Plant phenotype monitoring device and method

Abstract

The present invention discloses a plant phenotypic monitoring device and method, which relates to the field of plant growth monitoring technology. The monitoring device includes: an incubator, a data acquisition unit, and a full-band hyperspectral IoT monitoring terminal; the incubator is used to place vegetation to be measured and accommodate the data acquisition unit; the data acquisition unit includes: a slide fixing plate, a slider, a first depth camera, a second depth camera, a first external optical fiber, a second external optical fiber, and a three-axis gantry slide; the full-band hyperspectral IoT monitoring terminal is respectively connected to the first depth camera, the second depth camera, the first external optical fiber, the second external optical fiber, and the three-axis gantry slide for controlling the movement of the three-axis gantry slide; the device obtains spectral phenotypic data within a characteristic band range and two-dimensional appearance phenotypic data of each point to be measured. The present invention can realize plant phenotypic data monitoring with a high degree of automation, good phenotypic data diversity, and low cost.

Description

Technical Field

[0001] The present invention relates to the technical field of plant growth monitoring, and in particular to a plant phenotype monitoring device and method. Background Art

[0002] Plant phenotypic data collection infrastructure is the most effective way to acquire high-throughput plant phenotypic data. Building plant phenotypic infrastructure, within the framework of plant phenotypic data collection standards, can achieve standardized, high-throughput, and full-lifecycle plant phenotypic information collection. However, existing plant phenotypic monitoring systems often suffer from low automation, poor phenotypic data diversity, large site requirements, and high operating and maintenance costs, failing to meet the plant phenotypic research needs of most research institutions. Therefore, designing a plant phenotypic monitoring device and method with a high degree of automation, good phenotypic data diversity, and low cost has become a pressing technical challenge in this field. Summary of the Invention

[0003] The purpose of the present invention is to provide a plant phenotypic monitoring device and method, by which plant phenotypic data monitoring with high degree of automation, good phenotypic data diversity and low cost can be achieved.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] A plant phenotype monitoring device, comprising: an incubator, a data acquisition unit, and a full-band hyperspectral IoT monitoring terminal;

[0006] The incubator is used to place the vegetation to be measured and accommodate the data acquisition unit;

[0007] The data acquisition unit includes: a slide fixing plate, a slider, a first depth camera, a second depth camera, a first external optical fiber, a second external optical fiber and a three-axis gantry slide;

[0008] The slide fixing plate is arranged inside the incubator and is used to fix the three-axis gantry slide;

[0009] The slider is arranged on a three-axis gantry slide, and the movement of the three-axis gantry slide drives the slider to move;

[0010] The first depth camera is arranged on the top of the incubator and is used to obtain a first depth image and a first RGB image of the position of the three-axis gantry slide and the program zero point position;

[0011] The second depth camera is provided on the slider and is used to obtain a second depth image and a second RGB image of the vegetation to be measured;

[0012] The first external optical fiber is arranged above the second depth camera and fixed on the slider, and is used to collect the downward radiation emitted by the fill light on the top of the incubator;

[0013] The second external optical fiber is disposed below the second depth camera and fixed to the slider, and is used to collect the uplink radiation emitted by the vegetation to be measured;

[0014] The full-band hyperspectral IoT monitoring terminal is respectively communicated with the first depth camera, the second depth camera, the first external optical fiber, the second external optical fiber and the three-axis gantry slide, and is used to control the movement of the three-axis gantry slide according to the first depth image and the first RGB image obtained by the first depth camera or the second depth image and the second RGB image obtained by the second depth camera; pre-process the downlink radiation and the uplink radiation to obtain spectral phenotypic data in a characteristic band range; determine the appearance two-dimensional phenotypic data of the position of each test point according to the second RGB image, and the appearance two-dimensional phenotypic data includes: leaf stem height, leaf diameter, leaf area and leaf color.

[0015] The present invention also provides a plant phenotype monitoring method applied to the plant phenotype monitoring device described above, the monitoring method comprising the following steps:

[0016] Determine, based on the first depth image acquired by the first depth camera, whether the position of the three-axis gantry slide is at the program zero position, and obtain a first determination result;

[0017] If the first judgment result is no, controlling the three-axis gantry slide to move to the program zero position;

[0018] If the first judgment result is yes, obtaining a second depth image and a second RGB image of the vegetation to be measured according to the second depth camera;

[0019] Determining the positions of each to-be-measured point of the to-be-measured vegetation according to the second depth image and the second RGB image;

[0020] Controlling the three-axis gantry slide to move to the positions of the points to be measured respectively;

[0021] Determining the appearance two-dimensional phenotypic data of each of the test points based on the second RGB image, wherein the appearance two-dimensional phenotypic data includes: leaf stem height, leaf diameter, leaf area, and leaf color;

[0022] Acquire the downward radiation emitted by the fill light on the top of the incubator at each of the test points according to the first external optical fiber;

[0023] Acquiring uplink radiation emitted by the vegetation to be measured at each of the locations to be measured according to the second external optical fiber;

[0024] Preprocessing the downlink radiation and the uplink radiation of each of the test points to obtain spectral phenotype data within a characteristic wavelength range of each of the test points;

[0025] Determine whether the three-axis gantry slide has completed the movement of all the test points to obtain a second determination result;

[0026] If the second judgment result is no, return to the step of "controlling the three-axis gantry slide to move to the positions of the points to be measured respectively";

[0027] If the second judgment result is yes, the monitoring is completed.

[0028] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0029] The present invention provides a plant phenotypic monitoring device and method, the monitoring device comprising: an incubator, a data acquisition unit and a full-band hyperspectral Internet of Things monitoring terminal; the incubator is used to place vegetation to be tested and accommodate the data acquisition unit; the data acquisition unit comprises: a slide fixing plate, a slider, a first depth camera, a second depth camera, a first external optical fiber, a second external optical fiber and a three-axis gantry slide; the full-band hyperspectral Internet of Things monitoring terminal is respectively communicated with the first depth camera, the second depth camera, the first external optical fiber, the second external optical fiber and the three-axis gantry slide, and is used to control the movement of the three-axis gantry slide according to the first depth image and the first RGB image obtained by the first depth camera or the second depth image and the second RGB image obtained by the second depth camera; the downlink radiation and the uplink radiation are preprocessed to obtain spectral phenotypic data in a characteristic band range; and the appearance two-dimensional phenotypic data of each position of the test point is determined according to the second RGB image. Compared with traditional plant phenotypic monitoring systems, the present invention only needs to be set up on a full-band hyperspectral IoT monitoring terminal to realize automatic multi-point data collection, which can achieve high automation; moreover, compared with traditional plant phenotypic monitoring systems, the present invention can simultaneously collect non-imaging spectral data (i.e., the downlink radiation emitted by the fill light on the top of the incubator and the uplink radiation emitted by the vegetation to be measured), RGB image data, depth image data, and the appearance two-dimensional phenotypic data of each test point position generated based on the RGB image data and the depth image data, and has good phenotypic data diversity; in addition, the plant phenotypic monitoring device of the present invention is based on a single plant in terms of scale, which can reduce the volume occupied by the equipment, and a single device can monitor multiple plants, thereby reducing operating costs and maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention 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 invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0031] Figure 1 A structural diagram of a plant phenotype monitoring device provided in Example 1 of the present invention;

[0032] Figure 2 A front view of a plant phenotype monitoring device provided in Example 1 of the present invention;

[0033] Figure 3 This is a flow chart of a plant phenotype monitoring method provided in Example 2 of the present invention.

[0034] Explanation of symbols:

[0035] 1. Full-band hyperspectral IoT monitoring terminal; 2. Slide fixing plate; 3. Incubator; 4. Second depth camera; 5. First depth camera; 6. First external optical fiber; 7. Second external optical fiber; 8. Three-axis gantry slide; 9. Slider. DETAILED DESCRIPTION

[0036] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0037] The purpose of the present invention is to provide a plant phenotypic monitoring device and method, by which plant phenotypic data monitoring with high degree of automation, good phenotypic data diversity and low cost can be achieved.

[0038] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0039] Example 1:

[0040] See also Figure 1 and Figure 2 , the present invention provides a plant phenotype monitoring device, the monitoring device includes: an incubator 3, a data acquisition unit and a full-band hyperspectral IoT monitoring terminal 1;

[0041] The incubator 3 is used to place the vegetation to be tested and accommodate the data acquisition unit; wherein, the incubator 3 is also used to provide a constant incubation environment for the vegetation to be tested. The traditional plant phenotypic measurement process usually involves removing the plant and measuring the plant's phenotypic data using a spectrometer and a camera. However, this traditional measurement method changes the plant's growth environment and affects its normal growth. In the present invention, there is no need to remove the sample to be tested, thereby avoiding unnecessary ambient light loss and loss caused by changes in the plant's growth environment, allowing it to grow in a constant incubator environment, which is conducive to analyzing plant phenotypic data and establishing the accuracy of plant phenotype-related models.

[0042] The data acquisition unit includes: a slide fixing plate 2, a slider 9, a first depth camera 5, a second depth camera 4, a first external optical fiber 6, a second external optical fiber 7 and a three-axis gantry slide 8;

[0043] The slide fixing plate 2 is arranged inside the incubator 3 and is used to fix the three-axis gantry slide 8;

[0044] The slider 9 is arranged on the three-axis gantry slide 8, and the movement of the three-axis gantry slide 8 drives the slider 9 to move;

[0045] The first depth camera 5 is provided on the top of the incubator 3 and is used to obtain a first depth image and a first RGB image of the position of the three-axis gantry slide and the program zero position;

[0046] The second depth camera 4 is provided on the slider 9 and is used to obtain a second depth image and a second RGB image of the vegetation to be measured;

[0047] The first external optical fiber 6 is arranged above the second depth camera 4 and fixed on the slider 9, and is used to collect the downward radiation emitted by the fill light on the top of the incubator;

[0048] The second external optical fiber 7 is provided below the second depth camera 4 and fixed on the slider 9, and is used to collect the uplink radiation emitted by the vegetation to be measured;

[0049] The full-band hyperspectral IoT monitoring terminal 1 is respectively communicated with the first depth camera 5, the second depth camera 4, the first external optical fiber 6, the second external optical fiber 7 and the three-axis gantry slide 8, and is used to control the movement of the three-axis gantry slide 8 according to the first depth image and the first RGB image obtained by the first depth camera 5 or the second depth image and the second RGB image obtained by the second depth camera 4; pre-process the downlink radiation and the uplink radiation to obtain spectral phenotypic data in a characteristic band range; determine the appearance two-dimensional phenotypic data of the position of each test point according to the second RGB image, and the appearance two-dimensional phenotypic data includes: leaf stem height, leaf diameter, leaf area and leaf color. Specifically, the first depth camera 5 and the second depth camera 4 are respectively connected to the full-band hyperspectral Internet of Things monitoring terminal 1 through the USB-type-c interface; the three-axis gantry slide 8 is connected to the full-band hyperspectral Internet of Things monitoring terminal 1 through the aviation plug interface; the first external optical fiber 6 is connected to the SMA-SMA interface of channel I on the full-band hyperspectral Internet of Things monitoring terminal 1; the second external optical fiber 7 is connected to the SMA-SMA interface of channel II on the full-band hyperspectral Internet of Things monitoring terminal 1.

[0050] Among them, according to the first depth image and the first RGB image obtained by the first depth camera 5 or the second depth image and the second RGB image obtained by the second depth camera 4, controlling the movement of the three-axis gantry slide 8 specifically includes:

[0051] The first depth camera 5 runs a deep learning-based object detection network to identify the position of the three-axis gantry and the program zero position. It then uses its own infrared laser emitter and infrared sensor to obtain first depth images of the current three-axis gantry position and the program zero position. It then uses a color sensor to obtain first RGB images of these two positions. The full-band hyperspectral IoT monitoring terminal 1 receives the first depth image and first RGB image from the first depth camera 5 and, based on their corresponding relationship, converts them into depth coordinates at the center point through bilinear interpolation or other means. The rs.rs2_deproject_pixel_to_point function is used to obtain the coordinates of the program zero position, with the infrared laser emitter in the first depth camera 5 as the origin. The rs2_deproject_pixel_to_point function is then used to calculate the three-dimensional coordinates of the current three-axis gantry position. The difference between the two values ​​is then divided by the number of pulses required for a single stroke of the three-axis gantry. The result is converted into a PWM waveform and sent to the three-axis gantry, thereby controlling the movement of the three-axis gantry to the designated position. Among them, the rs.rs2_deproject_pixel_to_point function corresponds to the calculation of the program zero position coordinates; the rs2_deproject_pixel_to_point function corresponds to the calculation of the position coordinates of the end slider of the three-axis gantry slide. These two functions are open source codes for obtaining depth information from the D435 depth camera. The working principle of the first depth camera 5 is similar to that of the lidar, except that the laser is replaced by a light pulse beam. It uses the light pulse beam emitted by the depth camera to detect objects, and obtains the depth information of the object to be measured based on the time of the returned light pulse beam. After the depth coordinates are subsequently subtracted, the difference between the coordinates of the X-axis, Y-axis, and Z-axis is obtained. The difference is divided by the number of stepper motor pulses corresponding to the screw stroke (that is, the distance the screw moves forward when the stepper motor rotates one circle) to obtain the number of pulses required to reach the program zero position, thereby controlling the stepper motor to drive the slide to move to the program zero position.

[0052] The second depth camera 4 runs an instance segmentation network based on deep learning to identify the position of the vegetation to be measured, uses its own infrared laser emitter and infrared sensor to obtain the second depth image of the current position of the vegetation to be measured, and then uses the color sensor to obtain the second RGB images of the above multiple vegetation positions to be measured.

[0053] The full-band hyperspectral IoT monitoring terminal 1 receives the second depth image and the second RGB image from the second depth camera, and converts them into depth coordinates at the center point through bilinear interpolation and other means according to their corresponding relationship. The coordinate system coordinates of the current three-axis gantry slide position with the infrared camera in the depth camera II as the origin are obtained through the rs.rs2_deproject_pixel_to_point function. The three-dimensional coordinates of the position of the vegetation to be measured at each point are calculated through the rs2_deproject_pixel_to_point function. The difference operation is performed on the two, and the result of the operation is divided by the number of pulses required for the one-way stroke of the three-axis gantry slide, converted into a PWM wave, and sent to the three-axis gantry slide, which can control the three-axis gantry slide to move to the specified position.

[0054] Preprocessing the downlink radiation and the uplink radiation to obtain spectral phenotypic data within a characteristic wavelength range specifically includes:

[0055] Channel I and channel II are connected to the mechanical optical path switch, the mechanical optical path switch is connected to the spectrometer, and the spectrometer is connected to the central control board, wherein the mechanical optical path switch, the spectrometer, and the central control board are all components inside the full-band hyperspectral IoT monitoring terminal. The working principle is as follows: the central control board sends a 01 signal to control the channel I of the mechanical optical path switch to open and channel II to close, and the first external optical fiber obtains the optical signal (i.e., the downward radiation emitted by the fill light on the top of the incubator), and sends the collected signal to the spectrometer via channel I. After diffraction by the slit inside the spectrometer, it is converted into 3648 characteristic spectral data points in the range of 343 to 1036 nm, and after matching the characteristic bands with the characteristic spectral data points and sorting them in order according to the size of the characteristic bands, a total of 13 groups of character string matrices are formed with 261 in each group, which are stored in the central control board; the central control board sends a 10 signal to control the channel II of the mechanical optical path switch to open and channel I to close, and the external optical fiber II obtains the optical signal ( That is, the uplink radiation emitted by the vegetation to be measured), and the collected signal is sent to the spectrometer via channel II. After diffraction by the slit inside the spectrometer, it is converted into 3648 characteristic spectral data points in the range of 343 to 1036 nm. After matching the characteristic bands with the characteristic spectral data points and sorting them in order according to the size of the characteristic bands, 13 groups of string matrices are formed with 261 points in each group, and stored in the central control board. After the above-mentioned downlink radiation and uplink radiation are collected, a total of 26 groups of data of the downlink radiation string matrix and the uplink radiation string matrix are sent to the Internet of Things platform in the form of MQTT protocol. After a delay of 10 seconds, the central control board clears the downlink radiation string matrix and the uplink radiation string matrix and enters the standby state.

[0056] Determining the appearance two-dimensional phenotype data of each of the test points based on the second RGB image specifically includes:

[0057] The full-band hyperspectral IoT monitoring terminal 1 performs binarization processing on the second RGB image received from the second depth camera 4, that is, the grayscale value of the pixel point of the plant leaf image is set to 0 or 255 to form a black and white image, and performs edge detection on it to extract the outer contour of the plant leaf in the two-dimensional plane. The python opencv algorithm is used to obtain the minimum circumscribed matrix containing the feature image, and the matrix is ​​substituted into the minarerect algorithm to obtain the leaf stem height, leaf diameter and leaf area; the area divided by the edge detection is subjected to image fusion processing with the second RGB image, the color of each area of ​​the processed image is matched with the standard colorimetric card model, and the color value obtained from the color card of each area is restored to each area where the plant leaf is divided to obtain the leaf color.

[0058] The full-band hyperspectral IoT monitoring terminal 1 sends the spectral phenotypic data of the characteristic band range and the two-dimensional phenotypic data of the appearance including leaf stem height, leaf diameter, leaf area and leaf color data to the IoT platform, and then transfers it to the cloud database RDS through the set rule engine. The user uses the client software supporting the system to display and locally save the spectral data.

[0059] In this embodiment, the models of the first depth camera 5 and the second depth camera 4 are Intel RealSense D435; the network used by the first depth camera 5 is the YOLOv5 target detection network; the network used by the second depth camera 4 is the Mask-RCNN instance segmentation network; the three-axis gantry slide 8 includes: a KK40-S300-SG-1610 module, a 57 stepper motor and a DM542 driver; the first external optical fiber 6 and the second external optical fiber 7 are SMA905-SMA905 armored optical fibers; the incubator 3 is a cold light source dual-temperature zone artificial climate chamber 1400D-2LED; the full-band hyperspectral Internet of Things monitoring terminal 1 is a full-band hyperspectral Internet of Things monitoring terminal independently developed by Su Baofeng's team from the School of Mechanical and Electronic Engineering of Northwest Agriculture and Forestry University.

[0060] In summary, compared with the traditional plant phenotypic monitoring system, the present invention only needs to be set up on the full-band hyperspectral Internet of Things monitoring terminal to realize automatic multi-point data collection, which can achieve high automation; compared with the traditional phenotypic observation system, the present invention is combined with the Internet of Things system, and the collected data can be uploaded to the user-specified cloud server at a regular interval, and the client provided by the system displays and locally saves the data, ensuring the stability, continuity and accuracy of the data, and realizing highly unmanned operation; compared with the traditional phenotypic observation system, the present invention is combined with deep learning to control the vertical distance between the external optical fiber on the slider and the vegetation to be measured to a certain value, effectively avoiding problems such as inaccurate data caused by light coupling and measurement errors caused by optical path difference, thereby improving the accuracy of the data. Among them, the vertical distance is a certain value, which is mainly reflected in that after obtaining the coordinates of the vegetation to be measured, the coordinates of the Z axis are reduced by the set threshold value (such as 5 cm), so that after the three-axis gantry slide drives the slider to reach the coordinates of the vegetation to be measured at that point, the vertical distance therefrom in the Z axis direction is maintained at the set threshold value (such as 5 cm). This can effectively reduce the adverse effects of optical path difference and fiber coupling on the accuracy of spectral data. Moreover, compared with traditional plant phenotypic monitoring systems, the present invention can simultaneously collect non-imaging spectral data (i.e., the downward radiation emitted by the fill light on the top of the incubator and the upward radiation emitted by the vegetation to be measured) and RGB image data, and has better phenotypic data diversity; in addition, the plant phenotypic monitoring device of the present invention is based on a single plant in terms of scale, which can reduce the volume occupied by the equipment. Moreover, a single device can monitor multiple plants, thereby reducing operating costs and maintenance costs.

[0061] Example 2:

[0062] See also Figure 3 The present invention provides a plant phenotype monitoring method, which comprises the following steps:

[0063] S1: Determine whether the position of the three-axis gantry slide is at the program zero position according to the first depth image obtained by the first depth camera, and obtain a first determination result;

[0064] S2: If the first judgment result is no, controlling the three-axis gantry slide to move to the program zero position;

[0065] S3: If the first judgment result is yes, obtaining a second depth image and a second RGB image of the vegetation to be measured according to the second depth camera;

[0066] S4: determining the position of each to-be-measured point of the to-be-measured vegetation according to the second depth image and the second RGB image;

[0067] S5: Control the three-axis gantry slide to move to the positions of the points to be measured respectively;

[0068] S6: determining the appearance two-dimensional phenotypic data of each of the test points based on the second RGB image, wherein the appearance two-dimensional phenotypic data includes: leaf stem height, leaf diameter, leaf area, and leaf color;

[0069] S7: Obtaining downward radiation emitted by the fill light on the top of the incubator at each of the test points according to the first external optical fiber;

[0070] S8: Acquire, according to the second external optical fiber, the uplink radiation emitted by the vegetation to be measured at each of the locations of the vegetation to be measured;

[0071] S9: pre-processing the downlink radiation and the uplink radiation of each of the test points to obtain spectral phenotype data of a characteristic wavelength range of each of the test points;

[0072] S10: Determine whether the three-axis gantry slide has completed the movement of all the test points to obtain a second determination result;

[0073] S11: If the second judgment result is no, return to the step of "controlling the three-axis gantry slide to move to the positions of the points to be measured respectively";

[0074] S12: If the second judgment result is yes, the monitoring is completed.

[0075] In step S4, determining the position of each test point of the vegetation to be measured based on the second depth image and the second RGB image specifically includes:

[0076] S401: converting the second depth image and the second RGB image using a method such as bilinear interpolation to obtain depth coordinates;

[0077] S402: using the rs2_deproject_pixel_to_point function to solve the depth coordinates to obtain the three-dimensional coordinates of the positions of the various points to be measured of the vegetation to be measured.

[0078] In this embodiment, after step S12, the following steps are further included:

[0079] S13: Determine whether the interval monitoring time reaches the set time, and obtain a third determination result;

[0080] S14: If the third judgment result is yes, then determining whether to continue monitoring the vegetation to be tested, and obtaining a fourth judgment result; that is, when the set interval monitoring time is reached, if the acquired spectral phenotypic data within the characteristic band range and the two-dimensional phenotypic data of the appearance of each test point of the vegetation to be tested do not meet the requirements, then continuing to test the vegetation to be tested;

[0081] S15: If the fourth judgment result is yes, return to step S3;

[0082] S16: If the fourth judgment result is no, the monitoring is completed.

[0083] As a possible implementation manner, when the second judgment result is yes, the method further includes:

[0084] S17: Determine whether the position of the three-axis gantry slide is at the program zero position, and obtain a fifth determination result;

[0085] S18: If the fifth judgment result is no, controlling the three-axis gantry slide to move to the program zero position;

[0086] S19: If the fifth judgment result is yes, execute step S13.

[0087] It should be noted that, in this embodiment, the method of controlling the movement of the three-axis gantry slide, the method of preprocessing the downward radiation and the upward radiation to obtain spectral phenotypic data in a characteristic band range, and the method of determining the appearance two-dimensional phenotypic data of the position of each test point based on the second RGB image are the same as those in Example 1 and will not be repeated here.

[0088] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0089] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A plant phenotype monitoring device, characterized in that: include: Incubator, data acquisition unit and full-band hyperspectral IoT monitoring terminal; The incubator is used to place the vegetation to be measured and accommodate the data acquisition unit; The data acquisition unit includes: a slide fixing plate, a slider, a first depth camera, a second depth camera, a first external optical fiber, a second external optical fiber and a three-axis gantry slide; The slide fixing plate is arranged inside the incubator and is used to fix the three-axis gantry slide; The slider is arranged on a three-axis gantry slide, and the movement of the three-axis gantry slide drives the slider to move; The first depth camera is arranged on the top of the incubator and is used to obtain a first depth image and a first RGB image of the position of the three-axis gantry slide and the program zero point position; The second depth camera is provided on the slider and is used to obtain a second depth image and a second RGB image of the vegetation to be measured; The first external optical fiber is arranged above the second depth camera and fixed on the slider, and is used to collect the downward radiation emitted by the fill light on the top of the incubator; The second external optical fiber is disposed below the second depth camera and fixed to the slider, and is used to collect the uplink radiation emitted by the vegetation to be measured; The full-band hyperspectral IoT monitoring terminal is respectively communicated with the first depth camera, the second depth camera, the first external optical fiber, the second external optical fiber and the three-axis gantry slide, and is used to control the movement of the three-axis gantry slide according to the first depth image and the first RGB image obtained by the first depth camera or the second depth image and the second RGB image obtained by the second depth camera; pre-process the downlink radiation and the uplink radiation to obtain spectral phenotypic data in a characteristic band range; determine the appearance two-dimensional phenotypic data of each test point position according to the second RGB image, and the appearance two-dimensional phenotypic data includes: leaf stem height, leaf diameter, leaf area and leaf color.

2. The plant phenotype monitoring device according to claim 1, characterized in that: The models of the first depth camera and the second depth camera are Intel RealSense D435.

3. The plant phenotype monitoring device according to claim 1, characterized in that: The network used by the first depth camera is the YOLOv5 target detection network.

4. The plant phenotype monitoring device according to claim 1, characterized in that The network used by the second depth camera is the Mask-RCNN instance segmentation network.

5. The plant phenotype monitoring device according to claim 1, characterized in that: The three-axis gantry slide includes: KK40-S300-SG-1610 module, 57 stepper motor and DM542 driver.

6. The plant phenotype monitoring device according to claim 1, characterized in that: The first external optical fiber and the second external optical fiber are SMA905-SMA905 armored optical fibers.

7. The plant phenotype monitoring device according to claim 1, characterized in that: The incubator is a cold light source dual-temperature zone artificial climate box 1400D-2LED.

8. A plant phenotype monitoring method applied to the plant phenotype monitoring device according to any one of claims 1 to 7, characterized in that: The following steps are involved: Determine, based on the first depth image acquired by the first depth camera, whether the position of the three-axis gantry slide is at the program zero position, and obtain a first determination result; If the first judgment result is no, controlling the three-axis gantry slide to move to the program zero position; If the first judgment result is yes, obtaining a second depth image and a second RGB image of the vegetation to be measured according to the second depth camera; Determining the positions of each to-be-measured point of the to-be-measured vegetation according to the second depth image and the second RGB image; Controlling the three-axis gantry slide to move to the positions of the points to be measured respectively; Determining the appearance two-dimensional phenotypic data of each of the test points based on the second RGB image, wherein the appearance two-dimensional phenotypic data includes: leaf stem height, leaf diameter, leaf area, and leaf color; Acquire the downward radiation emitted by the fill light on the top of the incubator at each of the test points according to the first external optical fiber; Acquiring uplink radiation emitted by the vegetation to be measured at each of the locations to be measured according to the second external optical fiber; Preprocessing the downlink radiation and the uplink radiation of each of the test points to obtain spectral phenotype data within a characteristic wavelength range of each of the test points; Determine whether the three-axis gantry slide has completed the movement of all the test points to obtain a second determination result; If the second judgment result is no, return to the step of "controlling the three-axis gantry slide to move to the positions of the respective test points"; If the second judgment result is yes, the monitoring is completed.

9. The plant phenotype monitoring method according to claim 8, characterized in that: Determining the positions of each test point of the vegetation to be measured based on the second depth image and the second RGB image specifically includes: Converting the second depth image and the second RGB image using a bilinear interpolation method to obtain depth coordinates; The depth coordinates are solved using the rs2_deproject_pixel_to_point function to obtain the three-dimensional coordinates of the positions of the various points to be measured of the vegetation to be measured.

10. The plant phenotype monitoring method according to claim 8, characterized in that: After the "if the second judgment result is yes" step, the method further includes: Determine whether the interval monitoring time reaches the set time, and obtain a third determination result; If the third judgment result is yes, determining whether to continue monitoring the vegetation to be measured to obtain a fourth judgment result; If the fourth judgment result is yes, return to the step of "obtaining a second depth image and a second RGB image of the vegetation to be measured according to the second depth camera"; If the fourth judgment result is no, the monitoring is completed.

Images