capture the details of the root system according to the coordinate position of the root system, and simultaneously completes the panoramic stitching and phenotypic trait extraction of the root system on the current surface to be tested; S6: The rotating platform (4) rotates to the next limit point, photographs the next root surface to be tested, and completes the panoramic stitching and phenotypic trait extraction; S7, repeating steps S4 to S6 until the photography and trait extraction of the crop roots on the four root surfaces to be tested in the entire plant incubator (4) are completed, and the task is completed.

11. The method for detecting crop root 3D microscopic phenotype according to claim 10, characterized in that: The phenotypic trait extraction in step S5 specifically uses the Symphonies deep learning model to segment multiple parts including the crop main root, lateral roots, root hairs and root tip meristem, and uses computer vision technology to remove impurities.

12. The method for detecting crop root 3D microscopic phenotype according to claim 10, characterized in that: The root phenotypic traits extracted in step S5 include root biomass, root structure, lateral root number, root hair density and root tip meristem length.

Images