Method for evaluating fertility of lycium parent based on pollen vitality

Abstract

The present application relates to the technical field of plant breeding, and discloses a method for evaluating the fertility of a wolfberry parent based on pollen activity, which comprises establishing a time sequence of wolfberry pollen sampling, obtaining pollen microscopic images at each time point and performing adaptive white balance normalization processing, identifying effective active individuals based on the saturation characteristics of the HSV color space of the corrected images and forming a preliminary activity decay sequence, fitting the preliminary activity decay sequence using a constrained Weber model to extract shape parameters, correcting the contribution weight of sub-active particles using the weight coefficients determined by the shape parameters, generating a final activity decay data sequence, calculating the pollen activity half-life based on the sequence, determining the storage resistance grade by comparing with a preset storage resistance grading standard, and generating an evaluation result. The present application corrects the evaluation weight through kinetic parameter feedback, eliminates the false positive interference of environmental light and staining, realizes the quantitative grading of wolfberry pollen storage resistance, and accurately guides the selection and matching of breeding parents.

Description

Technical Field

[0001] This invention relates to the field of plant breeding technology, specifically a method for evaluating the fertility of wolfberry parent lines based on pollen viability. Background Technology

[0002] In the process of wolfberry hybridization breeding, accurate evaluation of pollen viability is a crucial step in parent selection and pollination decisions. Currently, pollen viability is mainly detected using chemical staining methods (such as TTC staining) combined with microscopic observation. However, this traditional method suffers from inconsistencies in optical environments during image acquisition and interpretation due to variations in microscope light source color temperature, light intensity fluctuations, and hardware differences between different devices. Relying on manual observation or simple grayscale analysis makes it difficult to accurately distinguish color depth differences under non-standardized lighting conditions, resulting in a highly subjective definition of highly viable and less viable pollen particles, which can easily lead to biases in the observation data.

[0003] Meanwhile, existing counting and statistical methods mostly employ a static cumulative model, treating all particles exhibiting a staining response (including deep red, strongly reactive particles and pink, weakly reactive particles) as equally viable individuals. This approach ignores the physiological decline dynamics of pollen populations during storage. Especially during pollen aging, sub-viable particles exhibiting a weak staining response often lose their actual fertilization capacity. Indiscriminately including them in the total count leads to a systematic overestimation of the actual pollination potential of the pollen population, failing to accurately reflect the physiological quality of the pollen.

[0004] Furthermore, current evaluations of storage tolerance are mostly based on single-point measurements or empirical qualitative descriptions, lacking quantitative analysis based on continuous time series. Due to the lack of a unified quantitative indicator (such as half-life) to characterize the rate of viability decay, breeders find it difficult to accurately determine how long the pollen of a particular wolfberry germplasm can maintain its effective pollination period. This makes it impossible to scientifically define in actual breeding work whether the germplasm is suitable for long-distance transportation across regions or limited to local immediate pollination, thus restricting the efficient utilization and scientific allocation of parental materials. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for evaluating the fertility of wolfberry parent lines based on pollen viability. This method solves the problem that existing methods for detecting pollen viability in wolfberry breeding rely on single-point measurements or manual observation, leading to highly subjective evaluation results, an inability to quantify the decline pattern of pollen storage tolerance, and consequently, difficulty in accurately guiding parent selection.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: a method for evaluating the fertility of wolfberry parent plants based on pollen viability, comprising establishing a sampling time series containing multiple discrete sampling time points, and realizing the quantitative classification of wolfberry pollen storage tolerance through a combination of image processing and dynamic model.

[0007] Specifically, this invention establishes a time series of wolfberry pollen sampling and acquires pollen microscopic images at each sampling time point. To address the light source fluctuations present during microscopic imaging, adaptive white balance normalization is performed on the original image using the statistical characteristics of the image background reference area, outputting a corrected image with consistent background brightness and a unified color standard.

[0008] Based on the corrected image, this invention extracts saturation features in the HSV color space. According to the color development characteristics of TTC staining products, thresholds for high and low viability are set, classifying pollen particles into high-viability, low-viability, and non-viable particles. In the initial quantization process, low-viability and high-viability particles are combined into effective viable individuals, and preliminary pollen viability observations at each sampling time point are calculated, forming a preliminary viability decay sequence.

[0009] Preferably, this invention utilizes a constrained Weber model to perform nonlinear regression fitting on the preliminary viability decay sequence, extracting shape parameters reflecting the decay rate characteristics. Weighting coefficients are determined based on these shape parameters, and these coefficients are used to correct the contribution weight of sub-viable particles in the overall viability calculation. Specifically, when the shape parameters indicate that the pollen population is in an accelerated aging stage, the weight of sub-viable particles is reduced, thereby suppressing the overestimation of viability caused by weakly reactive particles. Based on the corrected weighted data, the viability values ​​at each time point are recalculated to generate the final viability decay data sequence.

[0010] Furthermore, this invention calculates the pollen viability half-life based on the final viability decay data sequence. The calculation process automatically selects the calculation path based on the coefficient of determination of the data fit: when the goodness of fit meets a preset threshold, the half-life is calculated analytically using the model parameters obtained from quadratic fitting; when the goodness of fit is lower than the preset threshold, the half-life is calculated using linear interpolation. Finally, the calculated pollen viability half-life is compared with a preset storage tolerance grading standard threshold to determine the storage tolerance level of the tested wolfberry germplasm, and a fertility evaluation result including parental selection suggestions is generated.

[0011] Preferably, the adaptive white balance normalization process specifically uses the ratio of the statistical mean of the background reference area in the red, green, and blue channels to the target white point reference value to calculate the gain coefficient, and uses the gain coefficient to perform a linear transformation on the pixel values ​​of the pollen micrograph.

[0012] Preferably, the process of establishing the sampling time series includes: selecting mature flower buds in the whitening stage, removing the stamens and placing them in a constant temperature and dry environment until the anthers shed pollen, and using the day of pollen collection as the starting point to construct a time series covering the pollen viability decline process, ensuring that all batches of samples have a consistent physiological development state at the initial moment.

[0013] This invention provides a method for evaluating the fertility of wolfberry parent lines based on pollen viability. It has the following beneficial effects:

[0014] 1. This invention eliminates the interference of light source color temperature drift and brightness unevenness on TTC staining interpretation during microscopic imaging by combining adaptive white balance normalization processing and HSV color space quantization analysis. This method utilizes saturation channel features to replace subjective human judgment, achieving objective separation and counting of highly viable and less viable pollen particles, ensuring the accuracy and consistency of initial pollen viability observation data.

[0015] 2. This invention fits the viability decay sequence using a constrained Weber model and dynamically corrects the contribution weight of sub-viable particles using extracted shape parameters. This mechanism can automatically reduce the scoring weight of weakly reactive individuals when the pollen population is detected to be in an accelerated aging stage, thereby effectively calibrating the overestimation of viability caused by false positives in staining and improving the accuracy of the evaluation results in predicting the actual pollination potential of pollen.

[0016] 3. This invention transforms discrete time-series data into quantifiable storage duration indicators by calculating pollen viability half-life and establishing storage tolerance grading standards. This evaluation system can directly determine whether germplasm is suitable for long-distance inter-provincial introduction or limited to intra-field hybridization based on half-life values, providing an objective decision-making basis for parent selection in wolfberry hybridization breeding. Attached Figure Description

[0017] Figure 1 This is a flowchart of a method for evaluating the fertility of wolfberry parent plants based on pollen viability according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the morphology and sampling timing of the white bud stage of wolfberry flowers according to an embodiment of the present invention; Figure 3 This is a schematic diagram of a microscopic image acquisition and adaptive illumination correction process according to an embodiment of the present invention; Figure 4 This is a schematic diagram of pollen viability grading and counting logic in the HSV color space according to an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the principle of pollen viability decay kinetic model fitting and weight feedback correction according to an embodiment of the present invention; Figure 6This is a comparative diagram of the kinetic trends of pollen viability decline throughout the entire life cycle of two wolfberry germplasms in one embodiment of the present invention; Figure 7 This is a schematic diagram of the storage tolerance grading results based on pollen viability half-life in one embodiment of the present invention. Detailed Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] See attached document Figure 1 This invention provides a method for evaluating the fertility of wolfberry parent lines based on pollen viability. This method establishes a closed-loop evaluation system encompassing biological preparation, optical imaging correction, color space quantification, and kinetic parameter feedback to achieve quantitative grading of wolfberry pollen storage tolerance. The method mainly includes the following steps: Step S1: Construct a standardized goji berry pollen time-series sample library. During the peak flowering period of goji berries, select robust plants free from pests and diseases as the collection targets. To eliminate the interference of daytime high temperatures and sunlight on the initial state of pollen, the collection time is strictly limited to early morning, and the collected targets must be mature flower buds in the whitening stage, that is, the petals are slightly colored at the tip but not yet fully open, thus ensuring the consistency of initial physiological maturity of all batches of samples.

[0020] The collected flower buds were placed in a cool, shady indoor location to allow them to open naturally. Once the anthers were fully exposed, the stamens were removed. The removed stamens were then placed in a constant-temperature, dry environment with the temperature set to a standard level consistent with pollen physiological activity (e.g., 25±2℃) until the anthers naturally dehisced and released pollen. The collected pollen was then aliquoted and stored under room-temperature, moisture-proof conditions, and a sampling time series was established. This sampling time series, starting from the day of pollen collection, includes multiple discrete sampling time points covering the entire process of pollen viability from peak to complete decline, to address the technical limitation of existing single-point measurement methods that cannot predict storage tolerance.

[0021] Step S2: Acquire pollen microscopic images and perform adaptive illumination correction. At each sampling time point in the sampling time series, the corresponding sample is taken out and subjected to triphenyltetrazolium chloride (TTC) staining. The staining process follows standard incubation conditions to allow the dehydrogenases in viable pollen grains to fully react with the staining solution to generate formazan precipitate.

[0022] The stained slides were imaged using an optical microscope connected to a digital image acquisition device to obtain raw red, green, and blue images. To address fluctuations in light source intensity, differences in aperture size, or color temperature drift that occurred during the imaging process of different batches, adaptive white balance normalization based on background was performed on each raw red, green, and blue image. This process first identifies regions with high brightness and extremely low variance in the image as background reference areas, and calculates the average grayscale value of these regions in the red, green, and blue channels as a benchmark measure of ambient light intensity.

[0023] Assuming the background data is valid, the gain coefficient for each channel is calculated based on the ratio of the preset target white point reference value to the average gray value of each channel. The calculated gain coefficients are then used to perform a linear transformation on the corresponding channel values ​​of all pixels in the image, outputting a corrected image with consistent background brightness and a unified color reference. This step ensures that subsequent color-based vibrancy determination is not affected by changes in the hardware environment.

[0024] Step S3: Initial quantitative classification of pollen status based on the Hue, Saturation, and Value (HSV) color space. Given that the red-green-blue color space is sensitive to light and makes it difficult to intuitively separate hue and value, the corrected image is converted from the red-green-blue color space to the HSV color space, and the saturation channel, which is most sensitive to pigment concentration, is selected as the feature analysis channel.

[0025] Based on the biological characteristic that the color depth of the TTC reaction product formazan is positively correlated with the pollen viability intensity, a dual threshold is set on the saturation channel to finely classify pollen particles in the field of view into high-viability particles, low-viability particles, and non-viability particles according to the saturation level.

[0026] In the initial quantification stage, to address the technical limitation of traditional visual observation that easily misses pink, weakly reactive particles, this step utilizes image algorithms to identify both sub-viable particles (weakly reactive) and highly reactive particles (strongly reactive). Although there are chromatic differences between the two in their physiological state, when calculating the overall viability of the pollen population, according to standard measurement principles, sub-viable and highly reactive particles are counted together as effective viable individuals, thus avoiding underestimation of viability due to human visual errors. Combining the total number of identified viable particles with the total number of particles, preliminary pollen viability observations at each sampling time point are calculated, forming a preliminary viability decay sequence.

[0027] Step S4: Perform kinetic parameter feedback and weight correction using a constrained Weber model. The initial vitality decay sequence generated in Step S3 is fitted using a two-parameter Weber distribution function via nonlinear regression. The constrained Weber model can effectively describe the lifespan decay pattern of organisms. During the fitting process, a nonlinear optimization algorithm with boundary constraints is used to iteratively optimize the model parameters. Biologically meaningful physical boundary constraints are applied to the initial vitality, scale, and shape parameters to prevent the mathematical solutions from diverging or producing negative values.

[0028] The shape parameter estimate reflecting the decay rate trend is extracted from the fitting results. Based on this shape parameter estimate, a corrected weighting coefficient is calculated using a preset feedback function. The core logic of this feedback mechanism is to dynamically adjust the contribution weight of sub-viable particles to overall viability according to the decay acceleration or deceleration trend indicated by the shape parameter, thereby making the evaluation results more consistent with the true physiological decay characteristics of the pollen population. The corrected weighting coefficient is used to replace the initial weighting coefficient in step S3, and the pollen viability values ​​at each sampling time point are recalculated to generate the final viability decay data sequence.

[0029] Step S5: Calculate storage tolerance evaluation indicators and output grading results. Substitute the final vigor decay data sequence generated in Step S4 back into the boundary-constrained Weber model for a second fitting to obtain the final model parameters and coefficients of determination. Based on the distribution characteristics of the data determined by the coefficients of determination, choose to calculate the pollen vigor half-life using the final model parameters or by using linear interpolation. This half-life represents the time required for the vigor value to drop to half of its initial value (target vigor threshold). Based on the calculated half-life values, and comparing them with the preset grading standard thresholds corresponding to the breeding practice time nodes, classify the storage tolerance of the tested wolfberry germplasm into different levels (e.g., weak, medium, relatively strong, and strong). Finally, output the grading evaluation results of the germplasm, and mark the germplasm that meets the preferred grading standard as the target paternal material to guide the selection of parents in hybridization breeding.

[0030] See attached document Figure 1 and attached Figure 2 When implementing this invention to construct a standardized Lycium barbarum pollen time-series sample library, a strict biological material pretreatment and time-series sampling mechanism needs to be established. This mechanism aims to eliminate the interference of non-genetic factors (such as environmental heat shock and initial developmental differences) on the storage tolerance evaluation results, thereby ensuring that the final evaluation results only reflect the genetic characteristics of the germplasm itself.

[0031] Regarding the detailed implementation of step S1, this embodiment does not employ simple random sampling. Instead, it sets specific developmental stage locking and environmental control parameters based on plant physiological principles. Specifically, during the peak flowering period of wolfberry (usually May to June each year), for each wolfberry germplasm to be evaluated, plants with strong field growth, free from pests and diseases, and of uniform age are selected as test mother trees. A biological replicate of at least 10 plants is set to reduce statistical errors caused by individual differences. The time window for collection is strictly limited to early morning each day, specifically from sunrise until a significant rise in ambient temperature. This is to avoid unexpected loss of in vivo pollen viability due to high daytime temperatures and strong ultraviolet radiation.

[0032] The collection was strictly limited to mature flower buds in the "whitening" stage. The whitening stage refers to a specific morphological stage where the petals at the tip of the bud are just beginning to show color, the corolla has not yet opened, and the anthers have not yet dehisced. The physical basis for choosing this stage as the sole source of samples is that at this time, the pollen grains have completed physiological development but have not yet come into direct contact with the external environment, ensuring that all batches of samples are at the initial stage (i.e.,...). They have highly consistent initial pollen viability and water content, thus providing a unified baseline zero point for subsequent comparisons of decay kinetics.

[0033] After collection, the flower buds should be immediately placed in a cool, well-ventilated indoor area until the anthers are fully exposed. Then, the stamens should be manually removed. The removed stamens should be evenly spread on a clean carrier and placed in a constant-temperature drying device for drying. The drying environment temperature is set at 25±2℃. The purpose of this temperature parameter is to simulate the dehiscence process of the anthers under suitable natural conditions. This temperature range effectively promotes the dehydration and contraction of the anther wall cells to release pollen, while preventing the inactivation of dehydrogenases in the pollen due to heat shock above 30℃, ensuring that the enzymatic basis of the TTC staining reaction is not disrupted. The drying time is usually controlled to about 3 hours, or until the anthers are observed to naturally disperse as a yellow powder.

[0034] Collect scattered fresh flower pollen, remove filament remnants and impurities, and place it into well-ventilated microcentrifuge tubes or dedicated pollen storage bottles. Place the aliquoted pollen samples in sealed containers containing desiccants such as silica gel, and store the containers in a light-protected environment at a constant temperature of 25±2℃. This storage condition aims to create a controlled, room-temperature, moisture-proof environment, shielding the pollen from the effects of humidity fluctuations in the natural environment on its hygroscopic inactivation; retention time is the only dominant variable leading to pollen viability decline.

[0035] During sample storage, periodic sampling is performed according to a preset sampling time sequence. To address the limitation of existing technologies where single-point measurements cannot predict storage tolerance, this embodiment establishes a sampling time sequence covering the entire decay period (mathematically represented as a set of discrete sampling times). The construction of this sampling time series follows the principle of dynamic monitoring, that is, taking the day of pollen collection as the baseline starting point, and then sampling at preset time intervals within the storage period. Continuous monitoring will be conducted.

[0036] Define sampling time interval The following conditions must be met to ensure that the data point density is sufficient to fit the shape of the Weiber distribution: .

[0037] Based on the above interval rules, the discrete sampling time set corresponding to the sampling time series Defined as a combination of a series of discrete sampling time points: ; In a preferred embodiment of this invention, to balance observation cost and curve fitting accuracy, the following non-uniformly distributed sampling time series is specifically adopted: (Unit: days); in: This indicates the initial time (the day when pollen collection and drying are completed) used to determine the initial pollen viability observation. ; to This indicates that the intervals between subsequent sampling and monitoring points strictly follow the above. Scope constraints; The observation period is set to end on a 15-day date, which typically covers the entire life cycle from high viability to complete inactivation, based on the physiological characteristics of wolfberry pollen.

[0038] The discrete sampling time set corresponding to the sampling time series Each sampling time point Pollen samples from the corresponding batch were taken out of the constant temperature and moisture-proof container and immediately subjected to subsequent color development and slide preparation operations to ensure that the observation data truly reflected the storage-tolerant physiological state of the pollen at the sampling time point.

[0039] See attached document Figure 1 and attached Figure 3 In step S2, which involves acquiring pollen microscopic images and performing adaptive illumination correction, a rigorous staining and image preprocessing procedure is required to accurately convert the results of the biochemical reaction into a digital signal that can be processed by a computer and to eliminate environmental noise in the optical system.

[0040] Step S2 is specifically implemented based on the chemical colorimetric reaction principle of pollen viability. This is done within a discrete sampling time set. Each specific sampling time point Pollen samples from the corresponding batch were taken from a temperature- and moisture-proof storage environment and processed using the TTC staining method. According to the preferred embodiment, a 1.0% TTC solution (pH 7.0-7.4) was used as the staining solution, ensuring sufficient substrate for the dehydrogenase reaction. The specific staining process strictly followed these steps: a small amount of pollen sample was taken with a cotton swab and evenly dotted onto a clean glass slide; then 1-2 drops of TTC staining solution were added, and a coverslip was gently placed on top to avoid air bubbles. After slide preparation, the slide was incubated at 35±0.5℃ in the dark for 15 minutes. Under these conditions, succinate dehydrogenase in viable pollen grains reduced the colorless TTC to red, insoluble triphenylmethanesulfonate (TPF), while inactive pollen remained colorless.

[0041] After the colorimetric reaction is complete, digital image acquisition is performed. Using an optical microscope system connected to a high-resolution CCD or CMOS image sensor, following the sampling principles of randomness and uniform coverage, the slide is mechanically moved at a magnification of 100-200x (e.g., 10× eyepiece with 10× or 20× objective lens) to sequentially select at least five non-overlapping fields of view for imaging: upper left, upper right, center, lower left, and lower right. The specific hardware connection method between the optical microscope and the image sensor, the data transmission interface, and the low-level code writing of subsequent image processing algorithms (such as using OpenCV library functions) can be implemented by those skilled in the art using existing general-purpose equipment and programming techniques. The specific hardware structure and code implementation details are well-known in the field and will not be elaborated upon here.

[0042] During the acquisition process, the brightness, color temperature, and aperture of the microscope's light source may fluctuate slightly depending on the equipment status or operation time, resulting in slight variations in the original red, green, and blue images. Color cast occurs. To eliminate system errors caused by the hardware environment, this embodiment performs adaptive white balance normalization processing based on background statistical features on each original image. This processing logic first identifies areas in the image with high brightness and extremely low variance as background reference areas. This is based on the optical properties of blank areas in microscope transmission imaging.

[0043] After determining the background reference area, calculate the statistical mean of this area across the red (R), green (G), and blue (B) channels. For any color channel... Traverse the background reference area For all pixels within the range, sum their grayscale values ​​and divide by the total number of pixels. This allows us to obtain the average grayscale value of the background of that channel. .

[0044] The average gray value of the background This reflects the actual response intensity of the light source in each band under current imaging conditions. Based on the calculated background mean, a preset target white point reference value is introduced. (In 8-bit depth images, this is typically set between 245 and 250), calculate the gain coefficient for each channel. : ; in: Indicates the first The gain factor of each color channel, that is, the magnification factor required for the pixel value of that channel; This indicates the preset target white point reference value, representing the brightness level that an ideal white background should achieve (e.g., 250). The background reference area is indicated in the first position. The average gray value of the background on each color channel is used as the reference denominator for subsequent corrections. Represents a very small positive number (e.g.) ), as a denominator protection term, is used to prevent when When the value is zero, a mathematical singularity error of division by zero occurs.

[0045] The physical meaning of this gain coefficient is: the linear magnification factor required to stretch the actual background brightness of the current channel to the target white point reference value. After obtaining the gain coefficient, linear correction is performed on all pixels of the entire image to generate a corrected image. For any coordinate in the image The The correction calculation for the pixel values ​​of each color channel is as follows: ; in: This indicates the corrected image in coordinates. Place, No. Output pixel values ​​for each color channel; This represents a function that takes the minimum value and is used for numerical truncation. This represents the maximum brightness value allowed in an 8-bit image format, serving as a truncation threshold to prevent numerical overflow. Indicates the first The gain factor of each color channel, that is, the magnification factor required for the pixel value of that channel; This refers to the raw red, green, and blue images directly acquired by the image sensor (CCD / CMOS) of the microscopic imaging system before adaptive illumination correction is performed. In the middle, located at coordinates And belong to the first The original pixel grayscale values ​​of each color channel (typically ranging from 0 to 255).

[0046] Through the above processing, the background area is forcibly normalized to a standard bright white (red, green, and blue values ​​approaching a certain value). This method ensures that the color depth of pollen grains depends solely on their concentration, eliminating the interference of light source color temperature drift and providing standardized data input for subsequent quantitative analysis based on color space.

[0047] See attached document Figure 1 and attached Figure 4 In step S3, which is the initial quantification and grading of pollen status based on the HSV color space, color space transformation technology is used to convert the corrected image data into a quantitative index with biological semantics, so as to replace the subjective judgment of color depth by the naked eye.

[0048] The specific implementation of step S3 first involves the extraction of color features. Given that in the red-green-blue color space, the color information (chromaticity) of an image is highly coupled with the light intensity (brightness), directly using red, green, and blue values ​​for color segmentation is easily affected by residual brightness differences. To more accurately correspond to the biochemical characteristics of the TTC staining reaction, namely that the cumulative concentration of triphenylformazan (TPF) within pollen grains directly determines the saturation (purity) of the presented color, this embodiment uses the corrected image output in step S2... The color space was converted from red, green, and blue to HSV. In this space, the saturation channel (S channel) was selected as the core basis for feature analysis because its value is linearly positively correlated with the concentration of dehydrogenase-catalyzed reaction products in pollen.

[0049] In the S-channel image of HSV space, based on the color development mechanism of TTC staining, two key saturation judgment thresholds are set: a high viability judgment threshold. and low vitality threshold These two thresholds are set based on the statistical characteristics of a large number of samples, among which... The lower limit of saturation corresponding to the deep red precipitate (e.g., a normalized value of 0.45) indicates extremely high enzyme activity; The lower limit of saturation corresponding to a faint pink reaction (e.g., a normalized value of 0.15) is the lowest detectable limit characterizing the presence of enzyme activity.

[0050] Based on the above judgment threshold, the pixel-level classification logic for each pollen grain target within the field of view is as follows: If the average saturation within the pollen grain region It was identified as a highly active particle, corresponding to a deep red state, indicating that it has complete fertilization potential; If the average saturation within the pollen grain region These particles are classified as sub-active particles, corresponding to a pink or light red state, indicating that they are in a transitional stage of physiological function decline. If the average saturation within the pollen grain region These are classified as inactive particles, corresponding to a colorless or grayish-white state, indicating that they have lost their metabolic capacity.

[0051] After completing the classification and counting, in order to strictly adhere to the standard evaluation system of the TTC staining method and ensure that the subsequent storage tolerance grading results match the empirical thresholds (e.g., 50%) in breeding practice, the different color states identified in the images need to be mapped to standard pollen viability counting indicators. At the sampling time point... The number of each type of particle obtained through image analysis algorithms is denoted as follows: (Number of highly active particles) (Number of sub-active particles) and (Number of inactive particles).

[0052] According to the counting principle of this embodiment, all pollen exhibiting a red hue (including deep red and pink) indicates the presence of dehydrogenase activity. Although this method utilizes the HSV space to subdivide them into high-activity and low-activity pollen to achieve more accurate capture than human visual observation (especially to prevent the omission of light-colored low-activity particles), when calculating the viability ratio used for final classification, high-activity and low-activity particles are uniformly classified into the category of effective viable individuals. Specifically, the number of high-activity particles... With sub-active particle number Add them together to get the total number of individuals with effective vitality. At the same time, the total number of effective active individuals and the number of inactive particles are also included. Add them together to get the total number of particles in the current field of view. .

[0053] The formula for determining pollen viability based on digital image analysis is defined as follows: ; in: Indicates the sampling time point The calculated preliminary pollen viability observation value (i.e., pollen viability percentage) directly corresponds to the evaluation index specified in this invention and serves as the basic data for the feedback of kinetic parameters in the subsequent step S4. This represents the total number of individuals within the current field of view that are deemed to have effective vitality. This represents the total number of pollen particles within the current field of view. During algorithm execution, denominator protection logic needs to be implemented: if a pollen particle is detected... (i.e., empty field of view), then this calculation is invalid. ), are not included in the statistical sequence.

[0054] By analyzing the discrete sampling time set Zhong Cong to Repeat the above image acquisition and calculation process at all time points to generate a discrete dataset containing timestamps and corresponding vitality values. This discrete dataset quantifies the physiological decline of pollen populations over storage time, allowing it to be directly used in subsequent steps to evaluate storage endurance based on whether the viability value at key time points such as day 5 and day 9 is above 50%. The specific code implementation for the matrix operations of color space conversion and the connected component counting algorithm can be implemented by those skilled in the art using existing image processing libraries such as OpenCV; these are well-known techniques in the field and will not be elaborated upon here.

[0055] See attached document Figure 1 and attached Figure 5 During step S4, which involves using the constrained Weber model to perform dynamic parameter feedback and weight correction, a discrete dataset containing timestamps and preliminary pollen viability observations is obtained through previous steps. Subsequently, in order to reconstruct the continuous physiological decline trajectory of pollen populations under storage conditions from a limited number of discrete observation points, this embodiment introduces a survival analysis model from reliability engineering to perform nonlinear regression processing on the data. The core technical objective of this process is to quantify the decay rate characteristics of the pollen population and dynamically calibrate the weight of sub-viable particles in the final evaluation accordingly, thereby providing a more rigorous physiological state evaluation dimension than simple morphological counting.

[0056] As a preferred mathematical description method, this embodiment uses the survival rate form of the two-parameter Weiber distribution function to construct the pollen viability decay model. The physical basis for choosing this model is that its shape parameters can flexibly characterize various biological processes, from random failure (constant failure rate) to wear-down failure (incremental failure rate), making it particularly suitable for describing the evolution of viability of ex vivo biological materials over time. The constructed nonlinear regression equation is as follows: ; in: Indicates any continuous storage time (Unit: days) The theoretical viability retention rate (%) of the pollen population, which is intended to fit the true physiological level after eliminating observational noise; Represents the initial time (i.e.) The theoretical vigor benchmark value reflects the original quality of germplasm, and its value is constrained by biological limits. This represents the scale parameter, which physically corresponds to the characteristic lifetime, i.e., the time it takes for pollen viability to decline to its initial value. The duration of time experienced (approximately 36.8%) directly quantifies the time span of shelf life; Shape parameters are key kinetic indicators for determining pollen population aging patterns. The time corresponds to a constant failure rate (dominated by random factors), while This corresponds to an accelerated aging rate that increases over time (dominated by physiological depletion), among which The shape parameter threshold represents the boundary between constant failure rate and attrition failure based on the Weiber distribution.

[0057] Using the time-series data output from the preceding steps, the parameters of the above model are estimated using nonlinear least squares optimization algorithms such as trust region reflection or Levenberg-Marquardt. The specific code implementation and solver calls for the nonlinear least squares optimization algorithms can be accomplished by those skilled in the art using existing scientific computing libraries (such as SciPy or MATLAB). The underlying iterative logic and function interfaces are well-known technologies in the field and will not be elaborated upon here. To ensure that the fitting results conform to biological common sense and prevent the algorithm from getting trapped in meaningless local optima, strict physical boundary constraints are imposed on the parameter space during the iterative optimization process: initial vitality is required... It must fall within a closed interval between 0 and 100; at the same time, the scale parameter is required. and shape parameters All parameters are strictly greater than a very small positive number (e.g., 0.1) to avoid mathematical non-definiteness or division by zero errors. The optimal parameter vector characterizing the current germplasm characteristics is calculated by iteratively minimizing the sum of squared residuals. .

[0058] Based on the shape parameters obtained by fitting This involves performing adaptive weight adjustments for sub-active particles. The biological logic behind this feedback mechanism is that when... At this time, it indicates that the pollen population is in a wear-and-tear accelerated aging stage. At this point, the lighter-colored, less viable particles (pink) are often on the verge of irreversible death, and their fertilization function is lost faster than their enzyme activity. The baseline weight is used. Therefore, to improve the accuracy of the evaluation results in predicting actual pollination capacity, the contribution weight of sub-viability particles needs to be reduced. Correction weight coefficient. The calculation logic is as follows: ; in: These are the corrected weighting coefficients applied to sub-active particles. The physical meaning of this piecewise function is: when random failure characteristics are detected ( When ), the full contribution of subactive particles is retained; while when obvious wear-induced accelerated aging characteristics are detected ( When performing this process, a punitive weight decay that is inversely proportional to the aging rate is applied, which mathematically suppresses the risk of overestimation due to false positive staining (i.e., enzyme activity but no fertilization capacity).

[0059] Obtain the global corrected weight coefficients Then, for the discrete sampling time set Sampling time points The pollen viability values ​​are then weighted and calculated to generate the final dataset for grading and evaluation. The revised viability calculation formula is as follows: ; in: This indicates the time point corresponding to the sampling time after feedback correction from the dynamic model. The effective pollen viability value will serve as the direct basis for subsequent judgment on whether the 50% threshold is met; Indicates the sampling time point The number of observed deep red, highly active particles has a constant weight of 1; Indicates the sampling time point The observed pink subactive particle number contributes to the number of particles. Dynamic adjustment; This represents the total number of particles currently in the field of view, and needs to be verified before the calculation is executed. If the denominator is zero, the current field of view is deemed invalid and automatically removed.

[0060] Through this feedback mechanism, the present invention will transform the macroscopic time series decay characteristics (by...) The representation is inversely coupled to the value judgment of micro-entities, resulting in the final output of the final vitality decay data sequence. It not only reflects the current staining status, but also implies a prediction of the aging trend of the population, thus significantly improving the scientific nature and guiding significance of germplasm storage tolerance grading.

[0061] See attached document Figure 1 During step S5, which involves calculating storage tolerance evaluation indices and outputting grading results, the aim is to convert the final viability decay data sequence output in step S4, after kinetic correction, into a more robust and sustainable performance. This is transformed into a single physical indicator that can intuitively quantify the germplasm's storage potential: pollen viability half-life. ), and complete the automated grading of germplasm based on this indicator.

[0062] In this embodiment, to reconstruct the continuous decline trajectory of pollen viability from discrete observation data and to accommodate the inherent random fluctuations of biological data, a dual-path computation strategy is adopted, primarily using model fitting and secondarily using linear interpolation. The core of this strategy lies in solving for the time span during which pollen viability decays from its initial state to the critical biological threshold of 50%, i.e., the half-life.

[0063] First, a quadratic nonlinear regression analysis is performed on the input data sequence. The corrected data points are then substituted into the aforementioned constrained Weber model, and the final model parameter vector, including the final initial vitality, is solved using nonlinear least squares methods (such as trust region reflection or the Levenberg-Marquardt algorithm). Final scale parameters and final shape parameters During this process, the coefficient of determination for regression analysis is calculated simultaneously. This is used to quantify the interpretability of the model curve to the measured data. The specific code writing, solver calls, and determination coefficient calculation for the aforementioned nonlinear least squares fitting can be accomplished by those skilled in the art using existing scientific computing function libraries (such as the SciPy library in Python or MATLAB toolboxes). The underlying computational logic and function interfaces are well-known technologies in this field and will not be elaborated upon here.

[0064] Based on the calculation The value and the preset goodness-of-fit threshold (In this embodiment, the value is preferably set to 0.85), and the pollen viability half-life is automatically calculated by selecting any of the following paths. : Path 1: Parametric Model Solution Path When the judgment This indicates that the pollen viability decay process highly conforms to the Weiber distribution law, and analytical calculations are performed using model parameters. First, the final initial viability is verified. : like This indicates that the germplasm did not meet the breeding requirements at the theoretical initial moment, and therefore it was directly determined that... ; like Then, based on the inverse operation formula of the constrained Weber model, the analytical result is... The calculation formula is as follows: ; in: The calculated pollen viability half-life is expressed in days (d). This indicator eliminates the influence of uneven observation time intervals through mathematical derivation. This represents the final scale parameter obtained from the quadratic fitting, which physically corresponds to the feature lifetime; Indicates the initial time obtained from the fitting ( Theoretical pollen viability (%), i.e., final initial viability; This represents the final shape parameters obtained from the quadratic fitting. Represents the natural logarithm operation; This represents the target vitality threshold (i.e., 50%).

[0065] Path 2: Non-parametric linear interpolation path When the judgment When the data exhibits drastic fluctuations that do not conform to a typical S-curve, the system automatically switches to piecewise linear interpolation logic to prevent forced fitting from introducing computational errors.

[0066] The algorithm first detects the final vitality decay data sequence. The numerical distribution: If all sampling time points in the final vitality decay data sequence Corresponding effective pollen viability value If all values ​​are above 50%, the germplasm is determined to be extremely resistant to storage and is directly assigned a value. (or a value greater than 15); if all sampling time points Corresponding effective pollen viability value If all are below 50%, then directly judge .

[0067] For cases where the vitality value exceeds the 50% threshold, the algorithm searches for two adjacent sampling time points in the final vitality decay data sequence that first meet the condition. and (in and For adjacent The next sampling time), satisfying the condition is and Based on the linear assumption between these two points, estimate the time point when vitality crosses the 50% threshold: ; in: and These are adjacent sampling time points before and after the effective pollen viability value drops to 50% (e.g., day 5 and day 7). and These correspond to the sampling time points. and The effective pollen viability value. The denominator needs to be verified before performing the division operation. ( For example, take a very small positive number. If the denominator approaches zero, then take the value directly. .

[0068] Obtaining a unique quantitative indicator Subsequently, based on the actual operational window period of wolfberry hybridization breeding, this value was mapped to a pre-defined four-level evaluation standard table. This grading standard, based on discrete observation criteria from wolfberry breeding practice regarding day 5 (first storage tolerance threshold), day 9 (second storage tolerance threshold), day 13 (third storage tolerance threshold), and day 15, constructs a continuous half-life evaluation interval, determined logically as follows: Weak storage tolerance (Grade IV): The criteria for judgment are as follows This type of germplasm has a viability that drops below 50% before being stored at room temperature for 5 days, making it only suitable for immediate crossbreeding in the same field and not suitable for long-distance transportation.

[0069] Medium storage resistance (Grade III): Judgment criteria are as follows This type of germplasm remains viable on day 5, but becomes ineffective before day 9, thus meeting the needs for short-distance germplasm exchange and staggered pollination within a week.

[0070] Strong storage resistance (Grade II): Judgment criteria are as follows This type of germplasm has excellent storage resistance, can adapt to long-distance transportation across provinces, and is sufficient to cope with hybrid combinations where the flowering periods of the male and female parents differ by about 10 days, and is listed as a preferred candidate for male parent.

[0071] Strong storage resistance (Grade I): Judgment criteria are as follows This type of germplasm is an extremely durable material with a half-life exceeding 13 days. Logically, it seamlessly meets the upper limit of Grade II and fully covers the high-quality cases where viability remains above 50% even after storage up to day 15 within a regular observation period. It is the preferred paternal parent material for establishing pollen banks and conducting long-distance, large-span staggered hybridization.

[0072] Finally, the user interface generates a data set including germplasm name and pollen viability half-life. Electronic reports on storage tolerance grades are also provided. Germplasm assessed as having strong or relatively strong storage tolerance is automatically marked as a recommended paternal parent, thereby enabling a shift in breeding decisions from experience-driven to data-driven approaches.

[0073] See attached document Figure 6 and attached Figure 7 This specific embodiment uses the screening of superior, storage- and transport-resistant paternal lines in the Ningxia Yinchuan wolfberry germplasm resource nursery as an example to detail the entire process of implementing the method for evaluating the fertility of wolfberry parents based on pollen viability according to this invention. This embodiment aims to solve the technical problems of existing single-point observation methods being unable to distinguish false positives with high viability and unable to predict losses during cross-regional transportation.

[0074] Step S6: Construct a standardized Lycium barbarum pollen time-series sample library. During the peak flowering period of Lycium barbarum (mid-May in this example), two target germplasms to be evaluated were selected in the resource nursery and labeled as Yunnan Lycium barbarum (experimental group A) and Sichuan Lycium barbarum (experimental group B), respectively. Ten healthy plants of the same age were selected from each group.

[0075] On a sunny morning, after the dew on the flowers has completely dried (usually when the temperature rises to around 20℃), collect 5 flowers from each of the four directions of the tree canopy that are about to open (whitening stage). At this stage, the petals are slightly tinged with color at the tips, but the corolla is not yet fully open, and the anthers are not yet dehiscent. Transport the collected flower buds back to the laboratory and place them in a cool, shaded indoor area until the anthers naturally emerge. Remove the stamens, spread them evenly in sterile petri dishes, and place them in a thermostatic drying oven. Set the drying temperature to 25±0.5℃ for 3 hours, until the anther walls dry and shrink, releasing pure pollen. Collect the pollen and remove the filaments. Evenly distribute each type of pollen into labeled petri dishes and store them in a 25℃ thermostatic incubator to construct a sampling time set. (Unit: days), covering the entire cycle from fresh state to expected decline.

[0076] Step S7: Microscopic Image Acquisition and Adaptive Illumination Correction. At each time point in the sampling time set (e.g., day 5), the corresponding Yunnan and Sichuan wolfberry samples were removed from the incubator. A 1.0% triphenyltetrazolium chloride (TTC) solution was used as the staining agent. A small amount of pollen was smeared onto a glass slide using a cotton swab, 20 μL of staining solution was added, a coverslip was placed on top, and the slide was incubated in a humidified chamber on a 35°C water bath for 15 minutes in the dark to prepare the observation sample.

[0077] To illustrate the data processing flow in detail, this embodiment selects representative key time points from the sampling sequence. The data processing procedure (i.e., data stored on the 5th day) is demonstrated in detail.

[0078] Using an optical microscope at 100x magnification, five non-overlapping fields of view were randomly selected from each slide to obtain the original red, green, and blue images. To address light source fluctuations during imaging, fluctuations in the microscope's light source voltage were detected, leading to a darker background. The algorithm automatically identified the background reference area at the image edges and calculated... , , Set the target white point baseline value. Calculate the gain coefficient of the red channel. Perform the following on all pixels of the red channel of the image: Similarly, calculate the gains of the green and blue channels, perform a linear transformation on all pixels in the image, and output a corrected image with uniform background brightness (RGB values ​​approaching 250, 250, 250).

[0079] Step S8: Initial quantization grading based on HSV color space. Convert the corrected image to HSV space. Set the high vibrancy threshold. (Corresponding to deep red), low vitality threshold (Corresponding to pink).

[0080] For a typical field of view of Yunnan wolfberries on day 5: the image algorithm identifies the saturation. Deep red particles (high-activity particle count) Individual; identified The number of pink particles (sub-active particles) Individual; identified The number of colorless particles (inactive particles) The number of sub-active particles and the number of highly active particles are combined to calculate the total number of effectively active individuals. Calculate preliminary pollen viability observations (uncorrected): .

[0081] For a typical field of view of Sichuan wolfberry on day 5: a high number of highly active particles were identified. Number of sub-active particles Number of inactive particles The number of sub-active particles and the number of highly active particles are combined to calculate the total number of effectively active individuals. Calculate preliminary pollen viability observations (uncorrected): .

[0082] Step S9: Dynamic Model Fitting and Data Correction. Summarize the observation data from all time points (days 0-15) to construct a full lifecycle sequence. Based on the full-cycle measurements in this embodiment, the following key measured data were obtained: Yunnan wolfberry: 93% on day 5, 88% on day 9, 85% on day 13, and 58% on day 15. Sichuan wolfberry: 30% on day 5 and 0% on day 9.

[0083] The system performs a constrained Weber model fit on the above sequences: Yunnan wolfberry: Shape parameters obtained from fitting .because This indicates that the group exhibits typical wear-and-tear aging characteristics, and is in the accelerated aging stage (i.e., stable in the early stage, with an accelerated aging rate in the later stage). At this stage, a weight correction mechanism needs to be triggered to eliminate false positive interference. The corrected weight coefficients are calculated. The effective pollen viability value on the 5th day after correction is... .

[0084] Sichuan wolfberry: Shape parameters obtained from fitting It exhibits random failure characteristics; calculate the corrected weighting coefficient. This means no punitive adjustments will be made. Vitality remains at 30.0% on day 5.

[0085] After the above processing, the final viability decay data sequences of the two germplasms at each sampling time point were obtained. (See attached...) Figure 6 The graph is drawn in grayscale mode, with the horizontal axis representing storage time (days) and the vertical axis clearly representing pollen viability (%). To accurately display the viability status at key time points, black projected auxiliary dashed lines extending vertically from key data points to the X and Y axes respectively are introduced in the graph, and the specific values ​​of the key points are directly labeled in coordinate form (number of days, viability %). For Yunnan wolfberry: the projection line clearly shows that its viability reached 93% on day 5 (after correction), i.e., coordinates (5, 93%); until the end of the observation period on day 15, its viability remained at 58%, i.e., coordinates (15, 58%). This curve exhibits a distinct convex plateau characteristic, consistent with... The dynamic parameters are consistent.

[0086] For Sichuan wolfberry: the projection line shows a sharp decline in its vitality, which has dropped to 30% by day 5, i.e., coordinates (5, 30%).

[0087] Step S10: Data Analysis and Hierarchical Application. (See Appendix) Figure 6 and attached Figure 7 Based on the final viability decay data sequence output in step S9, the pollen viability half-life is calculated. ), and make a judgment based on the preset grading standards.

[0088] See attached document Figure 7 In addition to showing the half-life bar data for the two properties, the figure also clearly marks three key boundary lines that determine the classification: the first storage tolerance threshold (5d), the second storage tolerance threshold (9d), and the third storage tolerance threshold (13d).

[0089] Evaluation of Sichuan wolfberries: The effective pollen viability value was 55.0% on day 3. ), which dropped to 30.0% on day 5. Since the data crosses the 50% threshold in this range and exhibits severe decay unsuitable for model extrapolation, the system automatically selects linear interpolation to solve for the half-life. Heaven. (In the appendix) Figure 7 In this context, the value (3.4 days) is located to the left of the first storage tolerance threshold (5 days). Because... The system classified it as Level IV (weak storage tolerance).

[0090] Evaluation of Yunnan wolfberries: Until the end of the observation period (day 15), their vitality was 58.0%, consistently above 50%. This could not be directly obtained through interpolation. The system automatically selects the constrained Weber model parameter extrapolation method. Based on the parameters obtained from step S9, the predicted time point when its theoretical vitality drops to 50% is... Heaven. (In the appendix) Figure 7 In this context, the value (>15 days) is located to the right of the third storage tolerance threshold (13 days). Because... If the activity level is greater than 50% on day 15, the system classifies it as Level I (strong storage tolerance).

[0091] Data analysis and conclusions, combined with appendix Figure 6 and attached Figure 7 Data analysis shows that: Significant differences in kinetic characteristics: Vitality decay curves of Yunnan wolfberry (attached) Figure 6 The solid line exhibits typical high plateau and slow decay characteristics, and the fitted shape parameters... This confirms its extremely strong ability to resist physiological depletion. Conversely, the curve for Sichuan wolfberry (see attached curve)... Figure 6 (Dotted line) A precipitous drop occurred between day 3 and day 5.

[0092] Breeding decision guidance: The system automatically outputs breeding suggestions, marking Yunnan wolfberry as the target male parent material and recommending it for cross-provincial introduction or large-span staggered hybridization; while Sichuan wolfberry is only recommended for immediate hybridization within the same plot.

Claims

1. A method for evaluating the fertility of wolfberry parent lines based on pollen viability, characterized in that, Includes the following steps: Establish a sampling time series of wolfberry pollen containing multiple discrete sampling time points; Acquire pollen microscopic images at each of the sampling time points and perform adaptive white balance normalization processing to output corrected images; Based on the HSV color space saturation characteristics of the corrected image, effective viable individuals are identified and preliminary pollen viability observations are calculated to form a preliminary viability decay sequence. The shape parameters are obtained by fitting the preliminary vitality decay sequence using a constrained Weber model. The contribution weights of particles in the sub-vitality state in the effective vitality individuals are corrected using the weight coefficients determined by the shape parameters, thereby generating the final vitality decay data sequence. The pollen viability half-life is calculated based on the final viability decay data sequence, and the storage tolerance level is determined by comparing the pollen viability half-life with the preset storage tolerance grading standard threshold, thus generating the fertility evaluation results of the wolfberry parent line.

2. The method for evaluating the fertility of wolfberry parent lines based on pollen viability according to claim 1, characterized in that, The specific steps involved in forming a preliminary viability decay sequence include: Extract the saturation channel data of the corrected image in the HSV color space, and set high vibrancy and low vibrancy thresholds; Particles with an average saturation greater than or equal to the high vitality threshold in the saturation channel data are identified as high vitality particles, and particles with an average saturation between the low vitality threshold and the high vitality threshold are identified as particles in a sub-vitality state. The highly active particles and the sub-active particles are combined to form the effective active individual; The preliminary pollen viability observation value is obtained by calculating the proportion of the number of viable individuals to the total number of particles. The preliminary pollen viability observations corresponding to each sampling time point are arranged in chronological order to form the preliminary viability decay sequence.

3. The method for evaluating the fertility of wolfberry parent lines based on pollen viability according to claim 2, characterized in that, The specific steps for generating the final viability decay data sequence include: The preliminary vitality decay sequence is input into the constrained Weber model for nonlinear regression fitting to extract the shape parameters; When the fitted shape parameter is greater than the preset shape parameter threshold, the weight coefficient is set to the reciprocal of the shape parameter; When the fitted shape parameter is less than or equal to the shape parameter threshold, the weight coefficient is set to a preset baseline weight. The step of correcting the contribution weight of the sub-active particle using the weighting coefficient specifically includes: The pollen viability value is calculated by weighting the number of particles in a sub-viable state using the weighting coefficients and summing the weights with the number of highly viable particles. The pollen viability values ​​recalculated at each of the sampling time points are arranged in chronological order to generate the final viability decay data sequence.

4. The method for evaluating the fertility of wolfberry parent lines based on pollen viability according to claim 1, characterized in that, The constrained Weber model is a nonlinear regression model that imposes physical boundary constraints on the initial vitality parameter, scale parameter, and shape parameter.

5. The method for evaluating the fertility of wolfberry parent lines based on pollen viability according to claim 1, characterized in that, The specific steps for calculating pollen viability half-life include: Substitute the final vitality decay data sequence into the constrained Weber model for a second-order fitting and calculate the coefficient of determination. If the determination coefficient is greater than or equal to the preset goodness-of-fit threshold, the pollen viability half-life is calculated analytically using the model parameters obtained from the quadratic fitting. If the determination coefficient is less than the goodness-of-fit threshold, the pollen viability half-life is calculated using linear interpolation.

6. The method for evaluating the fertility of wolfberry parent lines based on pollen viability according to claim 5, characterized in that, The linear interpolation method involves searching for two adjacent sampling time points in the final vitality decay data sequence that first cross the target vitality threshold, and then performing linear interpolation calculations based on the vitality values ​​of the two adjacent sampling time points.

7. The method for evaluating the fertility of wolfberry parent lines based on pollen viability according to claim 1, characterized in that, The specific steps for performing adaptive white balance normalization include: The gain coefficient is calculated by using the ratio of the statistical mean of the red, green, and blue channels of the background reference area in the pollen micrograph to the target white point reference value. The gain coefficient is then used to perform a linear transformation on the pixel values ​​of the pollen micrograph to output the corrected image.

8. The method for evaluating the fertility of wolfberry parent lines based on pollen viability according to claim 1, characterized in that, The specific steps for establishing a sampling time series include: Mature flower buds in the whitening stage were selected, the stamens were removed and placed in a constant temperature and dry environment until the anthers were pollinated. The sampling time series covering the pollen viability decline process was constructed using the day of pollen collection as the starting point.

9. The method for evaluating the fertility of wolfberry parent lines based on pollen viability according to claim 1, characterized in that, The storage resistance grading standard thresholds include a first storage resistance judgment threshold, a second storage resistance judgment threshold, and a third storage resistance judgment threshold, wherein the first storage resistance judgment threshold is less than the second storage resistance judgment threshold, and the second storage resistance judgment threshold is less than the third storage resistance judgment threshold. The specific steps for determining the storage resistance level are as follows: When the pollen viability half-life is less than the first storage tolerance threshold, it is determined to be weak storage tolerance; when the pollen viability half-life is between the first storage tolerance threshold and the second storage tolerance threshold, it is determined to be moderate storage tolerance. When the pollen viability half-life is between the second storage tolerance threshold and the third storage tolerance threshold, it is determined to have strong storage tolerance. When the pollen viability half-life is greater than or equal to the third storage tolerance threshold, it is determined to be of strong storage tolerance.

10. The method for evaluating the fertility of wolfberry parent lines based on pollen viability according to claim 9, characterized in that, The specific steps for generating the fertility evaluation results of wolfberry parent lines include: Germplasm with strong or relatively strong storage tolerance is marked as preferred paternal parent material, and germplasm with weak storage tolerance is marked as material to be harvested and used immediately. The marked germplasm is determined as the fertility evaluation result of the wolfberry parent.

Images