A method for screening salt-tolerant varieties and a crop germination vigor monitoring system

Abstract

The application discloses a salt-tolerant variety screening method and a crop germination activity monitoring system, which utilizes a fluorescent tetrazolium salt solution to pretreat the seeds to be measured, and then places the seeds in a salt stress environment for germination tracking. The application utilizes the reduction reaction of dehydrogenase to tetrazolium and other substances generated in the growth and development process of the strong water absorption and respiration areas such as radicles and plumules in the seed germination process to mark the areas with strong germination activity in the seed with fluorescence. Thus, the application can track the fluorescence intensity in the process image of the seed germination and the concentration of the fluorescence marking in the target areas such as radicles and plumules, so that the varieties with strong germination activity in the salt stress environment can be screened before the seeds completely germinate. The monitoring system of the application can more accurately and efficiently screen the salt-tolerant varieties by tracking the characteristics of the seed plumules and radicles, and provides a detection basis for plant breeding.

Description

Technical Field

[0001] This invention belongs to the field of crop germination morphology monitoring technology, and particularly relates to a method for screening salt-tolerant varieties and a crop germination vigor monitoring system. Background Technology

[0002] Seed germination vigor is a crucial indicator of seed quality, directly impacting yield. Research on seed germination vigor in major crops such as wheat, corn, and cotton has been a global focus in agricultural research. Measuring the physical, chemical, and physiological characteristics of seeds can determine seed quality. Rapid determinations of seed vigor and germination ability include: rapid determination using tetrazolium staining, conductivity methods, and rapid germination methods that break dormancy and accelerate water absorption. Traditional seed germination vigor monitoring relies heavily on manual counting, which suffers from low efficiency, high subjectivity, potential damage to seed structure, cumbersome operation, and large errors, making it difficult to meet the demands of modern breeding's full-time, non-destructive testing requirements.

[0003] Existing detection technologies primarily focus on identifying the physical size of seed embryos and radicles. Under stress conditions, seed embryos and radicles germinate slowly and are smaller in size, resulting in low identification accuracy. Current technologies cannot predictively detect and screen the biological vitality of seeds before embryo and radicle germination, thus leading to low detection efficiency and susceptibility to identification errors and detection biases caused by various factors such as sampling and image processing. Summary of the Invention

[0004] The problem to be solved by this invention is to provide a method for screening salt-tolerant varieties and a crop germination vigor monitoring system. This application uses fluorescent tetrazolium salt to soak seeds and uses the fluorescent substances reduced by water absorption and respiration in key parts of the seed before germination of the plumule and radicle as the detection target. This allows for the detection and screening of germination vigor before the seed germinates the plumule and radicle.

[0005] To address the aforementioned technical problems, the present invention provides a method for screening salt-tolerant varieties, comprising the following steps: drying the disinfected and rinsed seeds; pretreating the dried seeds by immersing them in a fluorescent tetrazolium salt solution prepared according to preset salt stress conditions; uniformly placing the pretreated seeds in a culture container lined with filter paper, spraying the filter paper with a stress agent prepared according to preset salt stress conditions, and placing the container in an artificial climate chamber for cultivation; during the cultivation process, replenishing water at preset time intervals to maintain stable stress conditions, tracking, photographing, recording, and processing images of the seed germination process, and screening salt-tolerant varieties based on the fluorescence intensity and concentration in the target area of ​​the seed germination process images.

[0006] Optionally, in any of the above-described salt-tolerant variety screening methods, the step of tracking and capturing images of the seed germination process includes: turning on the excitation light at preset time intervals and capturing images of the seeds in the culture container under the excitation light environment; attaching a filter matching the wavelength of the excitation light to the lens and capturing fluorescence images of the seeds in the culture container; attaching a filter matching the wavelength of the emitted light to the lens and capturing ambient light images of the seeds in the culture container; the step of recording and processing the images of the seed germination process includes: labeling the category and shooting time of each captured seed image, fluorescence image, and ambient light image; preprocessing the seed image, fluorescence image, and ambient light image; inputting the preprocessed fluorescence information into a pre-trained model; and selecting seeds with the largest increase in fluorescence intensity during germination and whose fluorescence signal is concentrated in the target area, and marking them as salt-tolerant varieties.

[0007] Optionally, in any of the above-described salt-tolerant variety screening methods, the preprocessing steps for the seed images, fluorescence images, and ambient light images obtained from each capture include: identifying labels in the images; adjusting the deformation parameters of each image according to the pixel position of the label; aligning and determining the sampling area of ​​each image; sampling each image according to the pre-set sampling area; marking the type and sampling time of each sampled image; performing differential processing on the sampled seed image samples and ambient light image samples to obtain a differential image; performing gradient analysis on the differential image to extract the fluorescence-rich regions in the image; constructing a mask based on the fluorescence-rich regions; calculating the average brightness of the fluorescence regions in the sampled fluorescence image samples; enhancing and denoising the fluorescence signal in the image based on the average brightness; and extracting fluorescence information.

[0008] Optionally, in any of the above-described salt-tolerant variety screening methods, the model performs the following steps on each group of sampled seed image samples, fluorescence image samples, and ambient light image samples: first, identify and segment the target regions corresponding to each type of seed based on the ambient light image samples; encode the fluorescence information corresponding to each fluorescence image sequentially according to the sampling time sequence and input it into a convolutional neural network to extract dynamic features; perform nonlinear expansion based on the dynamic features to obtain fluorescence boundary regions; calculate the light intensity change of the target region in the fluorescence boundary region and extract the increase ratio of fluorescence intensity in the target region corresponding to the dynamic features; among the seeds with the highest increase ratio of fluorescence intensity, select the seeds with the highest proportion of the target region in the fluorescence boundary region, further track the germination time of their radicles and plumules, and select the seeds with the shortest average germination time of the radicles and plumules as salt-tolerant varieties.

[0009] Optionally, in any of the above-described salt-tolerant variety screening methods, the step of nonlinear extension based on the dynamic features in the model includes: fitting the edge positions of fluorescence information in each fluorescence image according to the dynamic features. Gradient variation during sampling time The coordinates of the edge of the fluorescent boundary region are expanded according to the following formula: ,in, Indicates the boundary position The intensity of fluorescence, This represents the adjustment coefficient. This represents the offset of the location of the same light intensity over time.

[0010] Optionally, in any of the above-described methods for screening salt-tolerant varieties, the fluorescent tetrazolium salt solution prepared according to the preset salt stress conditions is configured according to the following steps: a salt solution of the corresponding concentration is prepared according to the preset salt stress conditions; any one or any combination of the following reagent combinations is added to the salt solution: Reagent combination 1: dimethoxyazole yellow, phenazine dimethyl sulfate, resazurin; after mixing with the salt solution, the concentration of dimethoxyazole yellow is between 0.1–0.3 mg / mL, the concentration of phenazine dimethyl sulfate is between 0.1–0.3 mg / mL, and the concentration of resazurin is between 0.1–0.5 mM; Reagent combination 2: WST-8, 1-Methoxy PMS, 4-MUP; after mixing with the salt solution, the concentration of WST-8 is between 0.2–0.5 mM, the concentration of 1-Methoxy PMS is 1 / 40 of the concentration of WST-8, and the concentration of 4-MUP is between 0.1–0.3 mM. Between mg / mL; Drug combination 3: AIE type tetrazolium salt, Pluronic F-127; after being mixed with the salt solution, the concentration of AIE type tetrazolium salt is controlled between 5–20 μM, and the concentration of Pluronic F-127 is between 1 / 5 and 1 / 10 of the concentration of AIE type tetrazolium salt.

[0011] Meanwhile, to solve the above-mentioned technical problems, this application also provides a crop germination vigor monitoring system, characterized in that it includes: an artificial climate chamber for providing temperature, humidity and ambient light conditions for seed germination; an excitation light source, which is set in the artificial climate chamber for providing excitation light and exciting fluorescence signals when tracking and capturing images of the seed germination process at preset time intervals; an imaging device, which is set in the artificial climate chamber and equipped with filters respectively matched to the excitation light wavelength and emission light wavelength, for capturing images of seeds in a culture container, fluorescence images of seeds and ambient light images of seeds at preset time intervals; and a processing system for prompting corresponding processing steps according to the method described in claims 1-6, triggering the operation of the artificial climate chamber, excitation light source and imaging device, acquiring images obtained by the imaging device, and then: marking the category and shooting time of each captured seed image, fluorescence image and ambient light image; preprocessing the seed image, fluorescence image and ambient light image; inputting the preprocessed fluorescence information into a pre-trained model, screening out seeds with the largest increase in fluorescence intensity during germination and whose fluorescence signals are concentrated in the target area, and marking them as salt-tolerant varieties.

[0012] Optionally, in any of the above-described crop germination vigor monitoring systems, the target region is selected during model training based on the order of seed germination and / or the growth of the plumule and radicle during germination.

[0013] Optionally, in any of the above-described crop germination vigor monitoring systems, the identification information of each seed to be tested is marked at the same position in each of the culture containers.

[0014] Beneficial effects

[0015] This invention discloses a method for screening salt-tolerant varieties and a crop germination vigor monitoring system. The method involves pretreating test seeds with a fluorescent tetrazolium salt solution and then placing them under salt stress for germination monitoring. This application utilizes the reduction reaction of tetrazolium salt and other substances by dehydrogenases produced during the growth and development of regions with strong water absorption and respiration, such as the radicle and plumule, to mark areas with strong germination vigor with fluorescence. Therefore, by tracking the fluorescence intensity in images of seed germination and the concentration of fluorescent markers in target areas such as the radicle and plumule, this invention can screen for varieties that still exhibit strong germination vigor under salt stress before the seeds have fully germinated. The monitoring system, combined with the tracking of seed plumule and radicle traits, enables more accurate and efficient screening of salt-tolerant varieties, providing a basis for plant breeding. Attached Figure Description

[0016] The following detailed description, in conjunction with the accompanying drawings and embodiments of the present invention, is as follows:

[0017] Figure 1 This is a schematic diagram of the crop germination vitality monitoring system of this application;

[0018] Figure 2 This is a schematic diagram of the fluorescence image acquisition method in the climate chamber of this application;

[0019] Figure 3 This is a schematic diagram illustrating the concentration of fluorescent patterns in this application. Detailed Implementation

[0020] To make the objectives and technical solutions of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the described embodiments of this application without creative effort are within the scope of protection of this application.

[0021] Those skilled in the art will understand that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as herein.

[0022] The term "connection" as used in this application can mean a direct connection between components or an indirect connection between components through other components.

[0023] The meaning of "and / or" as used in this application includes situations where each exists alone or both exist simultaneously.

[0024] During germination, seed viability is typically manifested in the respiration rate of its embryo, enzyme activity, or its ability to absorb and convert soluble nutrients. However, with current technology, detecting these targets usually requires destroying the seed structure and then labeling the content of extracted enzymes and other substances. This detection method is cumbersome and cannot continuously track the subsequent developmental process of the seed. Therefore, the data obtained from such detection is difficult to provide a basis for predicting growth indicators in the later stages of seed development.

[0025] Non-destructive testing methods typically require culturing seeds until they germinate and develop radicle and plumule structures that extend to the outside of the seed. Qualitative conclusions about seed germination viability under stress are then obtained by monitoring the growth rate and other traits of these exposed radicle and plumule structures. However, such methods require long cultivation periods, cannot reveal the seed's internal response mechanisms to stress, and, due to the small size of newly germinated radicles and plumules, visual inspection requires complex detection models and inevitably introduces significant recognition and measurement errors.

[0026] To address the shortcomings of the aforementioned solutions, this application refers to... Figure 1 This invention provides a visual detection system for the reductive activity of dehydrogenases during seed germination based on fluorescent markers such as fluorescent tetrazolium salt solution. The system includes... Figure 1 Shown:

[0027] Artificial climate chamber 1, which is equipped with a temperature regulation module, a humidity regulation module 12 and an ambient light regulation module 11, is used to monitor the environmental conditions during the seed germination and cultivation process according to the experimental requirements and provide the temperature, humidity and ambient light conditions for seed germination according to the set indicators.

[0028] Excitation light source 2, which is set in an artificial climate chamber, is used to provide excitation light and excite fluorescence signals when tracking and capturing images of the seed germination process at preset time intervals;

[0029] The seeds to be tested are placed in culture containers. Each culture container is marked with identification information of the seeds to be tested at a fixed and uniform position, including their processing method, cultivation time, variety number, etc. Each culture container has a seed-laying groove 5 evenly arranged. The seeds are placed in the center of the depression of the seed-laying groove 5. Filter paper is laid around the seed-laying groove 5. The filter paper is pre-cut with holes according to the position and size of the seed-laying groove 5, so that the seeds in the center of the depression can maintain a humid environment with the filter paper around the holes without directly contacting the filter paper. Each culture container is made of light-transmitting material and has light-absorbing card paper at its bottom, so that the excitation light projected by the seed-laying groove 5 and the holes of the filter paper can be absorbed by the light-absorbing card paper at the seed position, reducing the reflection of the excitation light by the seeds and other detection targets and improving the signal-to-noise ratio of the fluorescence signal captured by the imaging device.

[0030] The imaging device 3 is set up in an artificial climate chamber and equipped with filters matched to the excitation and emission wavelengths, respectively, for capturing images at preset time intervals. Figure 2 Under the lighting conditions shown, images of the seeds, the fluorescence images of the seeds, and the ambient light images of the seeds in the culture container were captured at each time point t.

[0031] Among them, reference Figure 2 The imaging device 3 and the excitation light source 2 required for fluorescence signal excitation can be fixedly set on the imaging bracket facing each culture container, or can be adjusted to the appropriate imaging position in the climate chamber according to the imaging needs through a robotic arm or gantry structure.

[0032] Processing system 4 is used to prompt corresponding processing steps according to the method described below, and to trigger the operation of the temperature and humidity adjustment module, ambient light adjustment module, excitation light source, and imaging equipment in the artificial climate chamber according to a preset time cycle, to acquire images obtained by the imaging equipment and perform corresponding processing:

[0033] The operator is instructed to keep the disinfected and rinsed seeds in the breeding box and turn on the temperature and humidity control module to maintain ventilation and allow the seeds to dry.

[0034] The operator is instructed to pretreat the dried seeds by soaking them in a fluorescent detection solution, such as a fluorescent tetrazolium salt solution prepared according to the preset salt stress conditions.

[0035] The operator is instructed to place the pretreated seeds evenly in the culture container lined with filter paper. When laying the filter paper, the position of the seeding groove 5 in the culture container should be followed to avoid covering the groove. Then, spray the filter paper with the stress agent prepared according to the preset salt stress conditions. Place the culture container containing various seeds into the artificial climate chamber for cultivation.

[0036] During the cultivation process, the processing system triggers the humidity adjustment module to replenish water at preset time intervals to maintain stable stress conditions. It also triggers the imaging equipment, excitation light source, and corresponding gantry or robotic arm support to reach the top of each target cultivation container according to the sampling cycle requirements. The system tracks, captures, records, and processes seed images, fluorescence images, and ambient light images of the seed germination process, and marks the category and shooting time of each seed image, fluorescence image, and ambient light image captured.

[0037] Then, the seed image, fluorescence image, and ambient light image are preprocessed.

[0038] The pre-processed fluorescence information is input into a pre-trained model to screen out seeds with the largest increase in fluorescence intensity during germination and whose fluorescence signals are concentrated in the target area. These seeds are then marked as salt-tolerant varieties.

[0039] Therefore, this application can utilize the enrichment of external water and nutrients by the radicle and plumule within the seed during seed germination. It also utilizes the reduction effect of dehydrogenases and peroxidases produced during seed germination on fluorescent substances in the fluorescent detection solution, converting them into fluorescent substances in seed regions with strong germination activity. Then, by irradiating the seed with an external excitation light source, information on the distribution and accumulation of fluorescent substances within the seed can be obtained. Furthermore, this application can enhance different images to identify the accumulation of fluorescent substances, determining whether they are effectively concentrated at key germination locations. Thus, without damaging the seed structure, it can determine seed germination activity before the radicle and plumule structures, which are observable externally, thereby screening for seed varieties adapted to salt stress environments.

[0040] After obtaining data on fluorescent markers of germination activity within seeds under salt stress, the morphological characteristics of seed radicle and plumule structures can be further tracked based on this data to obtain full-process tracking data of seed germination, and target seeds that can adapt to salt stress and reduce yield loss or meet specific phenotypic requirements can be screened out.

[0041] During each shooting session, the processing system will adjust the excitation light source and shooting equipment to the top of the target culture container according to the preset time, turn on the excitation light, so that the excitation light shines on the seeds and is absorbed by the light-absorbing cardboard at the bottom of the container through the seeding groove where the seeds are located, reducing the amount of excitation light reflected to the camera when shooting.

[0042] Reference Figure 2 During each shooting session, the imaging device captures a set of three images in this excitation light environment:

[0043] Image of seeds in a culture container, taken without any filters applied to the lens of the imaging device;

[0044] Fluorescence image of seeds in a culture container. When taking this image, a filter matching the excitation wavelength needs to be installed in front of the lens.

[0045] An image of the ambient light in the seed culture container. When taking this image, the filter for the excitation wavelength mentioned above needs to be removed, and then a filter matching the emission wavelength needs to be installed in front of the lens.

[0046] For applications using dimethoxyazole yellow, dimethyl phenazine sulfate, and resazurin as fluorescent detectors, this application allows for the addition of dimethoxyazole yellow at a concentration of 0.1–0.3 mg / mL, dimethyl phenazine sulfate at a concentration of 0.1–0.3 mg / mL, and resazurin at a concentration of 0.1–0.5 mM to salt solutions prepared according to preset salt stress conditions. Before reduction by the reducing agent provided by the seed, dimethoxyazole yellow and dimethyl phenazine sulfate are primarily pale yellow to discolored, capable of absorbing ambient light of corresponding wavelengths, reducing the visual impact of the environment on the fluorescent patches. After the reaction, dimethoxyazole yellow absorbs orange ambient light, providing a marker for the colored area in the environmental image. Combined with the red fluorescence produced after the reduction of resazurin, this enhances the visual effect of image processing, facilitating visual observation and subsequent machine vision processing.

[0047] Therefore, by setting the excitation light source wavelength to the 540 nm or 571 nm range and setting the blocking wavelength range of the emission light wavelength filter to the 580–590 nm range, a differential image containing fluorescence signals and some excitation light residuals can be obtained by performing differential processing on the sampled seed image sample and the ambient light image sample. This allows for the maximum removal of noise from the ambient light and excitation light in the image, resulting in a differential image. In this image, the excitation light in the identification area where the seed is located can be projected onto the light-absorbing plate in the culture container and absorbed to the maximum extent, thereby extracting fluorescence information with a high signal-to-noise ratio with minimal computational overhead and hardware cost. Then, by performing gradient analysis on the differential image obtained in the aforementioned steps, fluorescence-rich regions in the image can be extracted. Simple gradient calculation can reduce the limitations of seed morphology, light intensity position, and computational overhead, enabling the labeling of fluorescence regions with minimal computational cost. As a result, the control unit can construct a mask based on the fluorescence-rich regions, calculate the average brightness of the sampled fluorescence image sample corresponding to the fluorescence region, and then enhance and denoise the fluorescence signal in the image based on the average brightness to extract the complete original fluorescence information. Therefore, this application can minimize the enhancement or weakening of fluorescence signals introduced by algorithm processing by controlling the consistency of ambient light during shooting, ensuring that the fluorescence signals extracted in each acquisition are consistent. This ensures that the extracted fluorescence signals not only retain the boundary and position information of the seed luminescent area to the maximum extent, but also avoid inconsistencies caused by the external environment or image processing mechanisms. Thus, in each extracted fluorescence signal, the intensity of seed germination vitality, respiration effect and enzyme activity can be expressed simply and intuitively using fluorescence intensity.

[0048] Therefore, this application can achieve non-destructive detection of seed germination viability by performing the following steps on each set of seed image samples, fluorescence image samples, and ambient light image samples obtained from each sampling:

[0049] First, the target regions corresponding to various seeds in the image are identified and segmented based on the ambient light image samples. The target regions corresponding to various seeds can be marked according to the concentrated areas of fluorescence signals in the seeds through pre-training of the model. This target region marking method can focus the model's attention on the concentrated distribution area of ​​germination vitality through fluorescence enrichment. By detecting and identifying the concentrated areas of seed germination vitality under different postures during training, the target regions in seeds of different postures can be labeled during detection.

[0050] The fluorescence information corresponding to each fluorescence image is sequentially encoded according to the sampling time sequence and input into a convolutional neural network to extract dynamic features that can reflect the changes in fluorescence signal intensity and position under that time sequence.

[0051] Based on the dynamic features, nonlinear extension is performed to track the rate of change of fluorescence intensity of fluorescent parts in the seed over time. Then, by using weighted smoothing of the curvature edge of the fluorescence boundary in the entire time series, interference noise in the image is removed, and the time series features of dynamic fluorescence changes are stably obtained. The improved Sigmoid gate is used to achieve soft threshold screening, thereby enhancing the acquisition of fluorescence boundary regions in the dynamically changing fluorescence time series signal.

[0052] Therefore, this application can directly calculate the proportion of the target region in the fluorescent boundary region based on the pixel coordinates of the region, and thus extract the increase ratio of fluorescence intensity in the target region corresponding to the dynamic features;

[0053] Then, among the seeds with the highest fluorescence intensity ratio, those with the highest proportion of the target area within the fluorescence boundary region are selected. This allows for the identification of seeds with strong vitality and germination vigor that require focused observation of radicle and plumule growth traits before the radicle and plumule germinate. Furthermore, this application can further track the germination time and phenotypic morphology of these seeds' radicles and plumules, thereby selecting the seeds with the shortest average germination time of their radicles and plumules as salt-tolerant varieties.

[0054] The nonlinear extension of the dynamic features in the above process can be specifically expressed by the following formula: Based on the dynamic features, the edge positions of the fluorescence information in each fluorescence image are fitted. Gradient variation during sampling time The coordinates of the edge of the fluorescent boundary region are expanded according to the following formula: ,in, Indicates the boundary position The fluorescence intensity at a given location, where t represents the time sequence. This represents the adjustment coefficient. This represents the shift in position of the same light intensity over time. The value of is between 1 and 1, representing the extended weight ratio. Therefore, this application can smoothly constrain the changes in fluorescence signal coordinates and intensity over time, dynamically screen the dynamic fluorescence change state after denoising using a soft threshold, and extract the changes in light intensity range within the fluorescence boundary region, thereby obtaining enhanced boundary change features. These features include both the coordinate position of the fluorescence boundary region and the change information in boundary fluorescence intensity. Thus, this application can extract the proportion of the target region within the fluorescence boundary region through model extraction, and then extract the increase ratio of fluorescence intensity in the target region corresponding to the dynamic features. Seeds that can still continuously and stably accumulate germination activity under salt stress are selected according to the fluorescence intensity increase ratio. Further screening is then conducted to select seeds whose fluorescence boundary regions conform to the distribution of the target region obtained through pre-training, ensuring that other interference errors are overcome. This allows for reliable prediction of the seed's resistance to salt stress before radicle germination without damaging the seed.

[0055] Therefore, this application can be used to further monitor the size and color of the germinated plumules and radicles of the seeds, and continue to track the relevant characteristics of plant growth.

[0056] In other implementations, the fluorescent detector used in this application for detecting seed germination activity can also be achieved by adding WST-8 to a solution under simulated salt stress at a concentration of 0.2–0.5 mM, adding 1-MethoxyPMS at 1 / 40 of the WST-8 concentration, and then supplementing with 4-MUP at a concentration of 0.1–0.3 mg / mL. This reagent combination can achieve fluorescent labeling by combining 4-MUP and WST-8, achieving 448 nm wavelength fluorescence labeling imaging in the 360 ​​nm to 505 nm excitation range, and utilizing the color development of the tetrazolium salt to reduce the masking and obstruction of the fluorescence signal by the seed surface color. This formulation can utilize WST-8 and 1-MethoxyPMS to provide colorimetric analysis for environmental and seed images, absorbing the cyan-green band of ambient light, reducing the influence of this band of light on the imaging in fluorescence imaging, and making the imaging effect of the blue fluorescence produced by 4-MUP after reduction during seed germination more significant.

[0057] The system can continue to process the time-series data obtained from sampling using the previous implementation method. The seed germination process images are recorded and processed as follows:

[0058] Based on the marking information set at the same position in each of the culture containers, the identification information of each seed to be tested carried in the container is obtained, and the category and shooting time of each seed image, fluorescence image and ambient light image captured are marked.

[0059] Preprocessing is performed on seed images, fluorescence images, and ambient light images. For example, the label coordinates in the images are first identified. Based on the pixel positions of the positioning markers at the four corners of the label, the offset of this sampling compared to the standard shooting posture is calculated, and the deformation parameters of each acquired image are adjusted accordingly. The deformed image is then scaled and rotated according to the offset of the current shooting angle, ensuring that the target position in each detected seed image is completely consistent, providing a unified coordinate reference for subsequent time-series tracking of the fluorescence region boundary. Then, the sampling area of ​​each image is determined according to the seed pixel range required for comparison, and each image is sampled according to the pre-set sampling area to obtain similar results. Figure 3 The seed images shown are labeled with the seed type, identification number, processing method and sampling time of each sampled image according to the tag information corresponding to its acquisition time;

[0060] The seed image sample obtained from each sampling is compared with the ambient light image sample to obtain a difference image;

[0061] Gradient analysis is performed on the difference image to extract... Figure 3 The region of fluorescence enrichment is indicated by the red area.

[0062] A mask is constructed based on the fluorescence-enriched region. The average brightness of the fluorescence region in the sampled fluorescence image is calculated. The fluorescence signal in the image is enhanced and denoised based on the average brightness to extract the fluorescence information from the current sample.

[0063] Finally, the preprocessed fluorescence information is encoded according to its sampling time sequence, and the fluorescence information containing the encoded identifier is input into a pre-trained model. The pre-trained model then filters out fluorescence signals that show the largest increase in fluorescence intensity during germination and are concentrated in areas such as... Figure 3 Seeds that show enhancement within the target area marked by the blue area, such as the radicle and plumule germination location, are marked as salt-tolerant varieties.

[0064] In another implementation, this application allows the addition of AIE-type tetrazolium salt and Pluronic F-127 to a salt solution prepared according to the required salt stress conditions during seed pretreatment. The concentration of the AIE-type tetrazolium salt is controlled between 5 and 20 μM, and the concentration of Pluronic F-127 is adjusted to 1 / 5 to 1 / 10 of the concentration of the AIE-type tetrazolium salt, thus obtaining a stress agent with fluorescent detection function. This formulation is more refined, and since the AIE-type tetrazolium salt serves as the sole chromogenic agent, and Pluronic F-127 does not add background color, the captured images allow for a more intuitive observation of the actual seed condition, completely eliminating the influence of external environmental factors on naked-eye vision.

[0065] The AIE tetrazolium salts in this stress agent are non-fluorescent before the action of reductases inside the seed, but produce strong AIE fluorescence under excitation light after being reduced to formazan. Therefore, in this embodiment, the excitation light source wavelength can be set in the range of 450 nm to 460 nm, and the emission wavelength filter can be set to block orange-red or red fluorescence in the wavelength range of 580–650 nm.

[0066] Therefore, after processing, seeds treated with stress agents can be sampled at preset time intervals throughout the entire seed germination and cultivation process. Each sampling involves acquiring three types of images: seed image, fluorescence image, and ambient light image. Then, using the coordinates of the labels on the images, distortions caused by device position deviations during image acquisition are corrected to ensure consistent field of view and seed coordinates in each acquired image. The images are then labeled with the seed identification code, processing method, and shooting time. Based on the labels, each set of images captured during the seed germination process is preprocessed. For each set of images: the seed image and the ambient light image are differentiated to obtain the difference image at that time point; gradient analysis is performed on the difference image to obtain the fluorescence-rich region; a mask corresponding to that time point is constructed based on the fluorescence-rich region; information about the fluorescence region in the real-world scene is extracted from the fluorescence image corresponding to that time point based on the mask position; the average brightness of the fluorescence region is used to enhance and denoise the fluorescence signal in the image, extracting the fluorescence information. Then, the fluorescence information is segmented into target regions corresponding to each seed type according to the seed coordinates. The fluorescence signals of each target region are sequentially encoded and labeled with their identity and temporal information according to the sampling time sequence. The time-labeled fluorescence information of the entire seed germination process is input into a convolutional neural network for dynamic feature extraction. The pixel coordinates of the seed fluorescence edge positions extracted based on the network lock at each time sequence t are then determined. Calculate the gradient of the sampling time sequence. Then, the calculated gradient is applied according to... This is further expanded to obtain a composite fluorescent boundary region that reflects the changes in light intensity range within the fluorescent boundary region. Therefore, this application can extract the increase ratio of fluorescence intensity in the target region corresponding to the dynamic characteristics by calculating the light intensity change in the target region within the fluorescent boundary region, and rank the various seeds accordingly. Then, among the seeds ranked high in fluorescence intensity increase ratio, the seeds with the highest proportion of the target region within the fluorescent boundary region are selected to remove interference biases caused by other factors in the fluorescence response. This method of first selecting fluorescence intensity and then removing the bias caused by fluorescence intensity at non-target locations reduces the uncertainty of the target location during model training due to sample bias or insufficient recognition accuracy, and simultaneously eliminates the interference of low-confidence non-target location fluorescence signals on the judgment results. After comprehensively selecting seeds according to the aforementioned two indicators, further information such as the germination time and growth rate of their radicles and plumules can be identified and tracked to select the seeds with the shortest average germination time and / or the highest growth rate and / or the morphological phenotype of the radicles and plumules that best meet the ideal expectations as salt-tolerant varieties.

[0067] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for screening salt-tolerant varieties, characterized in that the steps include: include: After disinfection and rinsing, the seeds to be tested were dried. The dried seeds were pretreated by soaking them in a fluorescent tetrazolium salt solution prepared according to the preset salt stress conditions. The pretreated seeds were evenly placed in a culture container lined with filter paper. Stress agents prepared according to the preset salt stress conditions were sprayed onto the filter paper and then placed in an artificial climate chamber for culture. During the cultivation process, water was replenished at preset time intervals to maintain stable stress conditions. Images of seed germination were tracked, recorded, and processed. Salt-tolerant varieties were screened based on the fluorescence intensity and concentration in the target area of ​​the seed germination process images.

2. The method for screening salt-tolerant varieties as described in claim 1, characterized in that, The steps involved in tracking and capturing images of the seed germination process include: The excitation light is turned on at preset time intervals, and images are taken in the excitation light environment: Images of seeds in a culture container; fluorescent images of seeds in a culture container are captured by attaching a filter to the lens that matches the excitation wavelength; ambient light images of seeds in a culture container are captured by attaching a filter to the lens that matches the emission wavelength. The steps for recording and processing images of the seed germination process include: Label the category and shooting time of each seed image, fluorescence image, and ambient light image captured; Seed images, fluorescence images, and ambient light images were preprocessed. The preprocessed fluorescence information was then input into a pre-trained model to screen out seeds with the largest increase in fluorescence intensity during germination and whose fluorescence signals were concentrated in the target area. These seeds were then marked as salt-tolerant varieties.

3. The method for screening salt-tolerant varieties as described in claim 2, characterized in that, The preprocessing steps for the seed image, fluorescence image, and ambient light image obtained from each capture include: The system identifies labels in the image, adjusts the deformation parameters of each image according to the pixel position of the label, aligns and determines the sampling area of ​​each image, then samples each image according to the pre-set sampling area, and marks the type and sampling time of each sampled image. The sampled seed image samples and ambient light image samples are subjected to differential processing to obtain a differential image; Gradient analysis is performed on the difference images to extract fluorescence-enriched regions. A mask is constructed based on the fluorescence-enriched region. The average brightness of the fluorescence region in the sampled fluorescence image is calculated. The fluorescence signal in the image is enhanced and denoised based on the average brightness to extract fluorescence information.

4. The method for screening salt-tolerant varieties as described in claim 3, characterized in that, The model performs the following steps on each group of seed image samples, fluorescence image samples, and ambient light image samples obtained from sampling: First, the target regions corresponding to various sub-sub ... The fluorescence information corresponding to each fluorescence image is sequentially encoded according to the sampling time sequence and then input into a convolutional neural network to extract dynamic features. Based on the aforementioned dynamic characteristics, a nonlinear extension is performed to obtain the fluorescence boundary region; Calculate the light intensity change in the target region within the fluorescence boundary region, and extract the increase ratio of fluorescence intensity in the target region corresponding to the dynamic features; Among the seeds with the highest fluorescence intensity ratio, those with the highest proportion of the target area in the fluorescence boundary region are selected. The germination time of their radicles and plumules is then tracked, and the seeds with the shortest average germination time of the radicles and plumules are selected as salt-tolerant varieties.

5. The method for screening salt-tolerant varieties as described in claim 4, characterized in that, The steps for nonlinear extension based on the dynamic features in the model include: Based on the dynamic features, the edge positions of fluorescence information in each fluorescence image are fitted. Gradient variation in the sampling time sequence The coordinates of the edge of the fluorescent boundary region are expanded according to the following formula: ,in, Indicates the boundary position The intensity of fluorescence, This represents the adjustment coefficient. This represents the offset of the location of the same light intensity over time.

6. The method for screening salt-tolerant varieties as described in claim 3, characterized in that, The fluorescent tetrazolium salt solution prepared according to the preset salt stress conditions was prepared according to the following steps: Prepare salt solutions of appropriate concentrations according to preset salt stress conditions; Add any one or any combination of the following reagents to the salt solution: Drug combination 1: dimethoxyazole yellow, phenazine dimethyl sulfate, and resazurin; after being mixed with the salt solution, the concentration of dimethoxyazole yellow is between 0.1 and 0.3 mg / mL, the concentration of phenazine dimethyl sulfate is between 0.1 and 0.3 mg / mL, and the concentration of resazurin is between 0.1 and 0.5 mM. Drug combination 2: WST-8, 1-Methoxy PMS, 4-MUP; after being mixed with the salt solution, the concentration of WST-8 is between 0.2 and 0.5 mM, the concentration of 1-Methoxy PMS is 1 / 40 of the concentration of WST-8, and the concentration of 4-MUP is between 0.1 and 0.3 mg / mL. Reagent combination 3: AIE type tetrazolium salt, Pluronic F-127; after being mixed with the salt solution, the concentration of AIE type tetrazolium salt is controlled between 5 and 20 μM, and the concentration of Pluronic F-127 is between 1 / 5 and 1 / 10 of the concentration of AIE type tetrazolium salt.

7. A crop germination vigor monitoring system, characterized in that, include: Artificial climate chambers are used to provide the temperature, humidity, and ambient light conditions necessary for seed germination. The excitation light source, which is set in an artificial climate chamber, is used to provide excitation light and excite fluorescence signals when tracking and capturing images of the seed germination process at preset time intervals. The imaging device is set up in an artificial climate chamber and is equipped with filters that are matched to the excitation wavelength and emission wavelength, respectively, to capture images of seeds in the culture container, fluorescence images of seeds, and ambient light images of seeds at preset time intervals. The processing system, according to the method described in claims 1-6, prompts corresponding processing steps, triggers the artificial climate chamber, excites the light source, and activates the imaging device to acquire the image obtained by the imaging device, and then: Label the category and shooting time of each seed image, fluorescence image, and ambient light image captured; Preprocess the seed image, fluorescence image, and ambient light image; The pre-processed fluorescence information is input into a pre-trained model to screen out seeds with the largest increase in fluorescence intensity during germination and whose fluorescence signals are concentrated in the target area. These seeds are then marked as salt-tolerant varieties.

8. The crop germination vigor monitoring system as described in claim 7, characterized in that, The target region is selected during model training based on the order of seed germination and / or the growth of the plumule and radicle during germination.

9. The crop germination vigor monitoring system as described in claim 7, characterized in that, Each of the culture containers has its own identification information for the seed to be tested marked at the same position.

Images