A high-throughput hydroponic crop 3D time-series imaging system and application method

Abstract

The present invention discloses a high-throughput hydroponic crop 3D time-series imaging system and an application method. The imaging system integrates a suspension rail, a suspension, an imaging device, a hydroponic device and a liquid replacement device; the application method includes an imaging method, a crop 3D reconstruction, a phenotypic extraction and a visualized prediction of crop growth. The liquid replacement device proposed in the present invention can achieve an unchanged root configuration, which is convenient for analyzing root growth changes; the imaging device can automatically and efficiently collect multi-view images of crops, and achieve high-throughput, three-dimensional analysis of the whole crop including the above-ground and underground parts; at the same time, the hydroponic device designed by the present invention is convenient for multi-view imaging to achieve three-dimensional phenotypic analysis. In terms of method, the present invention designs a technical process for crop 3D reconstruction, phenotypic extraction and visualized growth prediction, which realizes the intuitive display of crop three-dimensional phenotypic analysis and future morphological prediction. The invention is flexible and scalable, and can meet the needs of different application scenarios.

Description

Technical Field

[0001] The present invention belongs to the fields of agricultural automation equipment technology, computer vision technology and biometric identification technology, and specifically relates to a high-throughput hydroponic crop 3D time-series imaging system and an application method. Background Art

[0002] With the rapid development of modern agricultural technology, improving crop yields, optimizing planting management, and precision breeding have become important directions of agricultural research. Among them, high-throughput crop phenotyping technology plays a key role in crop genetic improvement, growth environment adaptability research, and yield estimation. However, traditional crop phenotyping methods often rely on manual measurement, which is not only time-consuming and labor-intensive, but also difficult to achieve continuous and non-destructive monitoring of the crop growth process. Therefore, it is particularly important to develop a high-throughput imaging device and its application method that can efficiently and accurately obtain the three-dimensional morphological structure of crops and their dynamic changes.

[0003] In recent years, the application of 3D image processing technology and computer vision in the agricultural field has become increasingly widespread. 3D reconstruction technology can reconstruct the 3D model of crops based on image information from multiple perspectives, thereby achieving an accurate description of the morphological structure of crops. Image processing technology and computer vision can further analyze the 3D model of crops and extract the 3D phenotypic characteristics of crops, such as key parameters such as plant height, leaf area, stem thickness, root length, and root dispersion. At the same time, mathematical models can be established based on key phenotypic parameters to achieve visual growth prediction of crops. The combination of these technologies provides strong support for the visualization of crop growth processes, the prediction of growth trends, and accurate decision-making for genetic improvement.

[0004] However, the existing high-throughput crop imaging devices and analysis methods still have some shortcomings. On the one hand, traditional imaging devices often cannot simultaneously meet the high-throughput, three-dimensional analysis requirements of the whole crop including the aboveground and underground parts, which limits its application in large-scale field trials. For example, Publication No. CN115239882A proposes a three-dimensional reconstruction method of crops based on weak-light images to achieve crop phenotypic analysis in weak-light scenes, but this method has low efficiency during image acquisition, is difficult to apply to high-throughput scenes, and does not consider the root structure of the underground part of the crop. Publication No. CN111165228A proposes an environmentally controllable device for cultivating crops to achieve the analysis of the relationship between the phenotype of the whole crop including the aboveground and underground parts and environmental factors, but this method cannot achieve three-dimensional phenotypic analysis compared to traditional methods, and it is difficult to achieve high-throughput and low-cost requirements. On the other hand, the existing three-dimensional reconstruction and phenotypic extraction methods are mostly limited to two-dimensional space, lacking versatility and adaptability to different crop types, different growth environments, and different growth stages. For example, publication number CN111165228A proposed a mobile cabin for crop phenotypic extraction and analysis, which can extract phenotypic information of plants including the root system and analyze it in combination with environmental information. However, this method can only extract two-dimensional phenotypes and has certain limitations in crop morphological analysis. In terms of crop three-dimensional reconstruction methods, publication number CN115564109A proposed a fruit tree growth prediction method based on an improved convolutional neural network, but this method is difficult to expand to other crop varieties, has certain limitations, and is difficult to meet the needs of high throughput.

[0005] In response to the above problems, the present invention proposes a high-throughput hydroponic crop 3D time-series imaging system and a method for three-dimensional reconstruction, phenotypic extraction and visual growth prediction of the entire crop including the root system, aiming to solve the shortcomings of existing crop imaging, three-dimensional phenotypic extraction and analysis methods. The system uses hydroponic technology to provide a stable growth environment for crops, and combines three-dimensional imaging technology to achieve continuous and non-destructive monitoring of the crop growth process. At the same time, through advanced image processing algorithms and computer vision technology, the acquired three-dimensional images are analyzed to extract the phenotypic characteristics of the crops, and based on these characteristics, the growth trend of the crops is predicted and visualized. The invention has broad application prospects and can be applied to agricultural scientific research, agricultural production, agricultural automation and other fields to promote the rapid development of modern agriculture. Summary of the invention

[0006] In crop phenotyping research, due to the opacity of the soil environment, most existing methods find it difficult to analyze root phenotypes; the reliance of 3D reconstruction technology on multiple perspectives brings more challenges to the design of crop cultivation methods and imaging devices; since crop phenotyping analysis often relies on specific devices, it is difficult to achieve phenotyping in high-throughput scenarios; at the same time, with the continuous development of crop growth monitoring and precision breeding, existing visualization technology is difficult to achieve an intuitive display of crop 3D morphological growth predictions, limiting its application in agricultural research and production.

[0007] To solve the above problems, the present invention provides the following technical solutions, and provides a high-throughput hydroponic crop 3D time-series imaging system and application method, as follows.

[0008] A high-throughput hydroponic crop 3D time-series imaging system, characterized in that the device comprises a suspension rail (1), a suspension (2), an imaging device (3), a hydroponic device (4) and a liquid replacement device (5);

[0009] The suspension guide rail (1) comprises two longitudinal guide rails parallel to each other and arranged horizontally on a plane;

[0010] The suspension (2) comprises two support rods and a transverse guide rail, the two support rods are respectively vertically mounted on the suspension guide rail, the transverse guide rail is mounted on the upper ends of the two support rods and spans between two longitudinal guide rails of the suspension guide rail; the suspension can be flexibly moved along the longitudinal guide rail;

[0011] The imaging device (3) comprises a suspension connecting member (6), an inner telescopic arm (7), an outer telescopic arm (8), an annular guide rail (9), a camera (10) and a background plate (11); the imaging device is connected to the suspension (2) via the suspension connecting member (6) and is located below the suspension, and can be flexibly moved along the transverse guide rail on the suspension; the upper ends of the inner telescopic arm (7) and the outer telescopic arm (8) are connected to the suspension connecting member (6); the lower end of the outer telescopic arm (8) is connected to the annular guide rail (9); the camera (10) and the background plate (11) are mounted on the annular guide rail (9), the camera and the background plate are mounted opposite each other, the centers of the two are located on the same straight line with the center of the annular guide rail, and the two can rotate synchronously and uniformly at the same speed along the annular guide rail around the central axis of the hydroponic device (4) to obtain crop images;

[0012] The hydroponic device (4) is located in the area between the two longitudinal rails of the suspension rail (1), and includes a plurality of hydroponic device units arranged in an array; each hydroponic device unit includes a plant sample (12), a transparent cylindrical growth tank (13), a plate (14), a sponge (15), a nutrient solution (16), a light shield (17) and a placement table (18); every two hydroponic device units are separated by a certain distance in the horizontal and vertical directions, so as to facilitate the imaging operation of the imaging device (3);

[0013] The liquid replacement device (5) comprises a siphon (20), a container (21) containing new nutrient solution, and a container (22) for collecting old nutrient solution, and is used to regularly replace the nutrient solution (16) of the hydroponic device (4).

[0014] More specifically, the suspension connecting member (6) of the imaging device is used to drive the imaging device (3) to move laterally along the suspension (2); the lower end of the inner telescopic arm (7) is used to remove and install the light shield (17) of the hydroponic device; the outer telescopic arm (8) is used to drive the annular guide rail (9) to move up and down; the annular guide rail (9) is centered on the hydroponic device (4) so ​​that the distance between the camera and the crop remains unchanged; the camera (10) uses manual focus and video shooting to obtain an image of the entire crop; the inner side of the background plate (11) is painted black to facilitate the acquisition of an image of the entire crop.

[0015] More specifically, in the hydroponic device (4), the flat plate (14) is of square design, and its outer side is fixedly connected to the transparent cylindrical growth tank (13) by interference fit with the inner wall of the transparent cylindrical growth tank (13); the flat plate (14) is provided with a fixing hole (19) penetrating from top to bottom; the sponge (15) is fixedly connected to the flat plate (14) by interference fit with the inner wall of the fixing hole (19); the sponge (15) is located at the center of the square of the flat plate (14); the liquid level of the nutrient solution (16) is flush with the flat plate (14); The transparent cylindrical growth tank (13) is used to contain components and provide a space for plant growth; the flat plate (14) is used to fix the plant at the center of the hydroponic device (4); the sponge (15) is used to provide water to the plant seed rootlets during the germination period and to fix the base of the plant root system in the later stage; the light shield (17) is wrapped around the outer periphery of the transparent cylindrical growth tank (13) in a detachable manner to prevent the external light from affecting the root system and is removed when the crop is imaged; the flat plate (14), sponge (15) and light shield (17) are made of black material.

[0016] More specifically, in the liquid replacement device (5), the liquid level of the container (21) containing the new nutrient solution is higher than the liquid level of the nutrient solution (16) in the hydroponic device (4); and the liquid level of the nutrient solution (16) in the hydroponic device (4) is higher than the liquid level of the container (22) for collecting the old nutrient solution.

[0017] More specifically, the siphon tubes (20) of the liquid replacement device (5) include two siphon tubes, one of which is connected to the top of the hydroponic device (4) and the container (21) containing the new nutrient solution, and is used to siphon the new nutrient solution into the hydroponic device (4), and the other siphon tube is connected to the bottom of the hydroponic device (4) and the container (22) for collecting the old nutrient solution, and is used to siphon the old nutrient solution in the hydroponic device (4) out into the container (22) for collecting the old nutrient solution; the two siphon tubes have the same siphon power, and can discharge the old nutrient solution while inputting the new nutrient solution, thereby ensuring that the liquid level of the nutrient solution (16) in the hydroponic device (4) remains unchanged during the water replacement, so that the root system configuration does not change due to the liquid replacement.

[0018] The present invention also relates to an application method of a high-throughput hydroponic crop 3D time-series imaging system, which uses the aforementioned high-throughput hydroponic crop 3D time-series imaging system to perform imaging, and then performs crop 3D reconstruction, phenotype extraction and crop growth visualization prediction.

[0019] The imaging steps are specifically as follows:

[0020] Step S11, the suspension (2) moves along the suspension guide rail (1) to the first row of the hydroponic device array, and the center axis of the suspension and the center of the hydroponic device unit in the first row are on the same vertical plane;

[0021] Step S12, the imaging device (3) moves along the suspension (2) to the top of the first hydroponic device unit, and the center of the imaging device and the center of the first hydroponic device unit are located on the same vertical line;

[0022] Step S13, the inner telescopic arm (7) automatically disassembles and temporarily moves away the light shield (17) of the hydroponic device unit directly below it;

[0023] Step S14, the outer telescopic arm (8) moves the annular guide rail (9) downward to be flush with the center of the hydroponic device unit;

[0024] Step S15, the camera (10) and the background plate (11) rotate at the same speed along the annular guide rail (9) around the central axis of the hydroponic device (4), and collect images of the entire crop plant;

[0025] Step S16, the outer telescopic arm (8) moves the annular guide rail (9) upward to reset;

[0026] Step S17, the inner telescopic arm (7) assembles the light shield (17) back into the hydroponic device;

[0027] Step S18, moving the suspension (2) and the imaging device (3) along the longitudinal guide rail and the transverse guide rail to the next hydroponic device unit, repeating steps S13-S17, and collecting whole-plant crop images of all hydroponic device units.

[0028] More specifically, the application method of the high-throughput hydroponic crop 3D time-series imaging system is characterized in that the steps of crop 3D reconstruction are specifically as follows:

[0029] Step S21, preprocessing the collected image to obtain an image containing only crops;

[0030] Step S22, performing distortion correction on the whole crop image obtained in step S21 to eliminate image distortion caused by camera imaging and medium refraction;

[0031] Step S23, arranging the corrected images obtained in step S22, and grouping the images of the same plant collected at the same time;

[0032] Step S24, performing pose estimation on each set of images to obtain pose information, that is, the position and orientation of the camera in the three-dimensional space when the images are captured;

[0033] Step S25, combining the corrected image obtained in step S22 and the posture information obtained in step S24, using the neural radiation field to perform three-dimensional reconstruction on each group of images to obtain a three-dimensional radiation field of the entire crop;

[0034] Step S26, converting the three-dimensional radiation field of the whole crop obtained in step S25 into a three-dimensional point cloud of the whole crop;

[0035] Step S27, analyzing the three-dimensional point cloud of the whole crop obtained in step S26 to extract the three-dimensional phenotypic information of the whole crop.

[0036] More specifically, the application method of the high-throughput hydroponic crop 3D time-series imaging system is characterized in that the step of phenotype extraction is specifically as follows:

[0037] Step S31, preprocessing the obtained three-dimensional point cloud of the whole crop;

[0038] Step S32, using a geometric calculation method to extract key points of the three-dimensional point cloud of the whole crop, and obtaining phenotypes including plant height, plant width, root length and root angle by calculating the relative positions of the key points in the three-dimensional space;

[0039] Step S33, using voxelization or triangulated meshing methods to calculate phenotypes including the surface area of ​​the crop above water, the volume of the crop above water, the root surface area, the root volume, and the minimum enclosing convex hull volume of the root system;

[0040] Step S34, calculating the ratio of the minimum root enclosing convex hull volume to the root volume, and obtaining phenotypic information of the root dispersion degree.

[0041] More specifically, the application method of the high-throughput hydroponic crop 3D time-series imaging system is characterized in that the step of crop growth visualization prediction is specifically as follows:

[0042] Step S41, establishing a growth prediction model based on the extracted three-dimensional phenotypic information of the whole crop, combined with time series information, variety information and environmental information;

[0043] Step S42, based on the growth prediction model obtained in step S41, predicting the future three-dimensional phenotypic information of the whole crop of the specific variety through the current three-dimensional phenotypic information of the whole crop, variety information and environmental information;

[0044] Step S43, using a three-dimensional geometric modeling technology, converting the three-dimensional phenotypic information of the future whole crop obtained in step S42 into spatial geometric information, so as to achieve a three-dimensional representation of the morphological structure of the future whole crop;

[0045] Step S44, using the texture model, the illumination model and the color rendering method, the three-dimensional characterization result obtained in step S43 is integrated to obtain the three-dimensional visualization growth prediction result of the whole crop in the future.

[0046] Compared with the prior art, the present invention has significant positive technical effects, which are specifically manifested in the following parts.

[0047] (1) Most of the existing root phenotyping technologies based on hydroponic environments directly and completely discharge the old nutrient solution and replace it with new nutrient solution when replacing the nutrient solution. This method of changing the solution will affect the original configuration and position of the root system, and bring errors to the subsequent root system temporal phenotyping analysis. The siphon-based solution changing device designed by the present invention can discharge and inject the nutrient solution at the same time with the same efficiency, so that the liquid level remains unchanged when changing the solution in the hydroponic scene, thereby ensuring that the root system configuration of the crop in the hydroponic device remains unchanged, so that the temporal imaging results can intuitively show the growth changes of the root system, and facilitate the analysis of the root system temporal phenotype.

[0048] (2) Most of the existing crop phenotyping devices in traditional soil cultivation environments are limited to two-dimensional levels, and rarely consider both the aboveground and underground parts at the same time, resulting in incomplete crop phenotyping. Most of the existing crop phenotyping systems in hydroponic environments only perform phenotyping on single crops. Existing high-throughput hydroponic systems cannot perform phenotyping well, and it is difficult to combine other technologies to achieve multi-view imaging and three-dimensional reconstruction. The crop hydroponic device and the matching movable, automatic acquisition imaging device designed by the present invention, wherein the transparent culture medium realizes the image acquisition and phenotypic analysis of the whole crop including the aboveground and underground parts, the hydroponic device array cooperates with the liquid replacement device to realize high-throughput hydroponic planting of crops, and the automated imaging device can efficiently realize multi-view imaging around the hydroponic environment, which is convenient for subsequent three-dimensional modeling and three-dimensional phenotypic analysis. Therefore, the device has the advantages of whole-plant phenotyping, three-dimensional analysis, high throughput, and automation, and realizes high-throughput and efficient three-dimensional phenotypic analysis of the whole crop including the aboveground and underground parts.

[0049] (3) Existing traditional crop phenotyping methods are often limited to phenotypic extraction and analysis on a two-dimensional plane, which makes it difficult to fully and accurately reflect the three-dimensional morphology and dynamic growth process of crops, resulting in one-sided information acquisition and lack of intuitive and in-depth insights. The present invention designs a technical process for crop 3D reconstruction, phenotypic extraction and visual prediction of future crop growth trends based on multi-view crop images. This technology achieves the improvement of crop phenotypes from two-dimensional to three-dimensional, making phenotypic analysis more accurate. At the same time, this technology also integrates crop temporal growth information and environmental information to establish a crop growth model, which can visualize the growth trend and possible morphology of crops in the future in three-dimensional form, enriching the means of crop phenotypic research, improving research efficiency and depth, providing precise guidance for crop genetic breeding, and providing technical support for growth monitoring and field management. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 FIG. 4 is an overall schematic diagram of the imaging system of the present invention.

[0051] Figure 2 Schematic diagram of the imaging device.

[0052] Figure 3 Schematic diagram of the hydroponic device.

[0053] Figure 4 Schematic diagram of the hydroponic device flat plate.

[0054] Figure 5 Schematic diagram of the fluid replacement device.

[0055] Figure 6 Flow chart of the application method of the high-throughput hydroponic crop 3D time-series imaging system.

[0056] Figure 7 Schematic diagram of the three-dimensional point cloud of the aboveground part of the crop.

[0057] Figure 8 This is a schematic diagram of the three-dimensional point cloud of the underground part of the crop. DETAILED DESCRIPTION

[0058] The present invention is further described below in conjunction with the accompanying drawings and implementation examples.

[0059] An exemplary embodiment of the present invention is now described in detail, and this detailed description should not be considered as a limitation of the present invention, but should be understood as a more detailed description of certain aspects of the present invention. It should be understood that the terms described in the present invention are only for describing a particular embodiment, rather than being limitations of the present invention. For the parameter value range in the present invention, it should be understood that the upper and lower limits of the range and their intermediate values ​​are specifically disclosed.

[0060] Unless otherwise specified, all techniques and terms used herein have the same meanings as commonly understood by those skilled in the art in the art to which the invention pertains. Although only one specific exemplary embodiment has been described herein, methods and materials similar or equivalent to those described herein may also be used in the practice or testing of the invention.

[0061] like Figure 1 As shown, the suspension rail is 10 meters long; the transverse rail of the suspension (2) is 10 meters long, the support rod is 1 meter high, and a motor is installed inside the support rod to control the support rod to drive the suspension to move along the suspension rail (1); 36 hydroponic devices (4) are located in the area between the two longitudinal rails of the suspension rail (1), arranged in an array of 6 rows and 6 columns, and the distance between every two hydroponic devices is 1.5 meters to ensure a sufficient imaging distance.

[0062] like Figure 2 As shown, the inner telescopic arm of the imaging device (3) is equipped with a motor to control the loading and unloading of the light shield (17); the outer telescopic arm is equipped with a click to control the movement of the circular guide rail (9); the circular guide rail (9) has a radius of 1.2 meters, and is equipped with a camera (10) and a background plate (11), and the motor controls the movement of the two on the inner side of the guide rail; the camera (10) uses a manual focus mode to ensure that the focal length remains unchanged, and uses a video recording method to collect the image of the whole corn plant; the background plate (11) is coated with black matte paint.

[0063] like Figure 3 , 4 As shown, the transparent cylindrical growth tank (13) of the hydroponic device (4) is made of glass, 40 cm high, 20 cm in outer diameter, and 5 cm thick; the flat plate (14) is made of a square black KT plate, 15 cm in side length, 0.5 cm thick, and a fixing hole (19) with a diameter of 3 cm is dug in the center; a black sponge is clamped on the stem above the root of the corn and placed in the fixing hole (19), thereby fixing the base of the corn without damage and providing moisture to the seeds; the nutrient solution (16) uses the corn Yoshida formula; the placing table (18) is made of hard plastic material, 30 cm long, 30 cm wide, and 5 cm high.

[0064] like Figure 5 As shown, when the liquid replacement device (5) is working, the two siphon tubes for liquid replacement are controlled by the stop valve on the siphon tube (20) to have the same liquid replacement efficiency. Each hydroponic device (4) replaces the liquid for 10 minutes to ensure that the nutrient solution is fully replaced.

[0065] An application method of a high-throughput hydroponic corn 3D time-series imaging device, using the above-mentioned high-throughput hydroponic corn 3D time-series imaging device for imaging, and then performing crop 3D reconstruction, phenotype extraction and corn growth visualization prediction. The flowchart of the method is as follows Figure 6 shown.

[0066] In the preparation stage, the corn seeds were germinated, and 36 germinated seeds were selected and transplanted into different hydroponic devices (4).

[0067] During the imaging phase, images of the maize grown in the device were collected every 3 days.

[0068] In the reconstruction phase, the steps of corn 3D reconstruction are as follows:

[0069] Step S21, performing threshold segmentation on the collected image, distinguishing the background part from the part where the plant is located, and setting them to black and white respectively, selecting the area where the white part is located to mask the collected image, and obtaining an image containing only corn;

[0070] Step S22, calibrating the camera using the Zhang Zhengyou calibration method, calculating the camera internal parameters, and performing distortion correction on the whole corn plant image obtained in step S21 in combination with the imaging distance and the refractive index of glass and water to eliminate the image distortion caused by camera imaging and medium refraction;

[0071] Step S23, arranging the corrected images obtained in step S22, and grouping the images of the same corn plant collected at the same time;

[0072] Step S24, using the SFM algorithm to estimate the pose of each group of images to obtain pose information, that is, the position and orientation of the camera in the three-dimensional space when the image is captured;

[0073] Step S25, dividing the corrected image obtained in step S22 into an above-ground part and an underground part, combining the posture information obtained in step S24, and using the neural radiation field to perform three-dimensional reconstruction on each group of images to obtain the three-dimensional radiation field of the above-ground part and the underground part of the corn;

[0074] Step S26, converting the three-dimensional radiation field obtained in step S25 into a three-dimensional point cloud. The schematic diagrams of the three-dimensional point clouds of the above-ground part and the underground part are as follows: Figure 7 and Figure 8 shown.

[0075] In the analysis phase, the steps for extracting the three-dimensional phenotype of corn are as follows:

[0076] Step S31, using a region growing algorithm to segment the three-dimensional point cloud of corn obtained in step S26;

[0077] Step S32, extracting edge points of the three-dimensional point cloud of corn obtained in step S31, and calculating geometric distances of the edge points to obtain phenotypes such as corn plant height and root depth;

[0078] Step S33, skeletonizing the three-dimensional point cloud of corn obtained in step S31, and using a B-spline curve fitting method to calculate phenotypes such as the number of main roots, root length, and leaf length;

[0079] Step S34, using a triangular mesh method to calculate the phenotypes of the corn above water surface area, crop above water volume, root surface area, root volume, and root minimum enclosing convex hull volume from the corn three-dimensional point cloud obtained in step S31;

[0080] Step S35, calculating the ratio of the minimum root enclosing convex hull volume to the root volume, and obtaining the phenotypic information of the root dispersion degree.

[0081] In the prediction stage, the steps of visual prediction of corn growth are as follows:

[0082] Step S41, establishing a growth prediction model based on the extracted three-dimensional phenotypic information of corn, combined with time series information and variety information;

[0083] Step S42, based on the growth prediction model obtained in step S41, input the three-dimensional phenotypic information of the current variety of corn, and predict the three-dimensional phenotypic information of the corn in the future;

[0084] Step S43, using a three-dimensional geometric modeling technology, converting the three-dimensional phenotypic information of the future corn obtained in step S42 into spatial geometric information, so as to achieve a three-dimensional representation of the morphological structure of the whole corn plant in the future;

[0085] Step S44, using the texture model, the illumination model and the color rendering method, the three-dimensional characterization result obtained in step S43 is integrated to obtain the three-dimensional visualization growth prediction result of the whole corn plant in the future.

[0086] Although the present invention has been described with reference to exemplary embodiments, it should be understood that the exemplary embodiments described in the present invention are merely illustrative of the spirit of the present invention. Those skilled in the art of the present invention may make various modifications, supplements or replace the described exemplary embodiments in a similar manner, but will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

Claims

1. A high-throughput hydroponic crop 3D time-series imaging system, characterized in that: The invention comprises a suspension rail (1), a suspension (2), an imaging device (3), a hydroponic device (4) and a liquid changing device (5): The suspension guide rail (1) comprises two longitudinal guide rails parallel to each other and placed on a plane; The suspension (2) comprises two support rods and a transverse guide rail, the two support rods are respectively vertically mounted on the suspension guide rail, the transverse guide rail is mounted on the upper ends of the two support rods and spans between two longitudinal guide rails of the suspension guide rail; the suspension can be flexibly moved along the longitudinal guide rail; The imaging device (3) comprises a suspension connecting member (6), an inner telescopic arm (7), an outer telescopic arm (8), an annular guide rail (9), a camera (10) and a background plate (11); the imaging device is connected to the suspension (2) via the suspension connecting member (6) and is located below the suspension, and can be flexibly moved along the transverse guide rail on the suspension; the upper ends of the inner telescopic arm (7) and the outer telescopic arm (8) are connected to the suspension connecting member (6); the lower end of the outer telescopic arm (8) is connected to the annular guide rail (9); the camera (10) and the background plate (11) are mounted on the annular guide rail (9), the camera and the background plate are mounted facing each other, the centers of the two are located on the same straight line with the center of the annular guide rail, and the two can rotate synchronously and uniformly at the same speed along the annular guide rail around the central axis of the hydroponic device (4) to obtain crop images; The hydroponic device (4) is located in the area between the two longitudinal rails of the suspension rail (1), and includes a plurality of hydroponic device units arranged in an array; each hydroponic device unit includes a plant sample (12), a transparent cylindrical growth tank (13), a plate (14), a sponge (15), a nutrient solution (16), a light shield (17) and a placement table (18); every two hydroponic device units are separated by a certain distance in the horizontal and vertical directions, so as to facilitate the imaging operation of the imaging device (3); The liquid replacement device (5) comprises a siphon (20), a container (21) containing new nutrient solution, and a container (22) for collecting old nutrient solution, and is used to regularly replace the nutrient solution (16) of the hydroponic device (4); The suspension connecting member (6) of the imaging device is used to drive the imaging device (3) to move laterally along the suspension (2); the lower end of the inner telescopic arm (7) is used to remove and install the light shield (17) of the hydroponic device; the outer telescopic arm (8) is used to drive the annular guide rail (9) to move up and down; the annular guide rail (9) is centered on the hydroponic device (4) so ​​that the distance between the camera and the crop remains unchanged; the camera (10) adopts manual focus and obtains the image of the whole crop in a video shooting manner; the inner side of the background plate (11) is painted black to facilitate the acquisition of the image of the whole crop.

2. The high-throughput hydroponic crop 3D time-series imaging system according to claim 1, characterized in that: In the hydroponic device (4), the flat plate (14) is of square design, and its outer side is fixedly connected to the transparent cylindrical growth tank (13) by interference fit with the inner wall of the transparent cylindrical growth tank (13); the flat plate (14) is provided with a fixing hole (19) penetrating from top to bottom; the sponge (15) is fixedly connected to the flat plate (14) by interference fit with the inner wall of the fixing hole (19); the sponge (15) is located at the center of the square of the flat plate (14); the liquid level of the nutrient solution (16) is flush with the flat plate (14); the transparent cylindrical The growth pot (13) is used to contain components and provide a space for plant growth; the flat plate (14) is used to fix the plant at the center of the hydroponic device (4); the sponge (15) is used to provide water to the plant seed rootlets during the germination period and to fix the base of the plant root system in the later stage; the light shield (17) is wrapped around the outer periphery of the transparent cylindrical growth pot (13) in a detachable manner to prevent the root system from being affected by external light, and is removed when the crop is imaged; the flat plate (14), the sponge (15) and the light shield (17) are made of black material.

3. The high-throughput hydroponic crop 3D time-series imaging system according to claim 1, characterized in that: In the liquid replacement device (5), the liquid level of the container (21) containing the new nutrient solution is higher than the liquid level of the nutrient solution (16) in the hydroponic device (4); and the liquid level of the nutrient solution (16) in the hydroponic device (4) is higher than the liquid level of the container (22) for collecting the old nutrient solution.

4. The high-throughput hydroponic crop 3D time-series imaging system according to claim 1, characterized in that: The siphon tubes (20) of the liquid replacement device (5) include two siphon tubes, one of which is connected to the top of the hydroponic device (4) and the container (21) containing the new nutrient solution, and is used to siphon the new nutrient solution into the hydroponic device (4), and the other siphon tube is connected to the bottom of the hydroponic device (4) and the container (22) for collecting the old nutrient solution, and is used to siphon the old nutrient solution in the hydroponic device (4) out into the container (22) for collecting the old nutrient solution; the two siphon tubes have the same siphon power, and can discharge the old nutrient solution while inputting the new nutrient solution, thereby ensuring that the liquid level of the nutrient solution (16) in the hydroponic device (4) remains unchanged during the water replacement period.

5. An application method of a high-throughput hydroponic crop 3D time-series imaging system, characterized in that: Using the high-throughput hydroponic crop 3D time-series imaging system described in any one of claims 1 to 4 to perform imaging, and then perform crop 3D reconstruction, phenotype extraction, and crop growth visualization prediction; The imaging steps are specifically as follows: Step S11, the suspension (2) moves along the suspension guide rail (1) to the first row of the hydroponic device array, and the center axis of the suspension and the center of the hydroponic device unit in the first row are on the same vertical plane; Step S12, the imaging device (3) moves along the suspension (2) to the top of the first hydroponic device unit, and the center of the imaging device and the center of the first hydroponic device unit are located on the same vertical line; Step S13, the inner telescopic arm (7) automatically disassembles and temporarily moves away the light shield (17) of the hydroponic device unit directly below it; Step S14, the outer telescopic arm (8) moves the annular guide rail (9) downward to be flush with the center of the hydroponic device unit; Step S15, the camera (10) and the background plate (11) rotate at the same speed along the annular guide rail (9) around the central axis of the hydroponic device (4), and collect images of the entire crop plant; Step S16, after the image acquisition is completed, the outer telescopic arm (8) moves the annular guide rail (9) upward to reset; Step S17, the inner telescopic arm (7) assembles the light shield (17) back into the hydroponic device; Step S18, moving the suspension (2) and the imaging device (3) along the longitudinal guide rail and the transverse guide rail to the next hydroponic device unit, repeating steps S13-S17, and collecting whole-plant crop images of all hydroponic device units.

6. The application method of the high-throughput hydroponic crop 3D time-series imaging system according to claim 5, characterized in that: The specific steps of crop 3D reconstruction are: Step S21, preprocessing the collected image to obtain an image containing only crops; Step S22, performing distortion correction on the whole crop image obtained in step S21 to eliminate image distortion caused by camera imaging and medium refraction; Step S23, arranging the corrected images obtained in step S22, and grouping the images of the same plant collected at the same time; Step S24, performing pose estimation on each set of images to obtain pose information, that is, the position and orientation of the camera in the three-dimensional space when the images are captured; Step S25, combining the corrected image obtained in step S22 and the posture information obtained in step S24, using the neural radiation field to perform three-dimensional reconstruction on each group of images to obtain a three-dimensional radiation field of the entire crop; Step S26, converting the three-dimensional radiation field of the whole crop obtained in step S25 into a three-dimensional point cloud of the whole crop; Step S27, analyzing the three-dimensional point cloud of the whole crop obtained in step S26 to extract the three-dimensional phenotypic information of the whole crop.

7. The application method of the high-throughput hydroponic crop 3D time-series imaging system according to claim 6, characterized in that: The specific steps of phenotype extraction are: Step S31, preprocessing the obtained three-dimensional point cloud of the whole crop; Step S32, using a geometric calculation method to extract key points of the three-dimensional point cloud of the whole crop plant, and obtaining phenotypes including plant height, plant width, root length and root angle by calculating the relative positions of the key points in the three-dimensional space; Step S33, using voxelization or triangulated meshing methods to calculate phenotypes including the surface area of ​​the crop above water, the volume of the crop above water, the root surface area, the root volume, and the minimum enclosing convex hull volume of the root system; Step S34, calculating the ratio of the minimum root enclosing convex hull volume to the root volume, and obtaining phenotypic information of the root dispersion degree.

8. The application method of the high-throughput hydroponic crop 3D time-series imaging system according to claim 7, characterized in that: The specific steps of crop growth visualization prediction are as follows: Step S41, establishing a growth prediction model based on the extracted three-dimensional phenotypic information of the whole crop, combined with time series information, variety information and environmental information; Step S42, based on the growth prediction model, predicting the future three-dimensional phenotypic information of the whole crop of a specific variety through the current three-dimensional phenotypic information of the whole crop, variety information and environmental information; Step S43, using a three-dimensional geometric modeling technology, converting the three-dimensional phenotypic information of the future whole crop obtained in step S42 into spatial geometric information, so as to achieve a three-dimensional representation of the morphological structure of the future whole crop; Step S44, using the texture model, the illumination model and the color rendering method, the three-dimensional characterization result obtained in step S43 is integrated to obtain the three-dimensional visualization growth prediction result of the whole crop in the future.

Images