Mobile portal frame phenotype acquisition device and application thereof

Abstract

The invention provides a mobile portal frame phenotype acquisition device and application thereof. The device can be used for collecting plant phenotype data. The device is provided with the transverse track and the longitudinal track at the same time, so that the device can move conveniently, and phenotypic data can be collected from one area to another area. In the device, the image acquisition device connected to the manual holder can comprise a camera and lighting equipment, and can be adjusted according to the requirement of image acquisition. In addition, in the device, the detection device is detachable and convenient to use.

Description

Technical Field

[0001] The present invention belongs to the field of agricultural robots. More specifically, the present invention relates to a mobile gantry phenotypic acquisition device and its application. Background Art

[0002] Mining key genes that control key plant traits has become the current focus of research in botany, especially plant genetics. In particular, with the rapid development of high-throughput sequencing technology, the pace of human exploration of key functional genes in plants is accelerating at an unprecedented speed. However, due to the highly complex plant genome, huge environmental variations, and extremely complex interaction relationships between genotypes and the environment, humans are still in the initial stage of elucidating the molecular mechanisms behind the complex plant phenotypes. In today's era of rapid development of sequencing technology, the academic consensus is that the development and application of high-throughput phenomics technology have become the bottleneck in current botany and even life science research. The relative lag of phenomics technology directly hinders the effective use of current advanced biotechnology represented by high-throughput sequencing for the benefit of humanity, and greatly hinders humans from making the most of plant resources to serve humanity. Therefore, countries around the world are actively establishing and developing phenomics platforms and technologies, striving to gain a strategic initiative in the field of current key functional gene mining and seize the dominant and discourse rights in future industrial competitions based on plant science.

[0003] High-throughput plant phenotype measurement will become the bottleneck in the current competition in the field of agricultural breeding.

[0004] Currently, major breeding companies and research institutions around the world have realized that in the future, to achieve a substantial increase in traits such as grain yield, quality, and resistance, molecular design breeding is the key. The core of molecular design breeding is to use modern biotechnology such as current genome sequencing, genome-wide association analysis, and genome editing to make full use of the natural variations in crops and design and create excellent varieties with specific traits. However, to effectively use these biotechnologies, the current "stuck neck" problem is to obtain large-scale phenotypes of different crop traits under different environmental conditions, and then combine genomic sequence information and modern molecular genetic means to clarify the genetic variations that control key traits. Summary of the Invention

[0005] The object of the present invention is to provide a mobile gantry phenotypic acquisition device and its application. This device can be used for the acquisition of plant phenotypic data. Since this device has both horizontal and vertical tracks, it can be moved conveniently and extended from one area to another field for phenotypic data acquisition. In this device, the image acquisition device connected to the manual turntable can include a camera and lighting equipment, and can be adjusted according to the needs of image acquisition. In addition, in this device, the detection device is detachable and easy to use.

[0006] In the first aspect of the present invention, a mobile gantry phenotype acquisition device is provided, and the device includes: a track part composed of a track beam and embedded parts under the track beam, and a gantry containing a detection device;

[0007] The track part includes: a horizontal (X) track and a vertical (Y) track;

[0008] The gantry includes: end beams, a main beam (Z) track, a lifting wheel set, lifting wheels, moving wheels, a platform vehicle traveling mechanism, and a trolley traveling mechanism; wherein, the main beam (Z) track is parallel to the ground, and a detection device is detachably and movably connected to the main beam (Z) track, and the detection device includes a trolley traveling mechanism for controlling the movement of the detection device on the main beam (Z) track; the end beams and the main beam track are connected together; the moving wheels can change directions so as to realize the switching between the horizontal (X) track and the vertical (Y) track; the platform vehicle traveling mechanism is connected to one side end beam for controlling the movement of the gantry on the vertical (Y) track; the trolley traveling mechanism is connected to the other side end beam for controlling the movement of the gantry on the horizontal (X) track.

[0009] In one or more embodiments, the lifting wheel set is connected to the lower end of the end beam for adjusting the height of the main beam track from the ground.

[0010] In one or more embodiments, the lifting wheel set is composed of a plurality of lifting wheels.

[0011] In one or more embodiments, the detection device further includes an adapter plate, and the detection device is fixed and disassembled by using the card slots on the adapter plate.

[0012] In one or more embodiments, the track beam is composed of a track beam main body, fixed parts and slide rails located on the track beam main body.

[0013] In one or more embodiments, the embedded parts are composed of embedded support parts, connecting parts and embedded brackets, wherein the connecting parts connect the embedded support parts and the embedded brackets together.

[0014] In one or more embodiments, there are 4 embedded brackets, which are respectively connected to 4 connecting parts of the embedded support parts.

[0015] In one or more embodiments, the detection device includes a trolley traveling mechanism, a manual pan-tilt connected to the trolley traveling mechanism, and an image acquisition device connected to the manual pan-tilt.

[0016] In one or more embodiments, the image acquisition device is detachably connected to the manual pan-tilt.

[0017] In one or more embodiments, the manual pan-tilt head includes a sphere, a pan-tilt main knob, a panoramic base knob, a 360° rotation base scale, a clamp base knob, a clamp, and a sleeve; the clamp base knob and the clamp cooperate with each other to adjust the tightness of the clamp and are connected to the trolley traveling mechanism; the sphere connects the clamp and the sleeve together and cooperates with the panoramic base knob and the 360° rotation base scale to rotate the sleeve; the pan-tilt main knob is used to rotate the manual pan-tilt head.

[0018] In one or more embodiments, the image acquisition device includes a camera and an optional lighting device.

[0019] In one or more embodiments, the image acquisition device further includes a heat insulation plate.

[0020] In a second aspect of the present invention, there is provided the use of the mobile gantry phenotypic acquisition device of the present invention in acquiring plant phenotypes.

[0021] In one or more embodiments, the plant is a crop.

[0022] In one or more embodiments, the plants include: rice, wheat, corn, and soybeans.

[0023] In a third aspect of the present invention, there is provided a method for acquiring plant phenotypic data, the method comprising: using the mobile gantry phenotypic acquisition device of the present invention to acquire plant phenotypic data.

[0024] In one or more embodiments, the method includes:

[0025] 1) Controlling the mobile gantry phenotypic acquisition device of the present invention to move to a suitable position;

[0026] 2) Using the mobile gantry phenotypic acquisition device to acquire plant phenotypic pictures;

[0027] 3) Obtaining plant phenotypic data of the plant by analyzing the plant phenotypic pictures obtained in 2).

[0028] In one or more embodiments, the plant is a crop.

[0029] In one or more embodiments, the plants include: rice, wheat, corn, and soybeans.

[0030] In one or more embodiments, the movement in step 1) includes: movement in the X-axis, Y-axis, and Z-axis directions.

[0031] In one or more embodiments, the plant phenotypes in step 2) include: pictures of plant phenotypes, the number of panicles at the heading stage of the plant, and the projected area of a single plant.

[0032] In one or more embodiments, in step 2), an image acquisition device of a mobile gantry phenotyping acquisition device is used to acquire pictures of plant phenotypes, and the coincidence degree of the pictures acquired at two adjacent sampling points is above 75%, above 80%, above 85% or above 90%.

[0033] In one or more embodiments, in step 2), an image acquisition device of a mobile gantry phenotyping acquisition device is used to acquire pictures of plant phenotypes, and the number of sampling points > 30, > 40 or > 50.

[0034] In one or more embodiments, step 3) includes: using Pix4D software to process the plant phenotype pictures obtained in step 2), and performing merging processing on the pictures to form plant three-dimensional reconstruction pictures.

[0035] In one or more embodiments, in step 1), a server (PLC) and a field industrial control computer are included in the mobile gantry phenotyping acquisition device to control the movement of the mobile gantry phenotyping acquisition device.

[0036] In one or more embodiments, the main program of the server (PLC) includes: a main thread and an acquisition thread.

[0037] In one or more embodiments, the main thread of the server (PLC) includes: sending a status inquiry instruction to the field industrial control computer to inquire about the position of the mobile gantry phenotyping acquisition device, so as to control whether the movement of the mobile gantry phenotyping acquisition device stops; the acquisition thread of the server (PLC) includes: according to the acquisition path, sending an acquisition instruction to the field industrial control computer, receiving the feedback of the field industrial control computer, if the picture already exists, the picture can be remotely downloaded or a picture deletion instruction is sent to the field industrial control computer, and the thread exits after the acquisition at the last acquisition point is completed.

[0038] In one or more embodiments, the main program of the field industrial control computer includes: receiving the instruction of the main program of the server (PLC), creating an acquisition thread according to the acquisition instruction, sending an instruction on whether to stop acquisition to the main program of the server after acquiring the picture, updating the real-time position of the mobile gantry phenotyping acquisition device according to the status inquiry instruction, and deleting the picture according to the picture deletion instruction.

[0039] In one or more embodiments, the acquisition thread of the field industrial control computer includes: updating the status and moving the mobile gantry phenotyping acquisition device, updating the status and acquiring images, updating the status and performing image processing, updating the status and saving the pictures, and updating the status and exiting the thread.

[0040] In one or more embodiments, the analysis in step 3) includes: rice heading stage analysis and rice single plant projected area analysis.

[0041] In one or more embodiments, the heading stage analysis includes:

[0042] 3.1) Manual annotation, including annotating the panicle part on the pictures of rice phenotypes collected in step 2) (e.g., using software such as Label studio, labelimg, etc.);

[0043] 3.2) Detection model construction: Use the yolo and rcnn models to train, test, and validate the pictures after manual annotation in 3.1), and construct a detection model for heading stage analysis;

[0044] 3.3) Application of the detection model: Analyze the heading stage according to the detection model in 3.2); for example: According to the collection time point, obtain the dynamic curve of the number of panicles over time, and define the date corresponding to 10% of the maximum number of panicles as the initial heading stage. Combine the pictures of rice phenotypes collected in step 2) and the detection model in 3.2) to analyze the heading stage.

[0045] In one or more embodiments, the analysis of the projected area of a single rice plant includes:

[0046] 3.1) Denoise the background of the pictures of rice phenotypes collected in step 2);

[0047] 3.2) Distinguish rice from other plants. For example, use the CIVE vegetation index combined with the Otsu threshold to separate the soil from the vegetation. For targets such as duckweed that are smaller in area compared to the main body of rice, remove them by setting a threshold;

[0048] 3.2) Obtain the area covered by the rice plants, and calculate the projected area of a single rice plant = the area covered by the rice plants ÷ the number of rice plants.

[0049] Other aspects of the present invention will be apparent to those skilled in the art from the disclosure herein. Brief Description of the Drawings

[0050] Figure 1 Structural diagrams of the track beam and embedded parts.

[0051] Figure 2 Track schematic diagram.

[0052] Figure 3 Schematic diagram of the mobile gantry.

[0053] Figure 4 Schematic diagram of the mobile gantry phenotype acquisition device.

[0054] Figure 5 Schematic diagram of the mobile gantry phenotype acquisition device and its scanning coverage range.

[0055] Figure 6, Schematic diagram of manual photo annotation and automatic identification of rice ears by deep learning model.

[0056] Figure 7 , Schematic diagram of background denoising for high-definition single-plant pictures.

[0057] Figure 8 , Camera external view diagram.

[0058] Figure 9 , External view diagram of the model MVL-LF5040M-F lens.

[0059] Figure 10 , Schematic diagram of the manual pan-tilt.

[0060] Figure 11 , Schematic diagram of the power cord and Ethernet waterproof connector of the access module.

[0061] Figure 12 , Example diagram of camera parameter calibration results.

[0062] Figure 13 , Example diagram of the field industrial computer.

[0063] Figure 14 , Acquisition flow chart.

[0064] Among them, each reference numeral corresponds to the following:

[0065] 1. Rail beam;

[0066] 101. Rail beam main body;

[0067] 102. Fixed part;

[0068] 103. Slide rail;

[0069] 2. Embedded part;

[0070] 201. Embedded support part;

[0071] 202. Connecting part;

[0072] 203. Embedded bracket;

[0073] 3. Rail;

[0074] 301. Vertical (Y) rail;

[0075] 302. Horizontal (X) rail;

[0076] 4. Gantry;

[0077] 401. End beam;

[0078] 402. Main beam (Z) rail;

[0079] 403. Lifting wheel set;

[0080] 404. Lifting wheel

[0081] 405. Movable wheel

[0082] 406. Platform truck traveling mechanism

[0083] 407. Gantry traveling mechanism

[0084] 5. Detection device

[0085] 501. Trolley traveling mechanism

[0086] 502. Manual pan - tilt head

[0087] 503. Image acquisition device

[0088] 601. Sphere

[0089] 602. Pan - tilt head main knob

[0090] 603. Panoramic base knob

[0091] 604. 360° rotation base scale

[0092] 605. Clamp base knob

[0093] 606. Clamp seat

[0094] 607. Sleeve Detailed implementation manners

[0095] After in - depth research, the inventor of the present invention provides a mobile gantry phenotypic acquisition device and a method for plant phenotypic acquisition using the mobile gantry phenotypic acquisition device.

[0096] Mobile gantry phenotypic acquisition device

[0097] The present invention provides a mobile gantry phenotypic acquisition device, which includes: a track part composed of a track beam and embedded parts under the track beam, and a gantry containing a detection device.

[0098] In one or more embodiments, the track part is composed of a track beam and embedded parts under the track beam, and is the basic part for the movement of the mobile gantry.

[0099] In one or more embodiments, the track beam is composed of a track beam main body, fixed components and a slide rail located on the track beam main body.

[0100] In one or more embodiments, the embedded parts are composed of embedded support components, connection components and embedded brackets.

[0101] In one or more embodiments, the embedded bracket is embedded in the concrete foundation.

[0102] In one or more embodiments, the embedded bracket is in a hook shape to increase the stability of the embedded part.

[0103] In one or more embodiments, the number of embedded brackets is 2, 3, 4, 5 or more.

[0104] In one or more embodiments, there are 4 embedded brackets, which are respectively connected to the 4 endpoints of the embedded support member.

[0105] In one or more embodiments, the connecting member connects the embedded support member and the embedded bracket together.

[0106] In one or more embodiments, the embedded support member is used to support the track beam.

[0107] In one or more embodiments, the track section includes: a horizontal (X) track and a vertical (Y) track.

[0108] In one or more embodiments, the track section is composed of a horizontal (X) track and a vertical (Y) track.

[0109] In one or more embodiments, the moving wheels of the mobile gantry can change directions, so as to realize the switching between the horizontal (X) track and the vertical (Y) track.

[0110] In one or more embodiments, the area of the track section is generally consistent with the area to be detected.

[0111] In one or more embodiments, the span of the gantry is consistent with the width of the vertical (Y) track of the track section.

[0112] In one or more embodiments, the area of the track section is an area with a width (Y track) of 20 m and a length (X track) of 80 m.

[0113] In one or more embodiments, the gantry includes: end beams, a main beam (Z) track, lifting wheel sets, lifting wheels, moving wheels, a platform vehicle traveling mechanism and a trolley traveling mechanism; wherein, the end beams and the main beam (Z) track are connected together, the platform vehicle traveling mechanism is connected to one side end beam and is used to control the movement of the gantry on the vertical (Y) track, and the trolley traveling mechanism is connected to the other side end beam and is used to control the movement of the gantry on the horizontal (X) track.

[0114] In one or more embodiments, the trolley traveling mechanism includes a drag chain electric tractor for towing the entire gantry to move on the horizontal (X) track.

[0115] In one or more embodiments, the walking mechanism of the platform vehicle is a rack-type platform vehicle, and the gantry is driven by a chain to move on the vertical (Y) track. In one or more embodiments, the end beams play a supporting role and are generally triangular, for supporting the entire gantry and maintaining the stability of the gantry.

[0116] In one or more embodiments, the main beam (Z) track is parallel to the ground.

[0117] In one or more embodiments, the maximum load capacity of the main beam (Z) track is 0.2t.

[0118] In one or more embodiments, a detection device is detachably connected to the main beam (Z) track.

[0119] In one or more embodiments, the detection device includes a trolley walking mechanism for controlling the movement of the detection device on the main beam (Z) track.

[0120] In one or more embodiments, the trolley walking mechanism of the detection device is detachably and movably connected to the main beam (Z) track.

[0121] In one or more embodiments, the lifting wheel set is connected to the lower end of the end beam for adjusting the height of the main beam track from the ground.

[0122] In one or more embodiments, the lifting wheel set consists of a plurality of lifting wheels.

[0123] In one or more embodiments, the height that the lifting wheel set can lift the gantry is 5m.

[0124] In one or more embodiments, the maximum load capacity of the gantry is 0.2t.

[0125] In one or more embodiments, a moving wheel is connected to the lower end of the lifting wheel set, and this moving wheel enables the entire mobile gantry phenotype acquisition device to move on the track and can switch between the horizontal (X) track and the vertical (Y) track.

[0126] It should be understood that in order to realize the automatic movement of the gantry, cables can be installed on the main beam, end beams, and track parts of the gantry, powered by a power supply to control the movement of the mobile gantry phenotype acquisition device. Further, the walking mechanism of the platform vehicle, the walking mechanism of the trolley, and the walking mechanism of the cart are all electric devices, powered by a power supply to control the movement of the mobile gantry phenotype acquisition device in the Y-axis, X-axis, and Z-axis directions. Usually, the moving distance of the gantry depends on the track. Those skilled in the art can adjust the moving speed and distance of the mobile gantry phenotype acquisition device in the Y-axis, X-axis, and Z-axis directions as needed.

[0127] In one or more specific embodiments, when the area of the track part is a region with a width (Y track) of 20 m and a length (X track) of 80 m, the operating parameters of the mobile gantry phenotype acquisition device include: the operating speed of the X-axis trolley traveling mechanism is 0 - 10 m / min, and the operating speed of the Y-axis platform trolley traveling mechanism is 0 - 10 m / min. Further, the operating motor power of the X-axis trolley traveling mechanism can be: 1.5 KW × 4 (servo motor); the operating motor power of the Y-axis platform trolley traveling mechanism can be: 1 KW × 1 (servo motor).

[0128] It should be understood that GPS positioning (for example, GPS positioning software can be used) can be utilized to adjust the specific positioning of the mobile gantry phenotype acquisition device, so as to accurately control the specific position where the device acquires the phenotype. In a specific embodiment, the positioning accuracy is < 1 mm.

[0129] In one or more embodiments, the detection device includes a trolley traveling mechanism, a manual pan-tilt connected to the trolley traveling mechanism, and an image acquisition device connected to the manual pan-tilt.

[0130] In one or more embodiments, the image acquisition device is detachably connected to the manual pan-tilt, for example, by a card slot connection method or a screw fit connection method. More specifically, in order to quickly detachably connect the image acquisition device, the detection device can also be provided with an adapter plate, and the image acquisition device is quickly fixed and detached by using the card slot on the adapter plate.

[0131] In one or more embodiments, the manual pan-tilt includes a sphere, a pan-tilt main knob, a panoramic base knob, a 360° rotation base scale, a clamp base knob, a clamp, and a sleeve.

[0132] In one or more embodiments, the clamp base knob and the clamp cooperate with each other to adjust the tightness of the clamp and are connected to the trolley traveling mechanism.

[0133] In one or more embodiments, the sphere connects the clamp and the sleeve together, cooperates with the panoramic base knob and the 360° rotation base scale, and is used to rotate the sleeve.

[0134] In one or more embodiments, the pan-tilt main knob is used to rotate the manual pan-tilt.

[0135] It should be understood that in order to achieve the automatic movement of the detection device, cables can be installed on the main beam of the gantry and the manual pan-tilt, powered by a power supply, and the detection device is controlled to move on the main beam (Z-axis) track of the gantry.

[0136] In one or more embodiments, the image acquisition device includes a visible light camera, a multispectral camera, and optionally an illumination device.

[0137] In one or more embodiments, in order to increase the clarity of image acquisition, the image acquisition device may include one or more (such as 2, more than 3, more than 4, more than 5 or more) illumination devices.

[0138] In one or more embodiments, in order to prevent the image acquisition device from overheating due to direct sunlight, the image acquisition device further includes a heat insulation plate.

[0139] In one or more embodiments, in order to increase the accuracy of image acquisition, the image acquisition device may include multiple (such as 2, more than 3, more than 4, more than 5 or more) cameras.

[0140] In one or more specific embodiments, the image acquisition device is 5 RGB cameras and a multispectral camera. This image acquisition device is also called a sensor module.

[0141] In one or more embodiments, in order to increase the accuracy of image acquisition, the image acquisition device may include multiple (such as 2, more than 3, more than 4, more than 5 or more) cameras.

[0142] In one or more embodiments, by building a camera module, developing a communication protocol, intervening in the development of a PLC host control protocol, building a bridge communication, integrating crop phenotype acquisition equipment, establishing a phenotype data collection workstation, planning the acquisition path of crop trait acquisition equipment, and performing high-throughput automatic acquisition of field crop phenotypes.

[0143] In one or more embodiments, the image acquisition device is detachably connected to the clamp of a manual pan-tilt head.

[0144] In one or more embodiments, the lens is detachably connected to the sleeve of a manual pan-tilt head.

[0145] Plant phenotype data acquisition method

[0146] The present invention provides a method for collecting plant phenotype data, including: using the mobile gantry phenotype acquisition device of the present invention to collect plant phenotype data.

[0147] In one or more embodiments, the method includes:

[0148] 1) Controlling the mobile gantry phenotype acquisition device of any embodiment of the present invention to move to a suitable position;

[0149] 2) Use a mobile gantry phenotyping acquisition device to collect plant phenotype pictures;

[0150] 3) Obtain plant phenotype data of the plant by analyzing the plant phenotype pictures obtained in 2).

[0151] In one or more embodiments, the plant includes crops, such as but not limited to rice, wheat, corn, and soybeans.

[0152] In one or more embodiments, the movement described in step 1) includes: movement in the X-axis, Y-axis, and Z-axis directions.

[0153] It should be understood that in step 1), in order to control the movement of the mobile gantry phenotyping acquisition device to a suitable position, a control system (such as a PLC and a field industrial control computer) can be installed in the mobile gantry phenotyping acquisition device of the present invention. Using the software program of the control system, control the movement of the mobile gantry phenotyping acquisition device on the horizontal (X) track and the vertical (Y) track, and detect the movement of the device on the main beam (Z-axis) track of the gantry.

[0154] The present invention also provides an exemplary software program schematic diagram, as Figure 14 shown. The software program includes a server (PLC) main program and a field industrial control computer main program.

[0155] The server (PLC) main program includes: a main thread and a collection thread; the main thread includes: sending a status inquiry instruction to the field industrial control computer to inquire about the position of the mobile gantry phenotyping acquisition device, so as to control whether the movement of the mobile gantry phenotyping acquisition device stops; the collection thread includes: according to the collection path, sending a collection instruction to the field industrial control computer, receiving the feedback from the field industrial control computer, if the picture already exists, remotely download the picture or send a picture clearing instruction to the field industrial control computer, and exit the thread after the collection at the last collection point is completed.

[0156] The field industrial control computer main program includes: receiving the instruction of the server (PLC) main program, creating a collection thread according to the collection instruction, sending an instruction on whether to stop collection to the server main program after collecting the picture, updating the real-time position of the mobile gantry phenotyping acquisition device according to the status inquiry instruction, and clearing the picture according to the picture clearing instruction; among them, the collection thread includes: updating the status and moving the mobile gantry phenotyping acquisition device, updating the status and collecting images, updating the status and image processing, updating the status and saving the picture, and updating the status and exiting the thread.

[0157] In one or more embodiments, the functions of the server (PLC) main program include: controlling the reset, movement, and positioning of the mobile gantry phenotyping acquisition device; controlling the light source switch of the lighting device; controlling the movement speed of the mobile gantry phenotyping acquisition device.

[0158] In one or more embodiments, the plant phenotypes in step 2) include: photos of plant phenotypes, the number of panicles at the heading stage of the plant, and the projected area of a single plant.

[0159] In one or more embodiments, in step 2), an image acquisition device of a mobile gantry phenotype acquisition device is used to acquire pictures of plant phenotypes, and the overlap degree of the pictures acquired at two adjacent sampling points is more than 75%, more than 80%, more than 85%, more than 90% or higher.

[0160] It should be understood that the overlap degree of the photos acquired at two adjacent sampling points determines the accuracy of the analysis in step 3). The higher the overlap degree, the higher the accuracy of the analysis in step 3). Moreover, the accuracy of the analysis in step 3) is also related to the focal length of the image acquisition device and the number of sampling points. The smaller the focal length and the more the number of sampling points, the higher the accuracy of the analysis in step 3). In one or more specific embodiments, when the plant is rice or wheat, since the leaf width of wheat and rice is much smaller than the leaf length, it is only necessary to ensure that there are enough sampling points in the leaf width direction, for example, the number of sampling points in the leaf width direction > 30, > 40, > 50 or more.

[0161] In one or more embodiments, according to the distance (ΔL) that the mobile gantry phenotype acquisition device moves along the X-axis, the distance (ΔW) that the mobile gantry phenotype acquisition device moves along the Y-axis, the speed (v), the image acquisition time (Tshoot), the preprocessing time (Tmod), the saving time (Tsave), and the main program remote download time (Tload), the acquisition time (Ti) of a single sampling point can be calculated, and the formula is as shown in Equation 1:

[0162] Ti = Tmove + Tshoot + Tmod + Tsave + Tload

[0163] Tmove = (ΔL + ΔW) / v

[0164] Tshoot + Tmod + Tsave ≈ 3s

[0165] Tload ≈ 5s(1)

[0166] In one or more embodiments, step 3) includes: using Pix4D software to process the plant phenotype pictures obtained in step 2), and performing merging processing on the pictures to form plant three-dimensional reconstruction pictures.

[0167] In one or more embodiments, the analysis in step 3) includes: rice heading stage analysis and rice single-plant projected area analysis.

[0168] In one or more embodiments, the heading stage analysis includes:

[0169] 3.1) Manual annotation, including annotating the panicle part of the rice phenotype pictures collected in step 2) (for example, using software such as Label studio, labelimg, etc.).

[0170] 3.2) Detection model construction: Use the yolo and rcnn models to train, test, and validate the pictures after manual annotation in 3.1), and construct a detection model for heading stage analysis.

[0171] 3.3) Application of the detection model: Analyze the heading stage according to the detection model in 3.2); for example: According to the collection time point, obtain the dynamic curve of the number of panicles with respect to time, and define the date corresponding to 10% of the maximum number of panicles as the initial heading stage, and combine the rice phenotype pictures collected in step 2) and the detection model in 3.2) to analyze the heading stage.

[0172] In one or more embodiments, the analysis of the projected area of a single rice plant includes:

[0173] 3.1) Denoise the background of the rice phenotype pictures collected in step 2).

[0174] 3.2) Distinguish rice from other plants. For example, use the CIVE vegetation index combined with the Otsu threshold to separate the soil and vegetation. For targets such as duckweed that are relatively small in area compared to the main body of the rice, remove them by setting a threshold.

[0175] 3.2) Obtain the area covered by the rice plants, and calculate the projected area of a single rice plant = the area covered by the rice plants ÷ the number of rice plants.

[0176] The advantages of the present invention are as follows:

[0177] 1) The mobile gantry phenotype acquisition device of the present invention, due to having both horizontal and vertical tracks, can be moved conveniently and extended from one area to another field for phenotype acquisition.

[0178] 2) In the mobile gantry phenotype acquisition device of the present invention, the image acquisition device connected to the manual turntable can include a camera and lighting equipment, and can be adjusted according to the needs of image acquisition.

[0179] 3) In the mobile gantry phenotype acquisition device of the present invention, the detection device is detachable and easy to use.

[0180] The following further elaborates the present invention in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and not to limit the scope of the present invention. The specific conditions not specified in the following embodiments are usually in accordance with conventional conditions or in accordance with the conditions recommended by the manufacturer.

[0181] Example 1, Mobile Gantry Phenotype Acquisition Device of the Present Invention

[0182] The present invention provides a mobile gantry phenotype acquisition device, which includes: a track part (3) composed of a track beam (1) and embedded parts (2) under the track beam, and a gantry (4) containing a detection device (5).

[0183] The track part (3) of the present invention is composed of a track beam (1) and embedded parts (2) under the track beam, and is the basic part for the movement of the mobile gantry. The track beam (1) is composed of a track beam main body (101), fixing components (102) and slide rails (103) located on the track beam main body (101). The embedded parts (2) are composed of embedded support components (201), connecting components (202) and embedded brackets (203). The embedded support component (201) has a length of 300 mm, a width of 200 mm, and a thickness of 10 mm. The length of the embedded bracket (203) is 250 mm. The embedded bracket (203) is embedded in the concrete foundation and is usually in a hook shape to increase the stability of the embedded parts. The concrete foundation has a width of 500 mm and a height of 500 mm. The distance between the embedded parts (2) is 600 mm. The diameter of the hook part of the embedded bracket (203) is 12 mm. The connecting component (202) connects the embedded support component (201) and the embedded bracket (203). The embedded support component (201) is used to support the track beam (1). The track part (3) is divided into a horizontal (X) track and a vertical (Y) track. The moving wheels (405) of the mobile gantry can change directions, so as to realize the switching between the horizontal (X) track and the vertical (Y) track. The area of the track part is usually consistent with the area to be detected. The present invention provides an exemplary track part, that is, an area with a width (Y track) of 20 m and a length (X track) of 80 m.

[0184] The gantry of the present invention includes: end beams (401), main beam (Z) rails (402), lifting wheel sets (403), lifting wheels (404), moving wheels (405), platform truck traveling mechanisms (406) and gantry traveling mechanisms (407); wherein, the end beams (401) and the main beam (Z) rails (402) are connected together. The end beams (401) play a supporting role and are usually triangular, used to support the entire gantry and maintain the stability of the gantry. The span of the end beams is 3500 mm. The main beam (Z) rails (402) are parallel to the ground, and a detection device (5) is detachably connected thereto. The platform truck traveling mechanism (406) is connected to one side end beam (401) and is used to control the movement of the gantry (4) on the vertical (Y) rails. The gantry traveling mechanism (407) is connected to the other side end beam (401) and is used to control the movement of the gantry (4) on the horizontal (X) rails. The gantry traveling mechanism (407) includes a drag chain electric tractor for towing the entire gantry (4) to move on the horizontal (X) rails. The platform truck traveling mechanism (406) is a rack-type platform truck, and drives the gantry (4) to move on the vertical (Y) rails through a chain. The lifting wheel sets (403) are connected to the lower ends of the end beams (402) and are used to adjust the height of the main beam rails (402) from the ground. The span of the lifting wheel sets is 4000 mm, and the lifting height is 50 mm. The lifting wheel sets (403) are composed of multiple lifting wheels (404). A moving wheel (405) is connected to the lower end of the lifting wheel sets (403), and this moving wheel (405) enables the entire gantry to move on the rails and can switch between the horizontal (X) rails and the vertical (Y) rails. It should be understood that in order to achieve the automatic travel of the gantry, cables can be installed on the main beam, end beams and rail parts of the gantry, powered by a power source, to control the movement of the mobile gantry phenotype acquisition device. Further, the platform truck traveling mechanism (406), the gantry traveling mechanism (407) and the trolley traveling mechanism (501) are all electric devices, powered by a power source, to control the movement of the mobile gantry phenotype acquisition device in the Y-axis, X-axis and Z-axis directions. Usually, the moving distance of the gantry is determined according to the rails. Those skilled in the art can adjust the moving speed and distance of the mobile gantry phenotype acquisition device in the Y-axis, X-axis and Z-axis directions as needed.The present invention provides an exemplary gantry, with parameters including: the gantry span is 20 m wide, the lifting height is 5 m, the maximum load capacity is 0.2 t, the X-axis travel length is 85 m (80 m for phenotype acquisition area + 5 m for maintenance area), the X-axis main vehicle power supply drag chain electric traction device, the gantry power supply cable reel, the X-axis main vehicle running speed: 0 - 10 m / min, the X-axis main vehicle running motor power: 1.5 KW × 4 (servo motor), the Y-axis travel length is 20 m, the Y-axis direction rack-type running platform vehicle, the Y-axis trolley running speed: 0 - 10 m / min, the Y-axis trolley running motor power: 1 KW × 1 (servo motor), the Z-axis direction - three-section electric telescopic hanger, the Z-axis direction carrying trolley, the travel is 3.5 m, and the X, Y, and Z six-direction automated control walking. The X, Y, and Z three-axis linkage integration debugging of the gantry meets the running speed of the X-axis main vehicle and the Y-axis trolley: 0 - 10 m / min, and the positioning accuracy: < 1 mm.

[0185] The trolley running mechanism (501) of the detection device (5) of the present invention is detachably and movably connected to the main beam (Z) track (402). The detection device (5) includes a trolley running mechanism (501), a manual pan-tilt head (502) connected to the trolley running mechanism (501), and an image acquisition device (503) connected to the manual pan-tilt head (502). The manual pan-tilt head (502) includes a sphere (601), a pan-tilt main knob (602), a panoramic base knob (603), a 360° rotation base scale (604), a clamp base knob (605), a clamp (606) and a sleeve (607). The clamp base knob (605) and the clamp (606) cooperate with each other to adjust the tightness of the clamp and are connected to the trolley running mechanism (501). The sphere (601) connects the clamp (606) and the sleeve (607) together and cooperates with the panoramic base knob (603) and the 360° rotation base scale (604) to rotate the sleeve (607). The pan-tilt main knob (602) is used to rotate the manual pan-tilt head (502). The image acquisition device (503) includes a visible light camera, a multispectral camera, and a lighting device. To increase the accuracy of image acquisition, the image acquisition device (503) may include multiple (such as 2, more than 3, more than 4, more than 5 or more) cameras. To increase the clarity of image acquisition, the image acquisition device (503) may include one or more (such as 2, more than 3, more than 4, more than 5 or more) lighting devices. Through the construction of the camera module and the development of the communication protocol, the intervention in the development of the PLC host computer control protocol, the construction of the bridge communication, the integration of the crop phenotype acquisition equipment, the establishment of the phenotype data collection workstation, the planning of the acquisition path of the crop trait acquisition equipment, and the high-throughput automatic acquisition of the field crop phenotype. The image acquisition device (503) is detachably connected to the clamp (606) of the manual pan-tilt head (502), and the lens is detachably connected to the sleeve (607) of the manual pan-tilt head (502). Exemplary cameras and lenses are respectively as Figure 8 , 9 shown.

[0186] Example 2. Method for collecting plant phenotype data

[0187] This embodiment provides a method for collecting plant phenotype data, including: using the mobile gantry phenotype acquisition device of the present invention to collect plant phenotype data.

[0188] Specifically, the method includes:

[0189] 1) Controlling the mobile gantry phenotype acquisition device of the present invention to move to a suitable position;

[0190] 2) Using the detection device to collect plant phenotype pictures;

[0191] 3) Obtain the plant phenotype data of the plant by analyzing the plant phenotype pictures obtained in 2).

[0192] More specifically, the mobile gantry phenotype acquisition device in step 1) includes an RGB camera and a multispectral camera. Step 2) includes: adding exif data to the picture, writing the information required for three-dimensional reconstruction of the camera, transmitting the picture back to the server using a wireless network, receiving and saving the picture on the server, and recording the picture saving path.

[0193] Example 3, Parameters of the Plant Phenotype Data Acquisition Method

[0194] 1. Overhead Crane Control Development

[0195] The first step of fully automatic acquisition is to intervene in the overhead crane control system. In this example, the central unit of the gantry (4) of the mobile gantry phenotype acquisition device is a PLC, model: Siemens S7-1200. This PLC has a built-in Ethernet interface and can be easily connected to the local area network.

[0196] Write a C# upper computer through the existing API interface, which can read and write the memory of the PLC. Communicate with Jingjiang Hoisting Equipment Factory to obtain the register address corresponding to the PLC function point. Modify the value in the function register to achieve the movement control of the gantry.

[0197] Functions to be realized:[[]]

[0198] Reset, movement, positioning (get / set) of the overhead crane

[0199] Specified IO control (get / set of the external light source switch)

[0200] Overhead crane movement speed setting (get / set)

[0201] 2. Sensor Module Development

[0202] A Camera Selection

[0203] The core of the image acquisition device (503) is 5 globally exposed RGB cameras and multispectral cameras.

[0204] Global exposure ensures the synchronization of the camera acquisition process, avoiding defects such as picture tearing and blurring caused by camera vibration and plant disturbance during camera shooting.

[0205] Considerations for lens selection: Since the leaf width of wheat and rice is much smaller than the leaf length, as long as there are enough sampling points in the leaf width direction of the image, sufficient details can be obtained for three-dimensional reconstruction. Currently, there are two specifications of available lenses for this camera, 35mm and 50mm. The detailed information corresponding to the lenses is shown in Table 1.

[0206] Table 1 Precision, Field of View, and Number of Sampling Points of Two Lenses at Different Distances

[0207]

[0208] As can be seen from Table 1:

[0209] If it is necessary to satisfy that within the working distance (2.5 - 5 m), the field of view covers at least one ridge (1.5 * 0.5 m), a lens with a focal length of 35 mm should be selected. At this time, the number of sampling points in the leaf width direction > 30, which is sufficient for 3D model reconstruction;

[0210] If higher-precision reconstruction is pursued, a 50-mm lens is used. At this time, the number of sampling points in the leaf width direction > 44, and the model has higher precision, but the required acquisition interval needs to be reduced synchronously to ensure the overlap degree of adjacent images.

[0211] Considering that the higher the precision of plant 3D reconstruction, the better, a 50-mm focal length lens is determined to be used. By reducing the shooting interval, the overlap degree of adjacent photos is ensured to be more than 75%. The appearance of the lens is as Figure 2 shown.

[0212] B Camera Pose Adjustment Bracket

[0213] 3D reconstruction tests were carried out on the software Pix4D. The test results show that this software can complete 3D reconstruction without knowing the exact pose of the camera. It is speculated that this software can identify the same feature points from multiple pictures and automatically calculate the rigid body transformation matrix of the corresponding cameras of two photos, thereby realizing 3D reconstruction.

[0214] Therefore, the camera does not need to be installed on the electric pan-tilt to precisely adjust the pose. It can be installed on the main knob (602) of the pan-tilt, which is convenient for the operator to adjust the camera pose at any time. After adjustment, lock the main knob (602) of the pan-tilt. The structure of the pan-tilt is as Figure 10 shown.

[0215] C Module Structure Design

[0216] The module is divided into upper and lower parts. The upper layer is the power supply and switch part, and the lower layer is the sensor module. The outer layer is enclosed by a sealed and waterproof box.

[0217] Input / Output (IO) Part: The upper-layer input wire harness is a 220V power line and a single network cable, which are connected to the module through a waterproof interface, and the waterproof level needs to reach IP67. The interface schematic diagram is as Figure 11 shown.

[0218] The module's upper layer leads out a network cable and a power line, which are directly connected to the camera after routing inside the module;

[0219] The lower layer of the module is designed as a hollow cage structure, which reduces weight and leaves enough space for camera attitude adjustment;

[0220] The bottom of the module uses high-transparency glass as the visible light window of the camera. The glass can be opened from below to facilitate the adjustment of the camera attitude. Finally, the glass light window is waterproof-sealed.

[0221] D Design of the module adapter board

[0222] 1) In order to better control the camera, a module adapter board is designed to be connected to the upper end of the camera of the image acquisition device (3);

[0223] 2) The adapter board is designed with a module bayonet, which can quickly plug the sensor module onto the adapter board;

[0224] 3) The adapter board is designed with light source mounting holes for fixing the external light source;

[0225] 4) The adapter board is designed with a heat insulation board, which is connected inside the image acquisition device (3) to prevent the module camera from being damaged due to overheating caused by direct sunlight on the module.

[0226] E Internal parameter calibration of the camera module

[0227] The Pix4D software can estimate the internal parameters and distortion coefficients of the camera from the input images and save the internal parameters and distortion coefficients to the camera model, thereby automatically calibrating and stitching the pictures. This software supports the combined processing of multiple different camera pictures, and it is necessary to perform matching calculations on the collected pictures of different cameras.

[0228] In order to reduce the difficulty of image matching calculation, after the image data is collected, the original image can be corrected for distortion using the pre-calibrated internal parameters and distortion coefficients of the camera and then handed over to Pix4D for processing, thereby increasing the image matching accuracy and reducing the 3D reconstruction time.

[0229] In this embodiment, a checkerboard and the Matlab camera calibration tool are used to calibrate the internal parameters and distortion coefficients of the camera, and the calibration results are stored in the configuration file for the correction program to call automatically. An example of the calibration results of a certain camera's parameters is as Figure 12 shown.

[0230] 3. Development of the field industrial computer

[0231] A Installation of the industrial computer

[0232] Furthermore, a PLC control cabinet can be set below the end beam (401), and the industrial computer is installed in the PLC control cabinet and can be quickly installed through the adapter board. With the help of the PLC control cabinet shell and the power supply line, additional wiring work is avoided, and it can better protect against rain and lightning.

[0233] B Interfaces of the industrial computer

[0234] Ethernet Interface: The industrial computer is equipped with two gigabit Ethernet interfaces. Network interface 1 (LAN1) is wired to the camera module to stably collect camera images; network interface 2 (LAN2) is wired to the bridge, and then through the WIFI network of the wireless bridge, it is bridged to the local area network where the server is located.

[0235] Serial Port: The industrial computer is equipped with an RS232 serial port, which can be used for serial communication of sensors.

[0236] USB Interface: By using a USB-to-Ethernet module, it can be used to connect to the Ethernet interface of the PLC.

[0237] Industrial Computer Software Development

[0238] 1) Host Computer Development for RGB Camera

[0239] ① Based on the official SDK of Hikvision cameras, develop a camera image acquisition program;

[0240] ② Based on the Opencv image recognition algorithm and call the camera calibration parameters to correct the distortion of the captured images;

[0241] ③ Write exif information to the distortion-corrected images.

[0242] 2) Development of Image Wireless Transmission Program

[0243] ① Solution 1: Use FTP communication. The advantage of FTP communication is higher efficiency in large file transfer. By deploying an FTP server on the industrial computer and an FTP client on the server side in the computer room, file data can be transferred between the two PCs.

[0244] ② Solution 2: Windows shared folder. By using a shared folder, the folder on the industrial computer that stores camera images can be shared to the local area network. In the main program on the server PC side, write a file stream reading program, input the username, password, and file path, and then the files in the shared folder can be accessed.

[0245] 3) Host Computer Development for PLC Control

[0246] Based on the third-party API (c#) of Siemens S7-1200 PLC, referring to the PLC point list provided by Jingjiang Crane Equipment Factory, develop the host computer program for the PLC to realize the functions of gantry reset, movement, positioning, and light source switching.

[0247] 4) System Framework Development

[0248] The above three software modules need to provide instruction interfaces. When the main program framework receives valid instructions, it will automatically call the corresponding software modules to achieve the specified functions.

[0249] Meanwhile, the main program framework needs to respond to the inquiry instructions from the remote end, feedback the execution status of each software module, and provide status flag signals for the main program running on the remote PC.

[0250] 4. Bridge Development

[0251] A Bridge Installation

[0252] Wireless bridges are installed in pairs: divided into the "camera end" and the "VCR end", and the two need to be installed opposite to each other to ensure signal strength. The following specific work needs to be done:

[0253] It is necessary to design an adapter to fix the "camera end" on the girder of the gantry and the "VCR end" outside the server hut;

[0254] The LAN port of the "camera end" of the bridge is connected to the LAN port of the field industrial computer through a network cable; the LAN port of the "VCR end" of the bridge is connected to the LAN port of the server through a network cable;

[0255] B Bridge Bandwidth Test

[0256] Static Test

[0257] Take 5 to 10 preset points in the length direction of the gantry, and measure the bandwidth of the gantry at each preset point through wireless network file copying.

[0258] Dynamic Test

[0259] During the movement of the gantry, measure the fluctuation of the bandwidth and determine the network stability according to the number of lost packets.

[0260] 5 Server Software Development

[0261] A Construction of the Main Program Function Framework

[0262] ① Design the "Function Instruction Set" to send function instructions to the main program of the field industrial computer to command it to implement complex functions such as image acquisition, processing, vehicle control, and picture transmission.

[0263] ② Design the "Status Instruction Set" to send inquiry instructions to the main program of the field industrial computer to obtain the running status of all subsystems to optimize the system operation process.

[0264] B Remote Copying and Clearing of Images

[0265] Remotely download the picture cache data in the field industrial computer, and delete the picture cache after copying to avoid hard disk storage overflow in the field industrial computer.

[0266] Save the downloaded pictures to the specified folder according to the naming rules, and write information such as picture name, save path, save time, scanning batch, and sensor ID to the Excel file.

[0267] C Path Planning

[0268] The software can freely plan multiple experimental areas. By inputting information such as the scanning start point, end point, interval, and height of each test area, the software can automatically calculate the optimal acquisition path and send acquisition instructions to the field industrial computer according to the path, thus realizing the acquisition process of collecting images.

[0269] D Acquisition Process Monitoring

[0270] 1) Real-time display of gantry position

[0271] 2) Real-time update of the status of each acquisition area: completed, in acquisition, to be acquired

[0272] 3) Display of acquisition progress

[0273] E Image Classification and Extraction

[0274] When using Pix4D to perform high-precision reconstruction on a specified area, all images of the specified area need to be extracted from the hard disk. This function can be achieved by developing UI software. In the software, specify the scanning batch and GPS range of the photos, so as to extract all photos that meet the search conditions and save them to a specified folder for 3D reconstruction.

[0275] Example 4, Plant Phenotype Data Acquisition Process and Efficiency Estimation

[0276] 1. Acquisition Process

[0277] In this example, the acquisition process is divided into two parts: the server main program and the field industrial computer main program. The former is responsible for planning and supervising the acquisition process, and the latter is responsible for performing underlying sensor communication, control, etc. The specific acquisition process is as Figure 14 shown.

[0278] 2. Single-point Acquisition Time

[0279] The instruction communication time between the two programs can be ignored. The main time consumption is in parts such as gantry movement, image acquisition, image processing, image storage, and remote download. The system acquisition time is shown in Formula 1.

[0280] In one or more implementation schemes, according to (△L), (△W), speed (v), the acquisition time (Ti) of a single sampling point can be calculated,

[0281] Ti = Tmove + Tshoot + Tmod + Tsave + Tload

[0282] Tmove = (ΔL + ΔW) / v

[0283] Tshoot + Tmod + Tsave ≈ 3s

[0284] Tload ≈ 5s(1)

[0285] Wherein, ΔL and ΔW are respectively the moving distances of the mobile gantry phenotype acquisition device on the X-axis (unit: m) and the moving distances of the mobile gantry phenotype acquisition device on the Y-axis (unit: m), v represents the traveling speed, the image acquisition time (Tshoot) includes the preprocessing time (Tmod) and the saving time (Tsave), the image acquisition time (Tshoot) totals approximately 3s, and the main program remote download time (Tload) is approximately 5s.

[0286] From this, it is estimated that: when the traveling crane moves at the maximum speed (10m / min), the minimum acquisition time for each point is:

[0287] T ≈ 6 * ΔD + 8(2)

[0288] Where ΔD is the moving interval, in meters.

[0289] 3 Estimation of the total acquisition time

[0290] Estimate the time taken for image acquisition of the entire experimental field (80 * 20m).

[0291] Considering that the RGB camera uses a 50mm focal length lens and the effective field of view angle is 26°×26°, in order to ensure the image matching accuracy, it is calculated according to the 70% overlap of adjacent photos. Then the total number of sampling points and the total acquisition time at different working distances are shown in Table 2.

[0292] Table 2 Sampling time estimation table

[0293]

[0294] Note: Table 2 is an estimate of the acquisition time for the entire experimental field. The actual acquisition time is related to the planting density of the plants. When the plants are sparse, the acquisition time can be reduced by reasonably planning the path and avoiding the acquisition of blank areas.

[0295] According to the number of acquisition points in Table 2, it can be known that: if 5 pictures are saved for each acquisition point, at least 109GB (3350 sampling points) of picture data is required, and at most 434GB (13400 sampling points) of storage space is required. It is recommended that the server storage space be greater than 5TB to support the storage requirements of 10 - 40 rounds of complete acquisition data. The actual storage space is related to the planting density of the plants. This example is an estimated value when the entire experimental field is fully sampled.

[0296] Example 5, Application of the mobile gantry phenotype acquisition device

[0297] In this embodiment, taking rice as an example, the mobile gantry phenotypic acquisition device of the present invention is used for image acquisition and processing. The mobile gantry phenotypic acquisition device of the present invention can cover a scanning area of 1,600 square meters. In a rice field (80 meters long and 20 meters wide), 481 varieties of single rice plants are planted, with 4 replicates for each variety, and growth phenotypic data is collected.

[0298] By taking photos with an RGB camera every day and performing photo processing and analysis, a series of phenotypic traits during the growth period of rice can be obtained, including the number of panicles at the panicle stage of rice and the projected area of a single rice plant.

[0299] For manual annotation, open-source software such as Label studio and labelimg is used to manually select the rice panicles. The manually annotated dataset is divided into a training set, a validation set, and a test set with a ratio of 8:1:1. A target detection model, such as the yolo series and rcnn series models, is trained on the training set, the parameters of the target detection model are selected on the validation set, and finally, the accuracy of the model in identifying rice panicles is evaluated on the test set.

[0300] After obtaining the rice panicle detection model, since the mobile gantry phenotypic acquisition device is used to photograph rice on a daily basis, a dynamic curve of the number of rice panicles at the panicle stage with respect to time can be obtained. The various heading stages of rice can be obtained by the corresponding dates given the percentage of the maximum number of panicles. The date corresponding to 10% of the maximum number of panicles is defined as the initial heading stage. After comparing with the initial heading stage traits investigated manually, the R 2 reaches 0.9387, indicating that the predicted heading stage results are highly consistent with the actual results.

[0301] In addition, the mobile gantry phenotypic acquisition device of the present invention can also obtain the projected area of a single rice plant. Exemplary methods include: First, denoise the background of the phenotypic data of the projected area of a single rice plant collected. Use the CIVE vegetation index combined with the Otsu threshold to separate the soil and vegetation. For targets such as duckweed with relatively small areas compared to the main body of the rice plant, they are removed by setting a threshold, and then the area covered by the rice plant is calculated.

[0302] The above-described embodiments only represent several implementation manners of the present invention. Their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention should be subject to the appended claims. At the same time, all the documents mentioned in the present invention are cited as references in this application, as if each document is cited separately as a reference.

Claims

1. A mobile gantry phenotype acquisition device, characterized in that, The device includes: a track part composed of a track beam and embedded parts under the track beam, and a gantry with a detection device; The track part includes: a horizontal track and a vertical track; The gantry includes: end beams, main beam tracks, lifting wheel sets, lifting wheels, moving wheels, platform vehicle traveling mechanisms, and gantry traveling mechanisms; among them, the main beam tracks are parallel to the ground, and a detection device is detachably and movably connected to the main beam tracks, and the detection device includes a trolley traveling mechanism for controlling the movement of the detection device on the main beam tracks; the end beams and the main beam tracks are connected together; the moving wheels can change directions so as to realize the switching between the horizontal track and the vertical track; the platform vehicle traveling mechanism is connected to one side of the end beam for controlling the movement of the gantry on the vertical track; the gantry traveling mechanism is connected to the other side of the end beam for controlling the movement of the gantry on the horizontal track.

2. The mobile gantry phenotype acquisition device according to claim 1, wherein The lifting wheel sets are connected to the lower ends of the end beams for adjusting the height of the main beam tracks from the ground; Preferably, the lifting wheel sets are composed of multiple lifting wheels.

3. The mobile gantry phenotypic acquisition device according to claim 1, wherein The detection device further includes an adapter plate, and the detection device is fixed and disassembled by using the card slots on the adapter plate.

4. The mobile gantry phenotype acquisition device according to claim 1, wherein The track beam is composed of a track beam main body, fixing components, and slide rails located on the track beam main body.

5. The mobile gantry phenotype acquisition device according to claim 1, wherein, The embedded parts are composed of embedded support components, connecting components, and embedded brackets, wherein the connecting components connect the embedded support components and the embedded brackets together.

6. The mobile gantry phenotype acquisition device according to claim 5, characterized in that, There are 4 embedded brackets, which are respectively connected to 4 connecting components of the embedded support components.

7. The mobile gantry phenotypic acquisition device according to any one of claims 1-6, characterized in that, The detection device includes a trolley traveling mechanism, a manual pan-tilt connected to the trolley traveling mechanism, and an image acquisition device connected to the manual pan-tilt; Preferably, the image acquisition device is detachably connected to the manual pan-tilt.

8. The mobile gantry phenotype acquisition device according to claim 7, wherein The manual pan-tilt includes a sphere, a main pan-tilt knob, a panoramic base knob, a 360° rotation base scale, a clamp base knob, a clamp, and a sleeve; the clamp base knob and the clamp cooperate with each other to adjust the tightness of the clamp and are connected to the trolley traveling mechanism; the sphere connects the clamp and the sleeve together and cooperates with the panoramic base knob and the 360° rotation base scale to rotate the sleeve; the main pan-tilt knob is used to rotate the manual pan-tilt.

9. The mobile gantry phenotype acquisition device according to claim 7, characterized in that The image acquisition device includes a camera and an optional lighting device; Preferably, the image acquisition device further includes a heat insulation plate.

10. Application of the mobile gantry phenotypic acquisition device according to any one of claims 1-9 in collecting plant phenotypes; Preferably, the plant is a crop; More preferably, the plants include: rice, wheat, corn, and soybean.

11. A method for collecting plant phenotype data, characterized in that The method includes: using the mobile gantry phenotypic acquisition device according to any one of claims 1-9 to collect plant phenotypic data; Preferably, the method includes: 1) Controlling the mobile gantry phenotypic acquisition device according to any one of claims 1-9 to move to a suitable position; 2) Using the mobile gantry phenotypic acquisition device to collect plant phenotypic pictures; 3) Obtaining plant phenotypic data of the plant by analyzing the plant phenotypic pictures obtained in 2). Preferably, the plant is a crop; more preferably, the plant includes: rice, wheat, corn, and soybean.

12. The method for collecting plant phenotype data according to claim 11, wherein the movement in step 1) includes: movement in the X-axis, Y-axis, and Z-axis directions; the plant phenotypes in step 2) include: pictures of plant phenotypes, the number of panicles at the heading stage of the plant, and the projected area of a single plant; in step 2), an image acquisition device of the mobile gantry phenotype acquisition device is used to collect pictures of plant phenotypes, and the overlap degree of the pictures collected at adjacent two sampling points is above 75%, above 80%, above 85%, or above 90%; in step 2), an image acquisition device of the mobile gantry phenotype acquisition device is used to collect pictures of plant phenotypes, and the number of sampling points > 30, > 40, or > 50; step 3) includes: using Pix4D software to process the plant phenotype pictures obtained in step 2), and performing merging processing on the pictures to form three-dimensional reconstructed pictures of the plant.

13. The method for collecting plant phenotype data according to claim 11, wherein in step 1), a server (PLC) and a field industrial control computer are included in the mobile gantry phenotype acquisition device to control the movement of the mobile gantry phenotype acquisition device; Preferably, the main program of the server (PLC) includes: a main thread and a collection thread; more preferably, the main thread includes: sending a status inquiry instruction to the field industrial control computer to inquire about the position of the mobile gantry phenotype acquisition device, so as to control whether the movement of the mobile gantry phenotype acquisition device stops; the collection thread includes: according to the collection path, sending a collection instruction to the field industrial control computer, receiving the feedback from the field industrial control computer, if the picture already exists, the picture can be remotely downloaded or a picture deletion instruction is sent to the field industrial control computer, and exiting the thread when the collection at the last collection point is completed; Preferably, the main program of the field industrial control computer includes: receiving the instruction of the main program of the server (PLC), creating a collection thread according to the collection instruction, sending an instruction on whether to stop collection to the main program of the server after collecting pictures, updating the real-time position of the mobile gantry phenotype acquisition device according to the status inquiry instruction, and deleting pictures according to the picture deletion instruction; more preferably, the collection thread includes: updating the status and moving the mobile gantry phenotype acquisition device, updating the status and collecting images, updating the status and image processing, updating the status and saving pictures, and updating the status and exiting the thread.

14. The method for collecting plant phenotype data according to claim 11, wherein the analysis in step 3) includes: analysis of the heading stage of rice and analysis of the projected area of a single rice plant; Preferably, the analysis of the heading stage includes: 3.1) Manual annotation, including annotating the panicle part of the pictures of the rice phenotypes collected in step 2); 3.2) Detection model construction: Using yolo and rcnn models to train, test, and verify the pictures manually annotated in 3.1) to construct a detection model for the analysis of the heading stage; 3.3) Application of the detection model: Analyzing the heading stage according to the detection model in 3.2); for example: according to the collection time point, obtaining the dynamic curve of the number of panicles with respect to time, defining the date corresponding to 10% of the maximum number of panicles as the initial heading stage, and combining the pictures of the rice phenotypes collected in step 2) and the detection model in 3.2) to analyze the heading stage; Preferably, the analysis of the projected area per rice plant includes: 3.1) Denoising the background of the rice phenotype images collected in step 2); 3.2) Distinguishing rice from other plants. For example, the CIVE vegetation index combined with the Otsu threshold is used to separate soil and vegetation. For targets such as duckweed with relatively small areas compared to the main body of the rice, they are removed by setting a threshold; 3.2) Obtain the area covered by the rice plants, and calculate the projected area per rice plant = the area covered by the rice plants ÷ the number of rice plants.

Images